Contents lists available at SciVerse ScienceDirect
Journal of Controlled Release
j ourna l homepage: www.e lsev ie r .com/ locate / jconre l
NANOMEDICIN
E
A micelle-shedding thermosensitive hydrogel as sustained release formulation
Albert J. de Graaf a, Inês I. Azevedo Próspero dos Santos a, Ebel H.E. Pieters a, Dirk T.S. Rijkers b,Cornelus F. van Nostrum a, Tina Vermonden a, Robbert J. Kok a, Wim E. Hennink a, Enrico Mastrobattista a,⁎a Utrecht Institute for Pharmaceutical Sciences, Pharmaceutics, Utrecht University, P.O. Box 80.082, 3508 TB Utrecht, The Netherlandsb Utrecht Institute for Pharmaceutical Sciences, Medicinal Chemistry & Chemical Biology, Utrecht University, P.O. Box 80.082, 3508 TB Utrecht, The Netherlands
In this paper it is shown that when a thermosensitive hydrogel based on poly(N-isopropylacrylamide)-poly(ethylene glycol)-poly(N-isopropylacrylamide) (pNIPAm-PEG-pNIPAm) was transferred into water,flower-like micelles were continuously released as long as the medium was regularly refreshed. On theother hand, if the medium was not refreshed the concentration of micelles reached an equilibrium. When thisgel was loaded with the cytostatic agent paclitaxel (PTX), the released micelles solubilized PTX, as evidenced bya PTX concentration in the release medium above its aqueous solubility. To test the applicability of thesemicelle-releasing gels for sustained and systemic delivery of PTX an in vivo experiment was performed intumor-bearingmice. pNIPAm-PEG-pNIPAmgels (without andwith 1.2% and 6.0% PTX loading)were administeredi.p. in nudemice bearing 14C human squamous cell carcinoma tumor xenografts to obtain doses corresponding toone and five times themaximum tolerated dose of PTX (when given i.v. as the standard formulation in CremophorEL/ethanol). All gel formulations were well tolerated and no signs of acute systemic toxicity were observed. Afterinjection of the highest dose, PTX levels in serum could be determined for 48 h with a comparatively long elimi-nation half-life of 7.4 h pointing to a sustained release of PTX. A bioavailability of 100% was calculated from thearea under the curve of plasma concentration vs time. Furthermore, at the highest dose, PTX was shown tocompletely inhibit tumor growth for at least 3 weeks with a single hydrogel injection. This promising conceptmay find application as a depot formulation for sustained, metronomic dosing of chemotherapeutics.
The objective of the present work was to develop a drug-loadedin situ gelling thermosensitive hydrogel that slowly interconverts intodrug-loaded polymeric micelles. The released micelles can solubilize ahydrophobic drug: the release kinetics of the drug are therefore likelycontrolled to a large extent by the dissolution rate of the gel.
This concept emerged from the notion that in an aqueous envi-ronment hydrophobic-hydrophilic-hydrophobic (BAB) triblock copoly-mers can self-assemble into hydrogels as well as flower-like micelles,depending on their concentration [1–10]. Polymeric micelles arewidelystudied as drug delivery vehicles for low-molecularweight hydrophobicdrugs [11–16], whereas (self-assembling) hydrogels are under investi-gation as depot formulations for various drugs, especially proteins[17,18]. It has been shown in a number of studies that at high concentra-tions of BAB block copolymers in water there is always an equilibriumbetween a gel (consisting of bridged micelles) and free micelles [7–9].This equilibrium implies that, if such a hydrogel would be formed invivo, it will release micelles over time.
Recently, there have been a number of reports on hydrogel/micellecomposite systems in which micelles are incorporated in hydrogels of
: +31 30 251 78 39.ttista).
rights reserved.
a different chemical composition [19–22]. Two hydrogel systems havebeen described that can be degraded by glutathione, leaving behind poly-mericmicellesmade of the samematerial as the parent gel [23,24]. How-ever, in the latter publications this micelle formation was only noted as apeculiarity and was not discussed in terms of applicability for making agradually releasing micellar drug delivery system. No papers have beenpublished yet, which describe a drug-loaded hydrogel that graduallyand spontaneously interconverts into drug-loaded micelles.
It might be envisioned that such a system is beneficial for thedelivery of cytostatic drugs, as (i) it could reduce the often used long in-fusions to a single injection and (ii) cytostatic drugs have repeatedlyshown a longer plasma half-life and increased tumor accumulation (dueto the enhanced permeability and retention effect) when formulated inpolymeric micelles as compared to the free drugs [11–16,25–27]. Fur-thermore, over the last decade there has been an increasing interest inthe use of so-calledmetronomic dosing schedules for chemotherapeutics.Withmetronomic dosing a continuous (up toweeks) dose of one ormorechemotherapeutic agents is given at a rate which is much lower thanfor conventional chemotherapy [28–31]. Paclitaxel appears to be a strongcandidate for metronomic chemotherapy given its broad-spectrum anti-tumor activity and its ability to inhibit endothelial cell functions relevantto angiogenesis in vitro at extraordinarily low concentrations (10 ng/mL)[29,32–37]. A hydrogel, which continuously releases PTX for a long timecould be advantageous in such a metronomic therapy.
583A.J. de Graaf et al. / Journal of Controlled Release 162 (2012) 582–590
NANOMEDICIN
E
In the present study, poly(N-isopropylacrylamide)-poly(ethyleneglycol)-poly(N-isopropylacrylamide) (pNIPAm-PEG-pNIPAm) thermo-sensitive block copolymers were used as model polymers for the in-vestigation of the release of flower-like micelles from thermosensitive(in situ gelling) hydrogels. The lower critical solution temperature ofpNIPAm (32 °C) allowed dissolution of the polymers at room tempera-ture and gel formation at 37 °C. Furthermore, these polymers havealready been shown to form gels at high concentrations and flower-like micelles at low concentrations [38,39].
In this paper, we report the in vitro release of micelles and PTXfrom pNIPAm-PEG-pNIPAm gels. Furthermore, release kinetics andefficacy were studied in an in vivo mouse model. As route of adminis-tration for the PTX-loaded hydrogels, i.p. injection was chosen as itallowed administration of a suitable amount of hydrogel leading torapid in situ gel formation. PTX i.p. is under investigation as local ther-apy for ovarian cancer [40,41], but there are several reports that afteri.p. injection PTX is also able to reach the systemic circulation via thelymphatic system, with reported bioavailabilities between 5% and37% [35,42–45]. When used as local therapy, high plasma levels ofPTX are considered undesirable. Only recently, a sustained systemicrelease after i.p. injection of the related compound docetaxel, was rec-ognized as advantageous for treatment of distal metastases [46].Based on these data, we hypothesize that PTX, solubilized in micellesthat are released from a gel, can be transported to the circulation andfrom there reach and affect distant tumors.
OO
On
OOBr
OO
OBr
HN Om
pNIPAm-PEG
C
Scheme 2. Schematic representation of polymerizatio
2. Experimental methods
2.1. Chemicals
PEG with molar mass of 6 kDa was purchased from Merck anddehydrated prior to use by azeotropic distillation using toluene. N-isopropylacrylamide (NIPAm; Aldrich, >99%), 2-bromoisobutyrylbromide(Aldrich), CuBr (Aldrich) and CuBr2 (Acros) were used as received.Tris(2-dimethylaminoethyl)amine (Me6TREN)was prepared accordingto a reported procedure [47].
2.2. Synthesis of PEG initiator
The synthesis is summarized in Scheme 1.Dehydrated PEG (5.0 g) was dissolved in 50 mL of dichloromethane
(CH2Cl2) dried on molecular sieves and degassed by flushing withnitrogen. The flask was put in an ice bath and subsequently 1.2 eqtriethylamine (to –OH groups) and 1.2 eq 2-bromo-isobutyrylbromide(to –OH groups) were added. The mixture was allowed to react over-night at room temperature. Afterwards, dichloromethanewas removedin vacuo. Tetrahydrofuran (THF, 50 ml) was added and the bromide saltwas filtrated off. The filtrate was concentrated in vacuo and dissolvedin aminimum amount of dichloromethane. The crude product was pre-cipitated in cold diethylether and then centrifuged (1000×g, 5 min,4 °C) and the obtained white precipitate was filtrated and dried byflushing with nitrogen to yield a white powder. The product was fur-ther dried overnight in a desiccator. 1 H NMR analysis was performedafter drying. A second spectrum was recorded after adding two drop-lets of trichloroacetyl isocyanate to the sample to derivatize the un-reacted OH-groups and thereby enable determination of the degreeof substitution.
The final yield was 85%. 1H NMR spectrometry indicated that 85% ofthe PEG –OH groups were functionalized and no excess 2-bromo-isobutyrylbromide was left. 1H NMR (300 MHz, CDCl3): δ 4.32 ppm(t, 4 H, OCH2), δ 3.63 ppm (t, 4n H, OCH2), δ 1.93 ppm (s, 12 H,C(CH3)2).
2.3. Synthesis of pNIPAm-PEG-pNIPAM
The polymerization reaction is summarized in Scheme 2.Screw-capped septum vials with the following content were pre-
pared: 0.3 g of PEG6kDa macroinitiator (0.05 mmol PEG; 0.1 mmol ini-tiating sites), 9.0 mg CuBr, 9.4 mg CuBr2 and 1.6 g or 3.2 g NIPAm
Br
On
OBr
OHN
m
OHN
+ 2 m
-pNIPAm
ATRP
H2O/CH3CN4°C
15 min
uBr/CuBr2/Me6TREN
n reaction of NIPAm onto the PEG macroinitiator.
584 A.J. de Graaf et al. / Journal of Controlled Release 162 (2012) 582–590
NANOMEDICIN
E
(target Mn: 16 kDa resp. 32 kDa). A stirring bar was introduced ineach vial and the vials were capped with a septum. The solutionswere degassed by flushing with N2 and placed in an ice bath. WhenNIPAm was dissolved, the reaction was started by adding 0.25 mL0.42 M Me6TREN solution in water (degassed by flushing with N2
prior to addition), which immediately turned the reaction mixtureblue. Periodically, 20 μL samples were taken, diluted in air-saturatedD2O and analyzed by 1H NMR to determine the NIPAm conversion.When the conversion had reached >95%, the reaction was quenchedbyflushingwith air for severalminutes. The crude productwas dialyzed(with dialysis cassettes of 10 kDa MWCO) against water at 4 °C over-night and subsequently lyophilized.
The polymers were characterized by 1H NMRwith a VarianMercuryPlus 300 MHz NMR instrument and by GPC. GPC was performed usingMixed-D columns (Polymer Laboratories), a solution of 10 mM LiClin DMF as eluent, a flow of 0.7 mL/min and a temperature of 40 °C. Arefractive index detector was used and linear PEG standards were usedfor calibration.
2.4. Determination of the Cloud Point (CP)
Samples of the polymers were prepared at a concentration of1 mg/mL in water and phosphate-buffered saline (PBS). A ShimadzuUV 2450 spectrophotometer equipped with a Peltier heating ele-ment was used with the Tm analysis software to measure the CPat a wavelength of 650 nm, raising the temperature from 20 °C to50 °C at 1 °C/min.
2.5. DLS
Micelles were formed by a heat shock procedure [48]. In short, anaqueous polymer solution (100 μL, 1 mg/mL) at room temperaturewas added to 900 μL of water which was pre-heated to 50 °C. The mi-cellar dispersions were equilibrated at 40 °C for 5 min before beingmeasured in a Malvern CGS-3 goniometer (Malvern Ltd., Malvern,U.K.) coupled to an LSE-5003 autocorrelator, a thermostated waterbath (set to 40 °C), a He-Ne laser (25 mW, 633 nm, equipped witha Uniphase 2500 remote interface controller,) and a computer withDLS software (PCS, version 3.15, Malvern). The measurement anglewas 90°. The solvent viscosity was corrected for the temperature bythe software.
2.6. Gel formation
For all gels, unless noted otherwise, the following procedure wasused. Polymer was dissolved to a final concentration of 25% (w/w)in water. This concentration was chosen because it offers a suitablebalance between fluidity at room temperature, gelation kinetics andgel strength above the CP. The required volume of polymer solutionwas placed on the bottom of a vial and kept for 10 min at 50 °C in awater bath, before being equilibrated at 37 °C.
2.7. Temperature-sweep rheology
To study the gel formation, 60 μL of a polymer solution (25% (w/w)in water or PBS) was introduced into an AR-G2 rheometer (TA Instru-ments Ltd) equippedwith a Peltier plate for temperature control usinga cone-plate geometry (steel, 20 mm diameter with an angle of 1°)and Rheology Advantage Instrument Control AR software. A solventtrap was used to prevent evaporation of the solvent. The storage (G’)and loss (G”) modulus were measured for a range of temperaturesfrom 20 °C to 60 °C with a heating rate of 1 °C/min. A strain of 1%and a frequency of 1 Hz were used. The data were analyzed throughRheology Advantage Data Analysis software.
2.8. Imaging the release of micelles from in situ formed gels
One mL of filtered demineralized water was equilibrated at 40 °Cin the measurement chamber of a Nanosight LM14-HS laser lightscattering microscopy system equipped with a 532 nm laser and anactively cooled EMCCD camera. Approximately 0.1 mL of a 25% (w/w)aqueous solution of polymer of approximately 20 °C was injected toform a hydrogel in situ, comparable to the way a hydrogel would beformed upon i.p. injection in vivo. Diffusion of the micelles releasedfrom the gel was followed in time, and their size distributionwas deter-mined from their Brownianmotion using nanoparticle tracking analysis(NTA) software.
2.9. Release kinetics of micelles from gels measured with DLS
Hydrogels were formed by heating 50 μL of a 25% (w/w) polymersolution from room temperature to 50 °C. After 10 minutes, 2 mL ofPBS (filtered through 0.2 μm and preheated to 37 °C) was added ontop of the hydrogel. Samples of 1 mL were collected every 15 minutesand transferred into a pre-heated DLS cuvette, and the initial solutionwas refilled with 1 mL of pre-heated PBS to maintain the volume. Thesamples were analyzed by DLS measurements at 40 °C.
2.10. Preparation of PTX-loaded gels
An aqueous polymer solution (6 mL, 50 mg/mL) was filteredthrough a 0.2 μm filter. The required volume of a 30 mg/mL solutionof PTX in CH3CN was quickly added to the polymer solution (atroom temperature) by a microliter syringe. Either 0, 120 or 600 μLwas added to yield PTX/polymer ratios of 0.0%, 1.2% and 6.0% (w/w),respectively. The mixture was briefly vortexed, after which an equalvolume of sterile filtered water (preheated to 60 °C) was added atonce. The resulting micellar dispersion was again briefly vortexedand equilibrated at 40 °C in awater bath for 15 min. Then, themicellardispersion was snap-frozen by transferring it dropwise to a 50 mLpolypropylene tube filled with liquid nitrogen. The resulting icebeads were freeze dried for 48 h and stored at −20 °C until use. Justbefore the experiment, the freeze dried micelles were allowed tothaw andwere swollen in sterile-filtered water (900 μL) at room tem-perature. Air bubbles were removed by shortly keeping the tube in anultrasonic bath. The gels without or with 1.2% PTX loading were clear,whereas the gel with 6.0% PTX loading had a slight bluish tinge.
2.11. In vitro release of PTX in water
A volume of pNIPAm16kDa-PEG6kDa-pNIPAm16kDa (N16P6N16) gel(20 μL) loaded with PTX (1.2% (w/w) to polymer) was prepared asdescribed above. Then, water of 40 °C (1 mL) was placed on top. After1, 2 and 3 h the concentration of PTX in the supernatant aqueousphase was determined as described below.
2.12. In vitro release of PTX in serum
Gels were loaded with PTX as described in the previous section. Avolume of gel (10 μL) was brought into 200 μL bovine serum, whichwas preheated to 37 °C. The vials were shaken at 37 °C in a waterbath. Every 15 min half of the serum was removed and replacedwith fresh, preheated serum. PTX concentration in the samples wasdetermined as described below.
2.13. Analysis of PTX by UPLC
CH3CN (40 μL)was added to each sample (20 μL). The sampleswerevortexed for 10 seconds and centrifuged for 5 min at 12,000 rpm, 4 °Cto remove precipitated plasma components. PTX amounts were deter-mined by UPLC on a Waters Acquity UPLC system using a Waters HSS
Table 1Characteristics of the synthesized pNIPAm-PEG-pNIPAm copolymers.
[a] at 0.1 mg/mL; [b] Cloud Point; [c] according to DLS at 40 °C; [d] at 25% (w/w) [e] GelPoint; [f] at the GP.
585A.J. de Graaf et al. / Journal of Controlled Release 162 (2012) 582–590
NANOMEDICIN
E
T3 1.7 μm, 2.1×50 mm column at 50 °C. The eluent was CH3CN/H2O/HCOOH 45/55/0.1 (v/v) at a flow rate of 1 mL/min. After each run of4 min the column was cleaned for 1.5 min with CH3CN/H2O/HCOOH90/10/0.1 (v/v) at 1 mL/min (ramp times 0.5 min) and re-equilibratedfor 2.5 min with the starting eluent. Per sample, 7 μL was injected.PTX was detected at a wavelength of 227 nm. Calibration was doneusing standard solutions of PTX (0.05-20 μg/mL) in CH3CN. In everysample set, a recovery test of PTX from spiked serum was included.Observed recoveries were consistently between 90-110%.
2.14. In vivo release of PTX and inhibition of tumor growth
The animal experiment was performed in accordance with localregulations and with consent of the local animal welfare committee.14C cells (human head and neck squamous cell carcinoma cells,1×106 in 0.1 mL cold PBS) were injected s.c. into the flank of femaleCD1-Foxn1 nude mice. When tumors had become palpable (averagetumor volume 37 mm3), the mice were divided over three groups(5 mice per group). Each mouse received an i.p. injection of 200 μLof a 25% (w/w) solution of N16P6N16 at room temperature containing0%, 1.2% or 6% PTX (w/w relative to polymer). These amounts corre-spond to 0, 20 or 100 mg PTX/kg body weight corresponding to 0×,1× or 5× the maximum tolerated dose (MTD) of PTX in mice whenit is given as i.v. injection formulated in cremophor EL/ethanol [43].After 1 h, 5 h, 24 h and 48 h, blood samples (approx. 200 μL) wereobtained via cheek punction (2 mice per group per time point),collected in K4-EDTA tubes and kept on ice. Within 30 minutes, thesamples were centrifuged for 10 min at 10,000 rpm, 4 °C. The super-natant plasma was stored at −20 °C until analysis. PTX concentra-tions in plasma were determined by UPLC as described above. Theequation
ln Ctð Þ ¼ ln C0ð Þ−k⋅t ð1Þ
was fitted by a linear least squares method to the plasma concentra-tions Ct at timepoints t=5 h, 24 h and 48 h of the high dose PTXformulation. From the obtained rate constant k, the half-life of theplasma concentrations t½ was calculated as
t1=2¼ ln 2ð Þ
kð2Þ
The area under the curve (AUC) was calculated using the trapezoi-dal approximation. From the AUC and the given dose D, the bioavail-ability F was estimated using 2.5 L·h-1·kg-1 as a reference value forthe clearance Cl of PTX [49,50]:
F ¼ AUCD
⋅Cl ð3Þ
Tumor growth was monitored daily using digital calipers. By mea-suring the largest diameter l and the smallest diameter b, tumor vol-ume V could be estimated as
V ¼ 12l·b2 ð4Þ
2.15. Tissue homogenization
The excised tissue was transferred into a 2 mL polypropylene tubecontaining zirconia beads. Cold phosphate buffer (100 μL of 0.5 M,pH 7.4) was added and the tube was placed in a Precellys 24 homog-enizer. A program consisting of 3 runs of 20 s at 5000 rpm, with 5 sbetween the runs, was used to homogenize the tissue. After homoge-nization, CH3CN (200 μL) was added, the tube was vortexed and thencentrifuged for 5 min at 12,000 rpm. The supernatant was analyzedas described above.
3. Results and Discussion
3.1. Synthesis of polymers
The characteristics of the triblock polymers are summarized inTable 1. Molecular weights corresponded to the calculated valuesbased on the feed and the dispersities were rather narrow, as expectedfor polymers synthesized by controlled radical polymerization.
3.2. Self-assembly behavior of the polymers
The properties of gels and micelles formed from the synthesizedpolymers are summarized in Table 2.
The CP of N32P6N32 was lower than that of N16P6N16 (33.0 vs35.2 °C in PBS). It has been shown before that the effect of a coupledhydrophilic PEG block on the thermal properties becomes significantfor polymers containing small pNIPAm blocks [38,39]. For both poly-mers, it can be observed that the presence of ions (as in PBS) leads toa decrease in CP of around 2.5 °C due to salting out. The polymersself-assembled into micelles above their CP and above their criticalmicelle concentration, which is 0.03 mg/mL for N16P6N16 [39]. Inwater, micelles of N32P6N32 were not significantly larger than thoseof N16P6N16 (24 vs 23 nm). However, using PBS leads to an increasein hydrodynamic radius Rh, probably due to a decrease in hydrationof the PEG chains due to partial ‘salting out’ leading to an increasedaggregation number per micelle.
When 25% (w/w) polymer solutions were heated, the viscositystarted to increase rapidly at a certain temperature (Tonset,visc, Fig. 1).Upon further heating, at a certain point tan δ≡G”/G’ became smallerthan unity. This temperature is often called the gel point (GP) [51].For both polymers, GP (as well as CP) was below 37 °C in PBS. The de-pendency of GP and Tonset,visc on the polymer composition and diluentwas the same as for the CP, with GP≥CP>Tonset,visc (Table 2). At tem-peratures above 34 °C, G’ and G” of the N32P6N32 gel in PBS sharplydecreased. This may well be explained by its larger hydrophobic char-acter, causing it to partially phase separate, and lose contact with therheometer plates (Fig. 1).
Gels of N16P6N16 were translucent, whereas thosemade of N32P6N32
had a white appearance (Fig. 2), possibly because the “micelles” in thelatter gel are more strongly dehydrated.
012345
AHDO
PEG
pNIPAm CH3
δ (ppm)
012345
BHDO
PEG
pNIPAm CH3
δ (ppm)
Fig. 3. 1HNMR spectra (in D2O) of 25% (w/w) solutions of (A) N16P6N16 and (B) N32P6N32.Dashed traces: 25 °C (shifted 0.3 ppm upfield for clarity), solid traces: 40 °C.
10 20 30 40 50 600.1
1
10
100
1000
10000A
T (°C)
T (°C)10 20 30 40 50 60
0.1
1
10
100
1000
10000B
G',
G"
(Pa)
G',
G"
(Pa)
Fig. 1. Rheograms of 25% (w/w) gels of (A) N16P6N16 and (B) N32P6N32, in PBS as a functionof temperature. Solid lines: storage modulus (G’), dashed lines: loss modulus (G”).
586 A.J. de Graaf et al. / Journal of Controlled Release 162 (2012) 582–590
NANOMEDICIN
E
3.3. Temperature-dependent NMR of gels
1H NMR spectra of the polymer solutions at 25 °C and the gels at40 °C are shown in Fig. 3. As the ratio between the integrals of thePEG peaks and the HDO peaks at 40 °C is equal to the ratio at 25 °C,it can be concluded that in both gels the PEG chains are still fully hy-drated above the GP. In contrast, the signals of the core-formingpNIPAm segments almost completely disappeared above the GP indi-cating that the pNIPAm segments self-assembled in hydrophobic do-mains. For both polymers at 40 °C the PEG peak has a narrow top(caused by a long T2) and a broad base (caused by a short T2) corre-sponding to observations of dilute dispersions of pNIPAm-PEG-pNIPAm flower-like micelles. The narrow top and broad base havebeen attributed to the PEG segments which are distal and proximal(respectively) to the cores of the flower-like micelles [39].
Fig. 2. Gels formed by heating 25% (w/w) solutions of N16P6N16 (left) and N32P6N32
(right) in water to 40 °C.
3.4. Imaging the release of micelles from in situ formed gels
Upon slow injection of the polymer solutions into the pre-heated(40 °C; above the LCST of the NIPAm blocks) measurement chamberof the Nanosight apparatus, instantaneous gel formation was ob-served. The intensity of scattered laser light was monitored andFig. 4 shows that after 10 min scattering was observed close to thegel edge, slowly increasing and progressing further away from thegel in time. This observation indicates that micelles are diffusinginto the aqueous phase that is in direct contact with the gel. Thesize distribution of the micelles was determined from their Brownianmotion once they reached the detection spot (Fig. 4B, C) and showedhomogeneous distributions of particles with average diameters of86 nm (N16P6N16) and 91 nm (N32P6N32). Converted to hydrodynam-ic radius, this corresponds to Rh=43 and 46 nm, i.e. about 2 timeslarger than the values found for dilute, heat-shocked dispersions ofmicelles. This finding indicates that the released micelles are tosome extent ‘bridged’, which is a normal phenomenon for flower-like micelles unless they are observed at very low concentrations(such as in the DLS experiment). In addition, the lower threshold ofthe NanoSight system (approximately 40–60 nm) may have attrib-uted to a higher average size distribution compared to DLS (Table 2)as the part of the population which is below this threshold is notdetected and is therefore not included in the calculation of themean size.
3.5. Release kinetics of micelles from gels measured with DLS
When gels were brought into PBS (pre-heated to 37 °C), release ofmicelles could be observed by eye after approximately 1 h (Fig. 5).
When the medium was refreshed every 15 min, it was observedby DLS that every time newmicelles were released. The amount of re-leased micelles (as indicated by the ratio of scattered versus incidentlight) increased during the first 1.5 h (6 medium refreshings) and
Fig. 4. Release of micelles from an in situ formed gel of N16P6N16 studied by nanoparti-cle tracking analysis. (A) Photographs showing the increased scattering of the laserbeam in time (from left to right). The hydrogel is positioned on the far left of the mea-surement chamber, outside the visible area. The area on the right side is the entry pointof the laser beam. (B) Microscopic image of the light scattering by individual micellesreleased from the in situ formed gel. The apparent particle size in this image reflectsthe scattering intensity of the particles rather than their actual size. (C) Size distribu-tion of the released micelles as calculated by nanoparticle tracking analysis software.
Fig. 5. Release of micelles from the N16P6N16 gel (at the bottom of the tube) after 1 h inPBS at 37 °C. The Rayleigh scattering which is observed above the gel indicates thepresence of micelles.
0.0 0.5 1.0 1.5 2.0 2.50.00
0.01
0.02
0.03
0.04
0.05
0.06A
Sca
tter
ing
rat
io (
a.u
.)
0.0 0.5 1.0 1.5 2.0 2.50
50
100
150
200B
t (h)
t (h)
Rh
Fig. 6. Characterization of the scattering ratio (A) and hydrodynamic radius (B) of themicelles thatwere released from the hydrogel in PBS at 37 °C. Squares: N16P6N16, diamonds:N32P6N32. Data are averages of two independent samples.
587A.J. de Graaf et al. / Journal of Controlled Release 162 (2012) 582–590
NANOMEDICIN
E
then the release leveled off (Fig. 6A). On the other hand, the size ofthe released micelles remained constant (Fig. 6B).
Micelles released from the N16P6N16 gel scattered more light thanthose from the N32P6N32 gel, whereas their sizes were similar. Assum-ing similar densities of the micelles, this would indicate a faster re-lease from the N16P6N16 gel than from the N32P6N32 gel. To release amicelle from a gel, all ‘bridging’ chains need to be extracted from itscore (or from its neighbor's core). It can be imagined that extractinga pNIPAm block of 16 kDa is easier than extracting a block of32 kDa, which would explain the difference in release rate.
When an N16P6N16 gel was kept in a DLS cuvette overnight withoutrefreshing the medium, an equilibrium scattering intensity (0.05 a.u.on the scale of Fig. 6)was reachedwithin an hour, afterwhich the scat-tering intensity did not increase significantly. Furthermore, the gel
was still present after overnight incubation. These observations ledto the conclusion that the release of micelles from the hydrogels isan equilibrium process, which can only be driven completely to theside of micelles by continuously refreshing the release medium.
3.6. PTX encapsulation and release
PTX was loaded into the gels by forming PTX-loaded micelles atrelatively low polymer concentration, lyophilizing these and subse-quently swelling them in a small volume of water (25% (w/w) poly-mer). Due to its hydrophobic character, PTX likely partitions intothe hydrophobic pNIPAm domains of the gels. The N16P6N16 gel,which showed the fastest release in the DLS experiment, was loadedwith 1.2% (w/w to polymer) PTX and was incubated in water at40 °C. The concentration of PTX in the water phase above the gelafter 1 h was only 0.6 μg/mL, which corresponds to its maximum
Fig. 8. Plasma PTX levels of mice injected i.p. with the N16P6N16 hydrogel at a dose of100 mg/kg PTX. The dotted line indicates the detection limit of 0.05 μg/mL. Data pointsrepresent individual mice (2 per timepoint); the curve is drawn through the averagedvalues.
588 A.J. de Graaf et al. / Journal of Controlled Release 162 (2012) 582–590
NANOMEDICIN
E
solubility in water [50,52,53]. After 2 h, however, the concentrationhad increased to 5.7 μg/mL, i.e. almost 10 times the aqueous solubilityof PTX. Clearly, the micelles which were released from the gel solubi-lized PTX which was also released from the gel.
Two mechanisms can be envisioned by which the PTX ends upin the micelles. One possible mechanism is that intact micellescontaining PTX bud off from the hydrogel. Another mechanism isthat unimers and PTX dissolve separately from the hydrogel, followedby self-assembly of the unimers into micelles once the concentrationexceeds the CMC. The dissolved PTX then partitions into these mi-celles. For in vivo application, the first mechanism would be desiredas it would protect PTX against aggregation or interaction with plas-ma components. The present data can be explained by both mecha-nisms. The observation of a lag time in the release is best explainedby mechanism 2, as it would take some time before the CMC isreached only after which PTX can be solubilized in micelles. On theother hand, the observation that the released particles are largerthan micelles prepared from a dilute solution, may indicate thatthey are in fact very small pieces of hydrogel (each consisting of afew bridged micelles) that are released from the hydrogel in intactform.
3.7. In vitro release of PTX in serum
Fig. 7 shows the cumulative release of PTX from hydrogels made ofthe two polymers. Upon regular sampling and refreshing the releasemedium it can be observed that the hydrogels of N32P6N32 showedthe slowest release, which is in line with the DLS measurementson the release of micelles. N16P6N16 hydrogels with different PTXloadings showed a nearly complete release in 4 h (i.e. after 15 timesrefreshing the medium), whereas the release from N32P6N32 gelswas much slower, amounting to only 10% release over the 4 h timeperiod. After this time, no N16P6N16 hydrogels could be observed any-more, while the N32P6N32 gels were still present. The N16P6N16 gelwith high PTX loading showed faster release kinetics than the samegel with lower PTX loading. The reason for this difference is unclear.The release behavior of the N16P6N16 gel with the low PTX loadingis nearly zero-order, indicating an erosion-controlled process. Therewas hardly any release in a control experiment in which sampleswere taken without refreshing the medium but only pipetting it upand down (Fig. 7, diamonds): the final concentration of PTX was1.2 μg/mL, whereas its solubility in serum is at least 5 μg/mL [49]. Atany time point, the PTX concentrations in serum that was in contactwith the N32P6N32 gels were also well below this value. These data in-dicate that (i) PTX is stably entrapped in the hydrogels, from which itis hardly extracted by serum in an equilibrium situation, (ii) the fastera hydrogel releases micelles in the non-equilibrium situation, thefaster it also releases PTX, and (iii) both PTX and micelles are released
0 1 2 3 40
25
50
75
100
t (h)
% o
f en
cap
sula
ted
PT
Xre
leas
ed
Fig. 7. In vitro release of PTX fromhydrogels in serumat 37 °C. Squares: N16P6N16, triangles:N32P6N32, closed symbols: 6.0% loading, open symbols: 1.2% loading. Diamonds representthe release from N16P6N16 1.2% loading when the medium was not refreshed. Values areindicated as mean±SD of three replicates.
simultaneously, in an erosion-controlled fashion which is governedby the rate at which the release medium is refreshed.
3.8. In vivo release of PTX and inhibition of tumor growth
The N16P6N16 gels were selected for an in vivo efficacy study, asit was anticipated that the low flow rate of intraperitoneal fluidcombined with the low release rate of N32P6N32 would lead toundetectably low PTX levels in plasma and thus too low (if any) ther-apeutic efficacy. The hydrogels were well tolerated by the mice, eventhe formulation which contained 100 mg/kg PTX (i.e. five times theMTD of PTX formulated in Cremophor EL/ethanol, when given i.p. ori.v.) [43,54,55]. There were no signs of acute systemic toxicity eitherdue to PTX or to any potentially accumulated polymer.
Fig. 8 shows the levels of PTX in the blood plasma of mice afterinjection of PTX-containing hydrogels. It can be concluded fromthese data that PTX (possibly loaded in micelles) is able to be carriedby the lymphatic flow via the lymph nodes to the circulation. It hasbeen shown before that both free PTX and PTX in micelles or othernanoparticles are able to be gradually transported (within severalhours) from the intraperitoneal cavity to the circulation, whereasPTX in microparticles is hardly transported [44,45]. For at least 48 hthe plasma PTX concentrations stayed well above the minimal effec-tive concentration which has been reported to be around 10 ng/mL(in vitro) [33–36]. From the data points at 5, 24 and 48 h thehalf-life t½ of the plasma concentrations was calculated to be 7.4 h,about twice as long as literature values for PTX/Cremophor i.p. inmice (range 2.9 - 3.7 h) [35,50]. This finding can be explained by ei-ther or both of the two following explanations. The first possible ex-planation is that released PTX is present in long-circulating micelles,which increases t½ due to reduced clearance. The second possible ex-planation is that the hydrogel gave a continuous release of PTX duringthis period. The apparent half-life t½ is then the result of the balancebetween release and clearance.
Using the trapezoidal method, an AUC0-48h of 45 μg·mL-1·hwas calculated. In order to calculate the bioavailability of PTX, theobtained AUC was compared to literature pharmacokinetic data onPTX/Cremophor formulations at relatively low doses, which in factproduced similar maximal plasma concentrations as found in thepresent study [43,49]. In this way, a bioavailability of 106% was calcu-lated for the 100 mg/kg formulation. Interestingly, literature valuesfor the bioavailability after i.p. injection of PTX formulated inCremophor/ethanol or in nanoparticles lie between 5% and 37%[35,42–45]. The high bioavailability found in our study may be relatedto the fact that the PTX was shown to be on one hand stably incorpo-rated in the hydrogel, preventing aggregation or diffusion into localtissues (mainly intestines, leading to toxicity [45]) before it can
0 10 20 30 400
25
50
75
100A
t (days)
% o
f m
ice
0 20 1000
250
5002250
2500B
Dose (mg/kg)
rela
tive
tu
mo
r vo
lum
e (%
)
Fig. 9. (A) Kaplan-Meier plot showing the percentage of mice of which the tumor sizehad increased less than 300%. Dotted line: 0 mg/kg, dashed line: 20 mg/kg, solid line:100 mg/kg. (B) Distribution of the relative tumor sizes of the three groups at 20 daysafter the start of treatment. Indicated is the median of each group.
589A.J. de Graaf et al. / Journal of Controlled Release 162 (2012) 582–590
NANOMEDICIN
E
reach the circulation, and on the other hand to be completely releasedover time.
The plasma levels of the 20 mg/kg group at 1 and 5 h amounted to13% of the plasma concentrations of the 100 mg/kg group. The plasmalevels of the 20 mg/kg group dropped below the detection thresholdafter 24 h so no further pharmacokinetic data could be obtained fromthis group.
The effect of the PTX-loaded hydrogels on the growth of the tu-mors is shown in Fig. 9. A 300% increase in tumor volume (averagetumor volume 144 mm3) was taken as the threshold point for theKaplan-Meier plot in Fig. 9A, whereas Fig. 9B displays the relativetumor volume after 20 days. No significant difference could be ob-served between the 20 mg/kg group and the control. Apparently,the plasma PTX concentrations that were reached in this groupwere insufficiently high, whether or not they were sustained for along time. In the 100 mg/kg group, however, no mice showed anymeasurable tumor growth for 40 days. The fact that this inhibitory ef-fect is observed, although the plasma concentrations of PTX are quitelow after 48 h, may indicate that the sustained release of PTX was in-deed beneficial, in line with already existing evidence and theories onthe advantages of metronomic dosing [28–32].
4. Conclusion
Taken together, this paper shows that thermosensitive triblock co-polymer hydrogels slowly interconverted into flower-like micelleswhen in contact with an aqueous environment. A hydrophobic drug(PTX) could be released in a sustained manner together with thesemicelles and be solubilized in the micelles. The release rate of micellesand PTX was governed by the refreshing frequency of the release me-dium, and a nearly quantitatively release was obtained both in vitroand in vivo. In vivo, this system gave a sustained release of PTX overat least 48 h and allowed to give at least a 5 times higher dose ofPTX than the maximum tolerated dose of the commonly used
formulation in Cremophor EL/ethanol. Although the exact mechanismof the release remains to be elucidated, as well as the question wheth-er or not the released PTX circulates in micelles in vivo, the concept ofa hydrogel which interconverts into micelles has been proven andmay find application as a depot formulation for metronomic chemo-therapy with poorly soluble cytostatic drugs.
References
[1] Z. Zhou, B. Chu, Macromolecules 27 (1994) 2025–2033.[2] G. Chang, C. Li, W. Lu, J. Ding, Macromol. Biosci. 10 (2010) 1248–1256.[3] S. Honda, T. Yamamoto, Y. Tezuka, J. Am. Chem. Soc. 132 (2010) 10251–10253.[4] S. Hua, Y. Li, Y. Liu, W. Xiao, C. Li, F. Huang, X. Zhang, R. Zhuo, Macromol. Rapid
Commun. 31 (2010) 81–86.[5] E.S. Lee, K.T. Oh, D. Kim, Y.S. Youn, Y.H. Bae, J. Controlled Release 123 (2007) 19–26.[6] S.J. Bae, J.M. Suh, Y.S. Sohn, Y.H. Bae, S.W. Kim, B. Jeong, Macromolecules 38
(2005) 5260–5265.[7] A.N. Semenov, J. Joanny, A.R. Khokhlov, Macromolecules 28 (1995) 1066–1075.[8] T. Fumihiko, J. Non-Cryst. Solids 307–310 (2002) 688–697.[9] P. Kujawa, H. Watanabe, F. Tanaka, F.M. Winnik, Eur. Phys. J. E Soft Matter 17
1760–1773.[11] V.P. Torchilin, AAPS J. 9 (2007) E128–E147.[12] E. Blanco, C.W. Kessinger, B.D. Sumer, J. Gao, Exp. Biol. Med. (Maywood) 234
(2009) 123–131.[13] V.P. Torchilin, Pharm. Res. 24 (2007) 1–16.[14] V.P. Torchilin, Cell. Mol. Life Sci. 61 (2004) 2549–2559.[15] X. Hu, X. Jing, Expert Opin. Drug Deliv. 6 (2009) 1079–1090.[16] C. Oerlemans, W. Bult, M. Bos, G. Storm, J.F. Nijsen, W.E. Hennink, Pharm. Res. 27
(2010) 2569–2589.[17] S.R. van Tomme, G. Storm, W.E. Hennink, Int. J. Pharm. 355 (2008) 1–18.[18] N.A. Peppas, P. Bures, W. Leobandung, H. Ichikawa, Eur. J. Pharm. Biopharm. 50
(2000) 27–46.[19] J.K. Kim, Y.-H. Won, K.S. Lim, Y.-H. Kim, Pharm. Res. 29 (2012) 525–534.[20] Y. Tang, C.L. Heaysman, S. Willis, A.L. Lewis, Expert Opin. Drug Deliv. 8 (2011)
1141–1159.[21] M. Chen, H. Tsai, C. Liu, S. Peng, W. Lai, S. Chen, Y. Chang, H. Sung, Biomaterials 30
(2009) 2102–2111.[22] L. Wei, C. Cai, J. Lin, T. Chen, Biomaterials 30 (2009) 2606–2613.[23] C. Li, J. Madsen, S.P. Armes, A.L. Lewis, Angew. Chem. Int. Ed. 45 (2006) 3510–3513.[24] J. Madsen, S.P. Armes, K. Bertal, H. Lomas, S. MacNeil, A.L. Lewis, Biomacromolecules
9 (2008) 2265–2275.[25] F. Alexis, E. Pridgen, L.K. Molnar, O.C. Farokhzad, Mol. Pharm. 5 (2008) 505–515.[26] Y. Matsumura, H. Maeda, Cancer Res. 46 (1986) 6387–6392.[27] H. Maeda, J. Wu, T. Sawa, Y. Matsumura, K. Hori, J. Controlled Release 65 (2000)
271–284.[28] R.S. Kerbel, B.A. Kamen, Nat. Rev. Cancer 4 (2004) 423–436.[29] R. de Souza, P. Zahedi, E.H. Moriyama, C.J. Allen, B.C. Wilson, M. Piquette-Miller,
Mol. Cancer Ther. 9 (2010) 1820–1830.[30] G. Gasparini, Lancet Oncol. 2 (2001) 733–740.[31] T. Browder, C.E. Butterfield, B.M. Kräling, B. Shi, B. Marshall, M.S. O'Reilly, J. Folkman,
Cancer Res. 60 (2000) 1878–1886.[32] J. Nabholtz, J. Gligorov, Expert Opin. Pharmacother. 6 (2005) 1073–1094.[33] N.M. Lopes, E.G. Adams, T.W. Pitts, B.K. Bhuyan, Cancer Chemother. Pharmacol. 32
(1993) 235–242.[34] M.A. Jordan, K. Wendell, S. Gardiner, W. Brent Derry, H. Copp, L. Wilson, Cancer
Res. 56 (1996) 816–825.[35] F. Innocenti, R. Danesi, A. di Paolo, C. Agen, D. Nardini, G. Bocci, M. del Tacca, Drug
4093–4100.[37] D. Belotti, V. Vergani, T. Drudis, P. Borsotti, M.R. Pitelli, G. Viale, R. Giavazzi, G.
Taraboletti, Clin. Cancer Res. 2 (1996) 1843–1849.[38] M. Teodorescu, I. Negru, P.O. Stanescu, C. Drăghici, A. Lungu, A. Sârbu, React.
Funct. Polym. 70 (2010) 790–797.[39] A.J. de Graaf, K.W.M. Boere, J. Kemmink, R.G. Fokkink, C.F. van Nostrum, D.T.S.
Rijkers, J. van der Gucht, H. Wienk, M. Baldus, E. Mastrobattista, T. Vermonden,W.E. Hennink, Langmuir 27 (2011) 9843–9848.
[40] G. Bajaj, Y. Yeo, Pharm. Res. 27 (2010) 735–738.[41] P. Francis, E. Rowinsky, J. Schneider, T. Hakes, W. Hoskins, M. Markman, J. Clin.
Oncol. 13 (1995) 2961–2967.[42] D. Soma, J. Kitayama, H. Ishigami, S. Kaisaki, H. Nagawa, J. Surg. Res. 155 (2009)
142–146.[43] J.L. Eiseman, N.D. Eddington, J. Leslie, C. MacAuley, D.L. Sentz, M. Zuhowski, J.M.
Kujawa, D. Young, M.J. Egorin, Cancer Chemother. Pharmacol. 34 (1994) 465–471.[44] H. Gelderblom, J. Verweij, D.M. van Zomeren, D. Buijs, L. Ouwens, K. Nooter, G.
Stoter, A. Sparreboom, Clin. Cancer Res. 8 (2002) 1237–1241.[45] M. Tsai, Z. Lu, J. Wang, T. Yeh, M. Wientjes, J. Au, Pharm. Res. 24 (2007) 1691–1701.[46] P. Zahedi, J. Stewart, R. de Souza, M. Piquette-Miller, C. Allen, J. Controlled Release
158 (2012) 379–385.[47] M. Ciampolini, N. Nardi, Inorg. Chem. 5 (1966) 41–44.[48] D. Neradovic, O. Soga, C.F. van Nostrum, W.E. Hennink, Biomaterials 25 (2004)
2409–2418.
590 A.J. de Graaf et al. / Journal of Controlled Release 162 (2012) 582–590
NANOMEDICIN
E
[49] A. Sparreboom, O. van Tellingen, W.J. Nooijen, J.H. Beijnen, Cancer Res. 56 (1996)2112–2115.
[50] A. Sparreboom, O. van Tellingen, W.J. Nooijen, J.H. Beijnen, Anticancer Drugs 9(1998) 1–17.
[51] Officially, the GP is defined as the temperature where both G’ and G” are frequencyindependent. However, the method used here (intersection point of G’ and G”) is acommonly used approximation.
[53] C.S. Swindell, N.E. Krauss, S.B. Horwitz, I. Ringel, J. Med. Chem. 34 (1991) 1176–1184.[54] G. Bajaj,M.R. Kim, S.I.Mohammed, Y. Yeo, J. Controlled Release 158 (2011) 386–392.[55] E. Auzenne, S.C. Ghosh, M. Khodadadian, B. Rivera, D. Farquhar, R.E. Price, M.
Ravoori, V. Kundra, R.S. Freedman, J. Klostergaard, Neoplasia 9 (2007) 479–486.
