Cancer Letters 334 (2013) 346–353

Contents lists available at SciVerse ScienceDirect

Cancer Letters

journal homepage: www.elsevier .com/ locate/canlet

Original Research Article

Polyisoprenoyl gemcitabine conjugates self assemble as nanoparticles,useful for cancer therapy

0304-3835/$ - see front matter � 2012 Elsevier Ireland Ltd. All rights reserved.http://dx.doi.org/10.1016/j.canlet.2012.08.023

⇑ Corresponding author. Address: Univ Paris-Sud, Faculté de Pharmacie, CNRSUMR 8612, Institut Galien Paris-Sud, 5, rue J.B. Clément, 92296 Châtenay-MalabryCedex, France. Tel.: +33 (0) 1 46 83 53 96; fax: +33 (0) 1 46 83 59 46.

E-mail addresses: andrey.maksimenko@u-psud.fr (A. Maksimenko), patrick.couvreur@u-psud.fr (P. Couvreur).

Andrei Maksimenko, Julie Mougin, Simona Mura, Eric Sliwinski, Elise Lepeltier, Claudie Bourgaux,Sinda Lepêtre, Fatima Zouhiri, Didier Desmaële, Patrick Couvreur ⇑Univ Paris-Sud, Faculté de Pharmacie, 5, rue J.B. Clément, 92296 Châtenay-Malabry Cedex, FranceCNRS UMR 8612, Institut Galien Paris-Sud, 5, rue J.B. Clément, 92296 Châtenay-Malabry Cedex, France

a r t i c l e i n f o

Keywords:Nanomedicine

Pancreatic cancerProdrugGemcitabineSqualeneTerpene

a b s t r a c t

A series of new polyisoprenoyl prodrugs of gemcitabine, which can be formulated as nanoassemblies aredescribed. These prodrugs were designed to improve gemcitabine efficacy and to overcome the limita-tions due to the systemic toxicity of this anticancer compound. In vitro biological assessment showed thatthese polyisoprenoyl gemcitabine nanoassemblies displayed notable cytotoxicity on several cancer celllines, including murine melanoma cell line B16F10, human pancreatic carcinoma cell line MiaPaCa-2,human lung carcinoma cell line A549 and human breast adenocarcinoma cell line MCF7. Interestingly,it was observed that the anticancer efficacy of these nanoassemblies was dependant on the size of poly-isoprenoyl moiety. The polyisoprenoyl prodrug of gemcitabine containing three isoprene units (2d) wasthe more active on all the cancer cell lines tested. The antitumor efficacy of the nanoassemblies (NAs)constructed with the most active prodrug 2d was further evaluated on a human pancreatic (MiaPaCa-2) carcinoma xenograft model in mice. The prodrug 2d NAs showed an increased antitumor efficacy ascompared to free gemcitabine or to squalene-gemcitabine (SQ-gem, 2a) nanoassemblies. Interestingly,MiaPaCa-2 tumors that did not respond to gemcitabine were inhibited by 76% after treatment with pro-drug 2d NAs, whereas SQ-gem-treated MiaPaCa-2 tumor xenografts decreased only by 41% compared tosaline or to gemcitabine-treated mice. Together, these findings demonstrated that the modulation of thelength of nanoassemblies polyisoprenoyl moiety made tumor cells more sensitive to gemcitabine treat-ment without flagrant toxicity, thus providing a significant improvement in the drug therapeutic index.

� 2012 Elsevier Ireland Ltd. All rights reserved.

1. Introduction

Cancer is currently the second leading cause of death in Europe,while it shows probably the highest clinical complexity [1]. Nano-medicine bears the potential to provide an effective answer to thecomplexity of the disease as it offers additional therapeuticoptions, compared to present conventional therapy [2]. New tai-lor-made nanodevices have the potential to revolutionize medicinebecause of their demonstrated ability to target the diseased organsand tissues at the molecular and cellular levels in a controlledmanner [3–6]. Nanotechnology now plays an important part indeveloping drug delivery methods, allowing the emergence ofnew treatments with an improved efficacy and/or reduced toxicity[7]. These new nanodevices can be tailor-made with particularfunctionalities and applications as a result of progresses in material

sciences, as well as in the preparation of colloidal systems with de-fined properties [8,9]. Nanomedical users and developers are, how-ever, faced with the challenge of balancing the medical and societalbenefits and risks associated with nanotechnology [10]. Up to now,the adequacy of available tools, such as physiologically-basedpharmacokinetic modeling or predictive structure–activity rela-tionships, in assessing the toxicity and risk associated with specificnanomaterials is unknown [10]. Successful development of futurenanomedicines requires a consolidated information base to selectthe optimal nanomaterial, understanding the toxicology and po-tential side effects associated with candidate materials for medicalapplications [10–12].

In this context, we have discovered the ‘‘squalenoylation’’ tech-nology, consisting in the linkage of squalene, a natural and well tol-erated triterpene, with anticancer and antiviral nucleosideanalogues [4,13]. Remarkably, the resulting bioconjugates werefound to be able to self-organize spontaneously as nanoparticlesin water whatever the nucleoside analogue used. This generic ap-proach has advantages of drug potentiation and minimal drugresistance, as the loading of the drug (approximately 50%) isradically improved compared to all the other currently available

A. Maksimenko et al. / Cancer Letters 334 (2013) 346–353 347

nanomedicines. This concept has been applied to gemcitabine, ananticancer compound, active against various solid tumors. Gemcit-abine has, however, some major limitations, such as a short biolog-ical half-life, a poor diffusion into cells and the induction ofresistances, which all restrict its therapeutic potential [14,15]. Toovercome these drawbacks, we made a covalent linkage betweengemcitabine and squalene to obtain 4-(N)-Tris-nor-squalenoyl-gemcitabine (SQ-gem). The resulting nanoassemblies (100 nmdiameter) were formed by the hexagonal molecular packing ofthe molecules as shown by X-ray diffraction and cryo-TEM exper-iments [13]. Moreover, these SQ-gem nanoassemblies with a drugloading of almost 50% (w/w) exhibited impressively greater anti-cancer activity than gemcitabine itself against either solid subcuta-neously grafted tumours (Panc-1, L1210 and P388) or aggressivemetastatic leukemia (leukemia L1210 wt, P388 and RNK-16 LGL)[4,16,17]. It also appeared that only SQ-gem nanoassemblies treat-ment resulted in long-term survivals at both equivalent and equi-toxic doses. Additional in vitro and in vivo studies have attributedthe impressive anticancer activity of this nanomedicine to: (i) thesignificantly higher deposition of SQ-gem nanoassemblies inspleen, liver and lungs, the major metastatic organs, (ii) the betterintracellular diffusion of SQ-gem, (iii) the selective activation of theprodrug by cathepsin B, a major enzyme in cancer cells and (iv) theincreased phosphorylation of gemcitabine when delivered as SQ-gem nanoassemblies. In the light of these results, the conceptwas also applied to other non nucleoside bioactive molecules, suchas siRNA, paclitaxel and penicillin [18–20].

The aim of the current study was to enlarge this new andground-breaking concept by varying the nature of the polyterpeneused for the drug conjugation. It was intended first to identify theinfluence of the isoprene length on the ability (or disability) of theresulting bioconjugates to form nanoassemblies and second toemphasize the impact of terpene’s nature on the in vitro andin vivo pharmacological activity, using the gemcitabine as a drugmodel. Thus, we focused on identifying the requisite chemicalfunctions in 1,10,2-tris-nor-squalenic acid which are really respon-sible for nanoassemblies formation and, we concentrated on themodulation of the length of the terpenoyl side chain, by synthesisof various polyisoprenic acids that will be coupled to gemcitabine.Each conjugate was submitted to nanoprecipitation conditions inorder to assess the minimal requisite size of the side chain. Basedon the biological activity of studied compounds in vitro and ontheir optimal structure for nanoassemblies formation, we selectedthe lead molecule which was evaluated on human pancreatic (Mia-PaCa-2) carcinoma xenograft model in mice.

2. Materials and methods

2.1. Reagents

Gemcitabine was purchased from Sequoia Research Product Ltd. (UK). D-(+)-Tre-halose dihydrate, squalene, 3-[4,5-dimethylthiazol-2-yl]-3,5-diphenyl tetrazoliumbromide (MTT), ethylchloroformate, tetra-butylammonium fluoride, tert-butyl dim-ethylchlorosilane, methanesulfonyl chloride, 20-deoxycitidine, sodium acetate andsodium hydride were purchased from Sigma–Aldrich Chemical Co. (St. Quentin Fal-lavier, France). Poly(ethylene glycol) monomethyl ether (Mn, 2000 g mol�1) waspurchase from Fluka (France). Tetrahydrouridine was purchased from Merck Chem-icals Ltd. (Nottingham, UK). All solvents were purchased from Carlo Erba (Val-de-Reuil, France). Annexin V-FITC apoptosis assay kit and cell culture reagents werepurchased from Invitrogen-Life Technologies (Cergy Pontoise, France).

2.2. Synthesis, purification, and characterization of the polyisoprenoyl gemcitabineprodrugs

The synthesis of squalenoyl-gemcitabine 2a or 1,10 ,2-tris-nor-squalenic acid 3ahas been previously reported (Fig. 1) [4]. Bioconjugates 2b–2g were obtained byreaction of the corresponding polyisoprenic acids 3b–3g with 30 ,50-bis TBDMS-pro-tected gemcitabine, followed by conventional tetrabutylammonium fluoride (TBAF)deprotection. Accordingly, the requisite acid was reacted with ethyl chloroformate

in THF at 0 �C to give the intermediate anhydride which was directly reacted withthe TBDMS-protected gemcitabine in THF for 12 h using DMAP as catalyst, followedby TBAF deprotection. The crude products were purified by column chromatogra-phy on silica gel providing the expected conjugate in fair yields (Table 1) (full syn-thetic details and characterization of the polyisoprenic acids 3b–3g and prodrugs2b–2g can be found in Supplementary material).

The synthesis of squalene coupled to polyethyleneglycol (SQ-PEG, Trisnorsqua-lene-MePEG) has been achieved in 60% yield as previously reported [21] byalkylation of the sodium alkoxide of MePEG with 1,10 ,2-trisnorsqualenylmethanesulfonate.

2.3. Preparation and characterization of the squalenoyl (2a) and polyisoprenoyl (2b–2g)gemcitabine nanoassemblies (NAs)

Two milligram of compounds 2a, 2b, 2c, 2e, 2f and 2g or 2 mg of compound 2dcompleted with 400 lg of SQ-PEG (20%, w/w) were dissolved in THF (500 lL, com-pounds 2a, 2c–2g) or ethanol (1 mL, compound 2b). The resulting solutions of com-pounds 2b and 2c were added drop-wise under stirring (1000 rpm) into 1 mL of anaqueous solution of trehalose (10%). Other solutions (i.e. 2a, 2d, 2e, 2f and 2g) wereadded drop-wise under stirring (1000 rpm) into 1 mL of distilled water. The forma-tion of the polyisoprenoyl gemcitabine nanoassemblies occurred spontaneously.THF or ethanol were then completely evaporated using a Rotavapor� at 37 �C undervacuum to obtain aqueous suspensions of pure polyisoprenoyl gemcitabine nanoas-semblies (final concentration 2 mg/mL). The samples of nanoassemblies 2a, 2b, 2e–2g were stored at 5 �C whereas 2c and 2d NAs were maintained at 37 �C beforein vitro and in vivo experiments. Nanoassemblies made of 1,10 ,2-tris-nor-squalenic(3a) or other (3b–3g) corresponding isoprenic acids alone were prepared in a sim-ilar manner. The final concentration of the aqueous suspension of squalenic orfarnesylacetic acids was 2 mg/mL.

The size of the nanoassemblies was determined at 20 �C (for compounds 2a, 2b,2e–2g) or at 37 �C (for compounds 2c and 2d) by quasi-elastic light scattering(QELS) with a nanosizer (Zêtasizer Nano ZS Malvern; Malvern Instruments SA, Or-say, France) after 1:10 dilution in water. All the polyisoprenoyl gemcitabine nano-assemblies prepared in conditions as described above gave stable nanoassembliesof 112–270 nm (PDI 0.2).

The morphology of the different polyisoprenoyl gemcitabine nanoassemblieswas examined by cryomicroscopy (cryo-TEM). Briefly, one drop (5 lL) of the poly-isoprenoyl gemcitabine nanoassemblies suspension (2 mg/mL) was deposited ontoa perforated carbon film mounted on a 200-mesh electron microscopy grid. Most ofthe drop was removed with a blotting filter paper and the residual thin filmsremaining within the holes were vitrified after immersion in liquid ethane. Thespecimen was then transferred using liquid nitrogen to a cryo-specimen holderand observed using a JEOL FEG-2010 electron microscope.

2.4. Drug release

To determine the kinetics of gemcitabine release from polyisoprenoyl gemcita-bine nanoassemblies, 100 lL of nanoassemblies 2a–2d, 2f and 2g (1 mg/mL) wereadded to 900 lL of heat-inactivated fetal bovine serum (FBS) (56 �C, 30 min) supple-mented with 200 lg/mL tetrahydrouridine (THU). Individual vials were used for eachtime point. The reaction mixture was incubated at 37 �C, and aliquots (100 lL) ofincubation medium were removed at predetermined time points (0, 2, 4, 8 and24 h), spiked with 10 lL of 200 lM 20-deoxycytidine solution (Internal Standard,IS) before addition of 1 mL of a mixture of acetonitrile/methanol (90/10, v/v) andultracentrifugated (15000 g, 20 min, 4 �C). Supernatant was then evaporated todryness under a nitrogen flow at 30 �C. The released drug was quantified by re-verse-phase HPLC (Waters, Milford, MA 01757, USA) with a C18 column as describedpreviously [22,23]. Briefly, the chromatographic system consisted of a Waters 1525Binary LC pump, a Waters 2707 Autosampler, a C18 Uptisphere column (3 lm,150 � 4.6 mm; Interchim), HPLC column temperature controllers (model 7950 col-umn heater and chiller; Jones Chromatography, Lakewood, CO), and a Waters 2998programmable photodiode-array detector. The HPLC column was maintained at30 �C. Detection was monitored at 270 nm. The HPLC mobile phase was metha-nol:water (5:95, by vol, with 0.05 M sodium acetate in water, pH 5, solvent A) andmethanol:water (97:3, by vol, with 0.05 M sodium acetate in water, pH 5, solventB). The residues were dissolved in 100 lL of solvent A and elution was carried outat a flow rate of 0.8 mL/min isocratically for 8 min with solvent A followed by a 1-min linear gradient to 100% solvent B. This was followed by a 16-min hold at solventB, and a 1-min linear gradient back to 100% solvent A. The system was held at 100%solvent A for 14 min for equilibration back to initial conditions.

2.5. Cell culture

Murine melanoma cell line B16F10, human pancreatic carcinoma cell line Mia-PaCa-2, lung carcinoma cell line A549 and breast adenocarcinoma cell line MCF7were obtained from the American Type Culture Collection and maintained as recom-mended. Briefly, A549 cells were maintained in F12-K medium. MiaPaCa-2 andB16F10 cells were grown in Dulbecco’s minimal essential medium (DMEM) – Gluta-mine medium. MCF7 cells were cultured in a mixture of DMEM/Ham’s F12 (1:1). All

Fig. 1. Structure of gemcitabine and of the polyisoprenoyl gemcitabine prodrugs.

Table 1Colloidal characteristics of the of polyisoprenoyl gemcitabine prodrugs.

Compound Number ofisoprenyl units

Overall chemicalyield (%)

CalculatedlogPc

Average size ofnanoassemblies (nm)

Polydispersityindex (PdI)

Drug loading (%)

SQ-gem (2a) 5 – 6.90 125 0.10 40.72b 1 74 0.79 250a 0.14 70.42c 2 65 2.32 145a 0.09 59.62d 3 53 3.85 270b 0.09 43.0d

2e 4 32 5.37 112 0.33 45.52f 5 64 6.90 116 0.14 40.72g 6 54 8.43 156 0.34 36.8

a Nanoassemblies were precipitated into 10% aqueous trehalose solution.b Co-nanoprecipitation of polyisoprenyol gemcitabine prodrugs with SQ-PEG into water.c ChemDrawBio Ultra 12.0 (CambridgeSoft).d Calculated taking account of the SQ-PEG added.

348 A. Maksimenko et al. / Cancer Letters 334 (2013) 346–353

media were supplemented with 10% heat-inactivated fetal bovine serum (FBS)(56 �C, 30 min), penicillin (100 U/mL) and streptomycin (100 lg/mL). Medium forMiaPaCa-2 cell line was supplemented additionally with 2.5% heat-inactivated horseserum (Gibco) (56 �C, 30 min). Cells were maintained in a humid atmosphere at 37 �Cwith 5% CO2.

2.6. Cell proliferation assay

MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide] wasused to test cytotoxicity of polyisoprenoyl gemcitabine prodrug nanoassembliesand cell viability. Briefly, 3 � 103 of MiaPaCa-2, 2 � 103 of B16F10, 5 � 103 ofA549 and 7 � 103 of MCF7 cells per well were incubated in 200 lL medium contain-

ing 10% FBS in 96-well plates for 24 h. The cells were then exposed to series concen-trations of polyisoprenoyl gemcitabine prodrug 2a–2g NAs, polyisoprenic acidnanoassemblies 3a–3g or free gemcitabine for 72 h. After drug exposition, the med-ium was removed and 100 lL of MTT solution (0.5 mg/mL in DMEM containing 10%FBS) was added to each well. The plates were incubated for 2 h at 37 �C and 100 lLof 20% SDS solution was then added to each well for 24 h at 37 �C. Absorbance wasmeasured at 570 nm using a plate reader (Metertech R 960, Fisher Bioblock, Illkirch,France). The percentage of surviving cells was calculated as the absorbance ratio oftreated to untreated cells. The inhibitory concentration 50% (IC50) of the treatmentswas determined from the dose–response curve. All experiments were set up in qua-druplicate to determine means and SDs. No toxic effects were observed for all poly-isoprenic acid 3a–3g NAs.

Table 2In vitro cytotoxicity (IC50) of nanoassemblies constructed with different polyisopre-noyl prodrugs of gemcitabine against various human and murine cancer cell lines.

Compound IC50 (nM)

A549 MCF-7 B16F10 Mia PaCa-2

Gem 15 15 250 252a 450 550 470 7002b 220 1200 620 5002c 30 100 200 602d 15 15 100 252e 80 35 230 1202f 350 500 980 14002g 200 100 230 250

A. Maksimenko et al. / Cancer Letters 334 (2013) 346–353 349

2.7. Apoptosis assay by Annexin V staining

To assess apoptosis, cells were stained with an Annexin V-FITC apoptosis assaykit (Invitrogen, France) following the protocol provided by the manufacturer. Anearly indicator of apoptosis is the rapid translocation and accumulation of thephospholipid phosphatidylserine from the cytoplasmic interface of the cell mem-brane to the extracellular surface. This event can be detected by Annexin V staining.Thus, after 72 h of treatment with polyisoprenoyl gemcitabine prodrug 2a–2d NAsat IC50 concentration, MiaPaCa-2 cells were washed and resuspended in 100 lL 1�Annexin V binding buffer. Subsequently, 5 lL of FITC-conjugated Annexin V and1 lL of propidium iodide (PI) solution (100 lg/mL) were added to each cell suspen-sion. After 15 min incubation in the dark at room temperature, stained cells wereanalyzed by flow cytometry (Accuri 6; Accuri, Ann Arbor, MI) using a BD AccuriCFlow Plus software. All the samples were assayed in triplicate, and apoptotic frac-tion was calculated as follows: apoptotic cell number/total cell number �100%.

2.8. In vivo study designs

Six-eight-week old female athymic nude mice were purchased from Harlan Lab-oratory. All animals were housed in appropriate animal care facilities during theexperimental period, and were handled according to the principles of laboratoryanimal care and legislation in force in France.

The farnesylacetyl gemcitabine prodrug nanoassemblies (2d NAs) were chosenfor efficacy study on the animals, as being the most active compound in vitro. Thesystemic toxicity of this drug was firstly investigated on healthy mice, compara-tively to gemcitabine and to SQ-gem NAs, by determining the maximum tolerateddose (MTD), after repeated intravenous injection into female nude mice. Practically,all groups (n = 5) received four intravenous injections on days 0, 4, 8 and 12 in thelateral tail vein (10 lL/g of body weight). Four groups of nude mice received 2d NAsat a gemcitabine equivalent dose of 10, 15, 20 or 25 mg/kg, three groups of nudemice received SQ-gem NAs at a gemcitabine equivalent dose of 7, 10 or 15 mg/kgand three groups of nude mice received gemcitabine at a dose of 15, 20 or25 mg/kg. The control groups received saline and 50 mg/kg of 1,10 ,2-tris-nor-squa-lenic (3a NAs) or farnesylacetic (3d NAs) acid nanoassemblies. Any toxicity of squa-lenic (3a NAs) or farnesylacetic (3d NAs) acid nanoassemblies was observed at doselevels tested. The body weight change and physical states of mice were monitoredfor a period of 29 days. The MTD was estimated based on the threshold at which allanimals survived with a body weight loss less than 10% (Table 3).

The antitumor efficacy of 2d NAs was further evaluated at equimolar doses (i.e.MTD for 2a NAs), comparatively to free gemcitabine or to SQ-gem NAs, on MiaPaCa-2-bearing mice. Practically, 200 lL of the MiaPaCa-2 cell suspension, equivalent to1 � 107, were injected subcutaneously into nude mice toward the upper portion ofthe right flank, to develop a solid tumor model. Tumors were allowed to grow untila volume �90 mm3 before initiating the treatment. Tumor length and width weremeasured with calipers, and the tumor volume was calculated using the followingequation: Tumor volume (V) = length �width2/2. Tumor-bearing nude mice wererandomly divided into 6 groups of 8 each and all groups received four intravenousinjections on days 0, 4, 8 and 12 in the lateral tail vein with either (i) SQ-gem nano-assemblies (2a NAs) at a gemcitabine equivalent dose of 7 mg/kg (MTD), (ii) farn-esylacetyl gemcitabine nanoassemblies (2d NAs) at a gemcitabine equivalentdose of 7 mg/kg, (iii) gemcitabine 7 mg/kg, (iv) farnesylacetic acid nanoassemblies(3d NAs) 50 mg/kg, (v) 1,10 ,2-tris-nor-squalenic acid nanoassemblies (3a NAs)50 mg/kg, or (vi) saline 0.9%. The injected volume was 10 lL/g of body weight.The mice were monitored regularly for changes in tumor size and weight.

3. Results

3.1. Synthesis, morphological characterization and physicochemicalproperties of polyisoprenoyl gemcitabine nanoassemblies

Bioconjugates 2b–2g (Fig. 1) were obtained by reaction of thecorresponding polyisoprenic acids with 30,50-bis TBDMS-protectedgemcitabine, followed by conventional TBAF deprotection.

When submitted to nanoprecipitation conditions at 2 mg/mLinto water media (for details see Section 2), all bioconjugates werefound to be capable of spontaneous self-assembly, providing nano-structures with size comprised between 112 and 270 nm (Table 1).However, the drugs 2b, 2c and 2d precipitated soon after nanoas-semblies formation. To stabilize them, bioconjugate 2d bearing 3isoprene units, was nanoprecipitated into water, together with20% (w/w) of SQ-PEG and bioconjugates 2b and 2c, containingrespectively 1 and 2 isoprene units, were nanoprecipitated into a10% aqueous solution of trehalose, providing stable NAs at 37 �C.Noteworthy, conjugation with such short oligoprenyl units (i.e.

2b–2d) provided nanoassemblies with impressively high drugloading with respect to squalene 2a prodrug (Table 1).

Interestingly, it was observed that cooling below 30 �C the sus-pensions of NAs 2c and 2d just after their preparation resulted intheir aggregation in a gel like structure whereas, heating againthe gels until 37 �C allowed resuspension of the nanoassembliesat the same size. Thus, 2c and 2d NAs were stored at 37 �C whereasthe other NAs were stored at 4 �C. NAs of bioconjugates 2a–2d and2g (2 mg mL�1) were imaged by cryo-transmission electronmicroscopy (cryo-TEM), revealing a population of spherical nano-assemblies with diameter ranging from 100 to 300 nm (Fig. 2).

3.2. Drug release

Incubation of nanoassemblies made of gemcitabine prodrugs2a–2d, 2f and 2g in 90% fetal bovine serum (FBS) culture mediumhas evidenced a zero order release of free gemcitabine (Fig. 3).However, the release rate was significantly faster for 2b, 2c and2d comparatively to 2a, 2f and 2g nanoassemblies. After 24 h ofincubation, approximately 29, 14 and 11% of total drug was re-leased in the presence of serum for polyisoprenoyl gemcitabineprodrugs 2c, 2d and 2b respectively, as compared with only 3%for SQ-gem (2a NAs) or, 6% for 2f and 2g nanoassemblies (Fig. 3).

3.3. In vitro efficacy studies

The nanoassemblies made of gemcitabine prodrugs 2b–2g wereevaluated for their cytotoxic activity in vitro against three humancancer lines (lung cell line A549, hormone-dependent breast celllines MCF-7 and pancreatic carcinoma cell line Mia PaCa-2) andone murine melanoma cell line B16F16, using free gemcitabineand SQ-gem (2a) NAs as controls. The cell viability was checkedby MTT assay. Results showing the concentrations required to inhi-bit cell growth by 50% (IC50 values) are presented in Table 2. In allcases, owing to their prodrug nature, studied nanoassemblies 2a–2g showed lower cytotoxicity than gemcitabine free. Interestingly,farnesylacetyl gemcitabine (2d) NAs displayed the best anticanceractivity on all tested tumor cell lines, as compared to the controlSQ-gem (2a) NAs and to the other polyisoprenoyl gemcitabine pro-drug 2b, 2c, 2e, 2f and 2g NAs. Thus, the antitumor efficacy of thenanoassemblies constructed with this prodrug (2d) has been fur-ther investigated in vivo on human pancreatic (MiaPaCa-2) carci-noma xenograft model in mice.

To further confirm gemcitabine-induced apoptosis, MiaPaCa-2cells were treated with polyisoprenoyl gemcitabine prodrug nano-assemblies 2a, 2b, 2c and 2d at IC50 (i.e. 700, 500, 60 and 25 nM,respectively) and compared with free gemcitabine (25 nM). After72 h of treatment, Annexin V and propidium iodide (PI) stainingshowed comparable induction of apoptosis in polyisoprenoyl gem-citabine 2a–2d NAs treated cells and in gemcitabine-treated cells(Fig. 4).

Fig. 2. Cryo-TEM appearance of polyisoprenoyl gemcitabine nanoassemblies 2a–2d and 2g (scale bar, 100 nm).

Fig. 3. Release of gemcitabine from 2a, 2b, 2c, 2d, 2f and 2g nanoassemblies at 37 �C in solution containing 90% FBS. Released gemcitabine was separated from polyisoprenoylgemcitabine nanoassemblies 2a–2d, 2f and 2g by centrifugation and quantified using HPLC (Section 2).

350 A. Maksimenko et al. / Cancer Letters 334 (2013) 346–353

3.4. Overall toxicity

The systemic toxicity of farnesylacetyl gemcitabine prodrug(2d) NAs was investigated in vivo, comparatively to gemcitabineand SQ-gem NAs, by determining the maximum tolerated dose(MTD) after repeated i.v. injection into female nude mice. TheMTD was estimated based on the threshold at which all animalssurvived with a body weight loss less than 10%; nanoassembliesof 1,10,2-tris-norsqualenic (3a) and farnesylacetic (3d) acids wereused as negative control. Whatever the dosing protocol, 2d NAsappeared dramatically (three times) less toxic than SQ-gem NAs(Table 3) and had MTD comparable to gemcitabine free.

3.5. In vivo anticancer efficacy

The antitumor efficacy of the three-isoprene units conjugate 2dNAs has been investigated on the human pancreatic carcinomaxenograft model MiaPaCa-2 at equimolar doses (i.e. MTD for 2aNAs), comparatively to gemcitabine free and to SQ-gem NAs. Aftertumors had grown to 80–100 mm3, the animals were subjected tothe different treatments using injection protocols, as explained inSection 2. As indicated in Fig 5A, the growth of MiaPaCa-2 tumorswas not affected neither by the treatment with the control nanoas-semblies made of 1,10,2-tris-norsqualenic (3a NAs) or farnesylace-tic (3d NAs) acid nanoassemblies nor with gemcitabine free, when

Fig. 4. Apoptotic fraction following gemcitabine or 2a–2d nanoassemblies exposure increased in MiaPaCa-2 cells. Cancer cells were treated with 2a–2d NAs or gemcitabine atIC50 (IC50 was 700, 500, 60 and 25 nM for polyisoprenoyl prodrug 2a, 2b, 2c and 2d NAs, respectively, and 25 nM for gemcitabine). Enhancement of gemcitabine-inducedapoptosis by prodrugs 2a–2d was confirmed by Annexin V staining. Data represent the mean ± SD from triplicate independent experiments (P < 0.01 for all prodrugs andgemcitabine versus nontreated cells).

Table 3Maximum tolerated dose (MTD) of free gemcitabine, SQ-gem (2a) and farnesylacetyl gemcitabine prodrug (2d) NAs.

Drugs MTD (mg/kg)

Gemcitabine 20a [80b]Prodrug 2a NAs 7a [28b]Prodrug 2d NAs 20a [80b]3a NAs ND3d NAs ND

ND – nondefined.a Equivalent to gemcitabine.b Total dose of treatment after five administrations.

A. Maksimenko et al. / Cancer Letters 334 (2013) 346–353 351

compared with saline-treated tumors. On the contrary, the treat-ment with 2a and 2d NAs reduced the tumor volume by 41% and76% respectively at day 34, (P < 0.01) (Fig. 5a). Noteworthy, notany absolute weight loss in all treated groups was observed(Fig. 5b).

4. Discussion

In previous reports [4], we have taken advantage of the remark-able dynamically folded conformation of squalene to chemicallyconjugate this lipid with various therapeutic molecules in orderto construct nanoassemblies of 100–300 nm with very high drugpayload. This so-called ‘‘squalenoylation’’ concept has been appliedwith success to gemcitabine, among other anticancer compounds.

In the current study, using gemcitabine as a drug model, wehave investigated whether the ‘‘squalenoylation’’ approach couldalso work using other terpenes for drug conjugation and furthernanoassemblage. In other words, the aim was to explore the influ-ence of the isoprene length on the ability (or disability) of theresulting bioconjugates to form nanoassemblies and to emphasizethe impact of terpene’s nature on the in vitro and in vivo pharma-cological activity. Thus, a series of 7 polyisoprenoyl prodrugs ofgemcitabine were synthesized. These bioconjugates were designedto improve gemcitabine efficacy and to overcome the limitationsdue to the systemic toxicity of this anticancer compound. Eachconjugate was submitted to nanoprecipitation conditions in orderto assess the minimal requisite size of the polyisoprenyl side chainto allow the formation of nanoassemblies. Only bioconjugates 2a

and 2e–2g, containing from 4 to 6 isoprene units, were able to formstable NAs in pure distilled water. On the contrary, the bioconju-gates with a polyisoprenoyl moiety less than four monomeric units(i.e. 2b, 2c and 2d) did not produce stable nanoassemblies in purewater: 2b and 2c prodrugs, containing respectively one and twoisoprene units, allowed the formation of NAs only in a 10% treha-lose aqueous solution whereas the bioconjugate 2d (bearing threeunits of isoprene) was able to form NAs only in the presence of thesurface active SQ-PEG. Noteworthy, at the end of NAs preparation,the evaporation of the organic solvent was performed at 37 �C (seeSection 2 for more details) and it was observed that cooling theNAs suspension below 30 �C resulted in a precipitated gel structurefor 2c and 2d NAs, suggesting these nanoassemblies needed moreenergy for structure stabilization. Thus, bioconjugates with smallisoprenyl chains (less than four isoprene units) were less proneto form stable nanoassemblies than those bearing longer chains,probably due to the fact that the amphiphilic character of theseprodrugs was less pronounced. Indeed, theoretical calculation oflogP gave values between 5.37 and 8.43 for bioconjugates madewith isoprene units 4–6, whereas they were comprised between3.85 and 0.79 for bioconjugates made with less than four isopreneunits. This certainly deserves further physico-chemical investiga-tions. On the other hand, conjugate 2f gave nanoassemblies witha quite similar stability than SQ-gem (2a), revealing that the sym-metrical distribution of the double bonds found in the squalenechain is not required for the NAs formation and that a regular poly-isoprenyl chain is equally effective.

In vitro biological assessment showed that all polyisoprenoylgemcitabine nanoassemblies displayed notable cytotoxicity (IC50

below the micromolar range) on the tested cancer cell lines,including murine melanoma cell line B16F10, human pancreaticcarcinoma cell line MiaPaCa-2, human lung carcinoma cell lineA549 and human breast adenocarcinoma cell line MCF7. Interest-ingly, it was observed that the anticancer efficacy of these NAsdepended on the length of the polyisoprenoyl moiety. The polyiso-prenoyl prodrug of gemcitabine containing three isoprene units(2d) was more active, displaying a cytotoxicity comparable to gem-citabine free (and even better in case of the B16F10 melanoma cellline). In particular, 2d NAs were found to display a dramaticallyimproved cytotoxicity comparatively to SQ-gem NAs (dependingon the cell line, IC50 of 2d NAs was five-times to 36-times lowerthan that of 2a NAs). This may be explained in the light of the drug

Fig. 5. Tumor growth inhibition after farnesylacetyl- (2d) and squalenoyl- (2a) gemcitabine nanoassemblies treatment of mice bearing human pancreatic (MiaPaCa-2) tumor.Tumor volume (a) and body weight (b) were regularly measured during the experimental period. The mice were treated with prodrug 2a NAs (4 � 7 mg/kg gemcitabineequivalent, MTD) (s) and prodrug 2d NAs (4 � 7 mg/kg gemcitabine equivalent) (d), gemcitabine (4 � 7 mg/kg) (.), farnesylacetic acid NAs (3d NAs, 4 � 50 mg/kg) (4),1,10 ,2-tris-nor-squalenic acid NAs (3a NAs, 4 � 50 mg/kg) (j) or, saline (NaCl 0.9%) (h), (P < 0.1; n = 8).

352 A. Maksimenko et al. / Cancer Letters 334 (2013) 346–353

release experiments, showing that the release of gemcitabine wasfaster for farnesylacetyl- (2d) and geranylacetyl-gemcitabine (2c)nanoassemblies comparatively to SQ-gem (2a) NAs (Fig. 3). Thequite rapid liberation inside cells of the parent active moleculefrom nanoassemblies 2c and 2d led to impressively greater anti-cancer activity of NAs in various cancer cell lines by cell apoptosisinduction. The only exception to this trend is nanoassemblies 2c,which released gemcitabine faster than 2d although being lesscytotoxic. However, other parameters like the size could also influ-ence cytotoxicity. Indeed, it has been observed that the volume ofencapsulated drug is correlated to the third power of nanoparticlesradius [24]. Therefore, although the extent of uptake of nanoas-semblies 2d is probably lower than that of smaller nanoassemblies[25,26], their higher entrapment efficiency led to higher intracellu-lar drug content, responsible of the enhanced cytotoxicity com-pared to the other nanoassemblies.

The antitumor efficacy of the most active 2d NAs has then beeninvestigated on human pancreatic (MiaPaCa-2) carcinoma xeno-graft model in mice, comparatively to the treatment using theSQ-gem NAs (2a) or gemcitabine as a free drug. Noteworthy, alltreatments were performed at the equimolar doses (7 mg of gem-

citabine equivalent/kg; MTD for 2a NAs). The nanoassemblies con-structed using gemcitabine bioconjugates bearing three isopreneunits (i.e. 2d) showed important antitumor efficacy whereas theparent drug (i.e. gemcitabine free) did not display any antitumoractivity on human MiaPaCa-2 pancreatic cancer. 2d NAs were alsosignificantly more active than the SQ-gem (2a) NAs (the tumornodule was decreased, on day 34, by 76% with 2d versus only41% with 2a). Noteworthy, 2d NAs revealed lower toxicity thanSQ-gem as seeing in body weight loss, which was expected sinceSQ-gem NAs were administered at MTD.

Now, the question is if the PEGylation of 2d nanoassembliesmay be responsible for the observed anticancer activity. Althoughit is possible, this hypothesis appears unlikely. Indeed, we havepreviously investigated in details the interaction of 2a nanoassem-blies with cancer cells and it was clearly demonstrated that thenanoparticles released 2a as single prodrug molecules, extracellu-larly [27]. Then, the gemcitabine–squalene conjugates interactedwith extracellular proteins, allowing the facilitated insertion ofthe prodrug into the cell membrane before intracellular diffusionand further gemcitabine release occur [28]. If considering that 2dnanoassemblies could follow similar pathway, 2d will be released

A. Maksimenko et al. / Cancer Letters 334 (2013) 346–353 353

extracellularly into the blood stream after administration, so thatSQ-PEG will not be able to induce ‘‘stealth’’ properties to the parti-cles. Therefore, our opinion is that SQ-PEG has only a transient sta-bilization effect for the preparation of nanoparticles.

In conclusion, this study demonstrated that more efficient ter-pene based anticancer nanomedicines may be designed by adjust-ing the length of the polyisoprenyl chain, allowing moving fromthe ‘‘squalenoylation’’ concept to the ‘‘terpenoylation’’ concept.

Role of the funding source

The research leading to these results has received funding fromthe European Research Council under the European Community’sSeventh Framework Programme FP7/2007–2013 Grant AgreementNo. 249835.

Acknowledgments

We thank the service of the animal experimentation from theIFR141 (Châtenay-Malabry) and Ghislaine Frébourg (service ofelectron microscopy from IFR of Integrative Biology, Paris) for thecryo-TEM analyse.

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.canlet.2012.08.023.

References

[1] E. Niederlaender, Causes of Death in the EU, EuroStat, 2006.[2] P. Couvreur, C. Vauthier, Nanotechnology: intelligent design to treat complex

disease, Pharm. Res. 23 (2006) 1417–1451.[3] L. Wang, W. Zhao, W. Tan, Bioconjugated silica nanoparticles: development

and applications, Nano Res. 1 (2008) 99–115.[4] P. Couvreur, B. Stella, H.L. Reddy, H. Hillaireau, C. Dubernet, D. Desmaële, S.

Lepêtre-Mouelhi, F. Rocco, N. Dereuddre-Bosquet, P. Clayette, V. Rosilio, V.Marsaud, J.-M. Renoir, L. Cattel, Squalenoyl nanomedicines as potentialtherapeutics, Nano Lett. 6 (2006) 2544–2548.

[5] Y. Matsumura, H. Maeda, A new concept for macromolecular therapeutics incancer chemotherapy: mechanism of tumoritropic accumulation of proteinsand the antitumor agent smancs, Cancer Res. 46 (1986) 6387–6392.

[6] R.A. Freitas, The future of nanofabrication and molecular scale devices innanomedicine, Stud. Health Technol. Inform. 80 (2002) 45–59.

[7] G.A. Hughes, Nanostructure-mediated drug delivery, Nanomed. Nanotechnol.Biol. Med. (2005) 22–30.

[8] F.J. Xu, K.G. Neoh, E.T. Kang, Bioactive surfaces and biomaterials via atomtransfer radical polymerization, Prog. Polym. Sci. 34 (2009) 719–761.

[9] Y. Zhou, Bio-Inspired Nanomaterials and Nanotechnology, Nova SciencePublishers, New York, 2010.

[10] I. Linkov, F.K. Satterstrom, L.M. Corey, Nanotoxicology and nanomedicine:making hard decisions, Nanomed. Nanotechnol. Biol. Med. 4 (2008) 167–171.

[11] L. Zuo, W. Wei, M. Morris, J. Wei, M. Gorbounov, C. Wei, New technology andclinical applications of nanomedicine, Med. Clin. North Am. 91 (2007) 845–862.

[12] I. Linkov, F.K. Satterstrom, G.A. Kiker, T.S. Bridges, S.L. Benjamin, D.A. Belluck,From optimization to adaptation: shifting paradigms in environmentalmanagement and their application to remedial decision, Int. Environ. Assess.Manage. 2 (2006) 92–98.

[13] P. Couvreur, L.H. Reddy, S. Mangenot, J.H. Poupaert, D. Desmaële, S. Lepêtre-Mouelhi, B. Pili, C. Bourgaux, H. Amenitsch, M. Ollivon, Discovery of newhexagonal supramolecular nanostructures formed by squalenoylation of ananticancer nucleoside analogue, Small 4 (2008) 247–253.

[14] J.L. Abbruzzese, R. Grunewald, E.A. Weeks, D. Gravel, T. Adams, B. Nowak, S.Mineishi, P. Tarassoff, W. Satterlee, M.N. Raber, Activity of gemcitabine in ahuman tumor cloning assay as a basis for clinical trials with gemcitabine, J.Clin. Oncol. 9 (1991) 491–498.

[15] R. Grunewald, H. Kantarjian, M. Du, K. Faucher, P. Tarassoff, W.J. Plunkett,Gemcitabine in leukemia: a phase I clinical, plasma, and cellular pharmacologystudy, J. Clin. Oncol. 10 (1992) 406–413.

[16] L.H. Reddy, C. Dubernet, S.L. Mouelhi, P.E. Marque, D. Desmaele, P. Couvreur, Anew nanomedicine of gemcitabine displays enhanced anticancer activity insensitive and resistant leukemia types, J. Control. Release 124 (2007) 20–27.

[17] S. Réjiba, L.H. Reddy, C. Bigand, C. Parmentier, P. Couvreur, A. Hajri, Squalenoylgemcitabine nanomedicine overcomes the low efficacy of gemcitabine therapyin pancreatic cancer, Nanomedicine 7 (2011) 841–849.

[18] N. Sémiramoth, C.D. Meo, F. Zouhiri, F. Saïd-Hassane, S. Valetti, R. Gorges, V.Nicolas, J.H. Poupaert, S. Chollet-Martin, D. Desmaële, R. Gref, P. Couvreur, Self-assembled squalenoylated penicillin bioconjugates: an original approach forthe treatment of intracellular infections, ACS Nano 6 (2012) 3820–3831.

[19] M. Raouane, D. Desmaele, M. Gilbert-Sirieix, C. Gueutin, F. Zouhiri, C.Bourgaux, E. Lepeltier, R. Gref, R. Ben Salah, G. Clayman, L. Massaad-Massade, P. Couvreur, Synthesis, characterization, and in vivo delivery ofsiRNA-squalene nanoparticles targeting fusion oncogene in papillary thyroidcarcinoma, J. Med. Chem. 54 (2011) 4067–4076.

[20] F. Dosio, L.H. Reddy, A. Ferrero, B. Stella, L. Cattel, P. Couvreur, Novelnanoassemblies composed of squalenoyl-paclitaxel derivatives: synthesis,characterization, and biological evaluation, Bioconjugate Chem. 21 (2010)1349–1361.

[21] F. Bekkara-Aounallah, R. Gref, M. Othman, L.H. Reddy, B. Pili, V. Allain, C.Bourgaux, H. Hillaireau, S. Lepêtre-Mouelhi, D. Desmaele, J. Nicolas, N. Chafi, P.Couvreur, Novel PEGylated nanoassemblies made of self-assembledsqualenoyl nucleoside analogues, Adv. Funct. Mater. 18 (2008) 1–11.

[22] H. Khoury, A. Deroussent, L.H. Reddy, P. Couvreur, G. Vassal, A. Paci,Simultaneous determination of gemcitabine and gemcitabine–squalene byliquid chromatography-tandem mass spectrometry in human plasma, J.Chromatogr. 858 (2007) 71–78.

[23] N.A. Parshina, T.V. Pleteneva, V.N. Baikova, M.N. Narimanov, S.A. Tyulyandin,Quantitative estimation of gemcitabine by HPLC in plasma, Pharm. Chem. J. 42(2008) 37–39.

[24] S.M. Fujikawa, I.A. Chen, J.W. Szostak, Shrink-wrap vesicles, Langmuir 21(2005) 12124–12129.

[25] S. Zhang, J. Li, G. Lykotrafitis, G. Bao, S. Suresh, Size-dependent endocytosis ofnanoparticles, Adv. Mater. 21 (2009) 419–424.

[26] H. Hillaireau, P. Couvreur, Nanocarriers’ entry into the cell: relevance to drugdelivery, Cell. Mol. Life Sci. 66 (2009) 2873–2896.

[27] L. Bildstein, V. Marsaud, H. Chacun, S. Lepêtre-Mouelhi, D. Desmaële, P.Couvreur, C. Dubernet, Extracellular-protein-enhanced cellular uptake ofsqualenoyl gemcitabine from nanoassemblies, Soft Matter 6 (2010) 5570–5580.

[28] L. Bildstein, C. Dubernet, V. Marsaud, H. Chacun, V. Nicolas, C. Gueutin, A.Sarasin, H. Bénech, S. Lepêtre-Mouelhi, D. Desmaële, P. Couvreur,Transmembrane diffusion of gemcitabine by a nanoparticulate squalenoylprodrug: an original drug delivery pathway, J. Control. Release 147 (2010)163–170.