Professional Documents
Culture Documents
Tumor-Penetrating Peptide Functionalization Enhances The Anti-Glioblastoma Effect of Doxorubicin Liposomes
Tumor-Penetrating Peptide Functionalization Enhances The Anti-Glioblastoma Effect of Doxorubicin Liposomes
doxorubicin liposomes
This content has been downloaded from IOPscience. Please scroll down to see the full text.
(http://iopscience.iop.org/0957-4484/24/40/405101)
View the table of contents for this issue, or go to the journal homepage for more
Download details:
IP Address: 129.173.72.87
This content was downloaded on 02/07/2014 at 07:19
Tumor-penetrating peptide
functionalization enhances the
anti-glioblastoma effect of
doxorubicin liposomes
Yiyi Yang1,2,5 , Zhiqiang Yan2,5 , Daixu Wei2 , Jian Zhong2 , Lu Liu2 ,
Lin Zhang3 , Fei Wang4 , Xiaoli Wei4 , Cao Xie4 , Weiyue Lu4 and
Dannong He1,2
1
School of Materials Science and Engineering, Shanghai Jiao Tong University, Shanghai 200240,
People’s Republic of China
2
National Engineering Research Center for Nanotechnology, Shanghai 200241,
People’s Republic of China
3
Department of Pharmacy, Shaoxing People’s Hospital, Shaoxing Hospital of ZheJiang University,
Shaoxing 312000, People’s Republic of China
4
Key Laboratory of Smart Drug Delivery, Ministry of Education and PLA, Department of
Pharmaceutics, School of Pharmacy, Fudan University, Shanghai 201203, People’s Republic of China
0957-4484/13/405101+08$33.00 1 c 2013 IOP Publishing Ltd Printed in the UK & the USA
Nanotechnology 24 (2013) 405101 Y Yang et al
been damaged by the invasion of tumor cells, and to glioblastoma in vivo. Then the growth inhibitory
endothelial gaps have formed on the microvessels of brain effect of RGERPPR-functionalized doxorubicin liposomes
tumors [5]. Then the enhanced permeability and retention (RGE-LS/DOX) to glioblastoma was investigated. Finally,
(EPR) effect comes into action, leading to the tendency of we studied the tumor-penetrating ability of RGE-LS by
nanoscale particles (for example, nanoparticles, liposomes immunofluorescence analysis.
and macromolecular drugs) to accumulate in tumor tissue
much more than they do in normal tissues [6]. However, 2. Materials and methods
the EPR effect of intracranial tumors is relatively weak
compared with that of peripheral tumors, as the critical pore 2.1. Materials
size for solute passage of intracranial tumors is smaller than
peripheral tumors [7, 6]. For example, the critical pore size Mal-PEG-DSPE was obtained from Laysan Bio Co. (USA).
for intracranial human primary glioblastoma cell line U87MG Fetal bovine serum (FBS) and DMEM medium were from
is as low as 7–100 nm [7]. The weak EPR effect decreases Gibco Co. (USA). PEG2000-DSPE and HSPC were supplied
the extravasation of nano drug delivery system (NDDS) from by Shanghai AVT Pharmaceutical (China). DiR was from
blood vessel to the glioblastoma tissue, thereby resulting in Invitrogen (USA). Cholesterol was supplied by Sinopharm
the low delivering efficiency to the targets [6]. Chemical Reagent Co. (China). Sephadex G50 was from GE
Another problem affecting the efficient delivery of NDDS Healthcare (UK). Rat anti-mouse CD31 was from Abcam
in glioblastoma is the difficulty of permeation from the tumor (UK). Goat anti-rat IgG-R was from SantaCruz (USA). DAPI
vessel vicinity deep into tumor tissue. This results from was purchased from Molecular Probes (USA). All chemicals
several delivering barriers, including the rapidly increased were analytic reagent grade. U87MG cell line was provided
IFP in the periphery of tumor tissue, dense stromal cells by Shanghai Institute of Cell Biology, which was cultured
and ECM between tumor vessels and cancer cells [8]. These according to Zhan et al [20]. Male Balb/c Nu/Nu mice
barriers prevent the transport of NDDS sufficiently deep to (four weeks old) of 20–25 g body weight were supplied
reach as many glioblastoma cells as possible. Several reports by Shanghai SLAC laboratory animal company (China) and
have proved that the NDDS can penetrate only a few cell maintained under SPF conditions. All manipulations were
diameters into the extravascular tumor tissue, leaving most performed in accordance with procedures approved by the
cancer cells in deep tumor tissue unable to be exposed to ethics committee (Shanghai Jiao Tong University).
antitumor drugs [9–11]. This problem has resulted in the
incomplete inhibition of tumor growth, arousing extensive 2.2. Synthesis of RGERPPR-PEG-DSPE
concern among researchers [12, 11].
Neuropilin-1 (NRP-1) is a receptor overexpressed on Firstly, the thiolated RGERPPR (C-RGERPPR) with was
both glioblastoma cells and tumor endothelium [13]. As synthesized via Boc-protected solid-phase peptide synthesis
the co-receptor of VEGF165, NRP-1 plays an essential strategy, followed by reaction with Mal-PEG-DSPE as
role in tumor angiogenesis, metastasis and regulation of described previously [21]. The reaction solution was purified
vascular permeability [13, 14]. The expressing level of NRP-1 by dialysis (MWCO 3.5 kDa cut off, Millipore) against
increases with the increase in the degree of glioblastoma distilled water, followed by lyophilization. The purity was
malignancy [15]. RGERPPR peptide, derived from phage verified by 1 H-NMR and FTIR.
display screens, is the specific ligand of NRP-1 [16, 17]. It
was proved to be a ‘tumor-penetrating peptide’ (TPP), able 2.3. Preparation of liposomes
to penetrate through tumor vessels and tumor stroma [18, 16].
Several reports demonstrated that TPP functionalization could The thin-film hydration and extrusion method was used
promote the transport of NDDS through blood vessel to to prepare the liposomes loaded with DOX, FAM or DiR
tumor tissue and the penetration through tumor stroma as previously [22, 23]. HSPC, cholesterol, mPEG-DSPE
and deep into the whole tumor tissue, which had been and RGERPPR-PEG-DSPE were used with the molar ratio
evidenced in pancreatic, breast, prostatic and liver cancer of 55:40:5:1 (for RGERPAR-modified liposomes, including
[18, 16, 19]. Therefore, TPP makes it possible to solve the RGE-LS/FAM, RGE-LS/DiR and RGE-LS/DOX) or 0 (for
two above-mentioned problems of NDDS in the targeting unmodified liposomes, including LS/FAM, LS/DiR and
therapy of glioblastoma. On the one hand, the tumor LS/DOX). DOX was loaded into liposomes by the ammonium
vessel-penetrating ability of TPP-functionalized NDDS may sulfate gradient method according to the procedure described
compensate for the weak EPR effect of glioblastoma; on the previously [24].
other, the tumor stroma-penetrating ability may compensate
for the low delivering efficiency of NDDS in glioblastoma. 2.4. Characterization of liposomes
In this study, we prepared RGERPPR-functionalized
liposomes (RGE-LS) to enable them to penetrate through The vesicle size of liposomes was measured by dynamic
tumor vessel and tumor stroma, thereby enhancing the light scattering method (Zetasizer Nano-ZS, Malvern Instru-
inhibitory effect to glioblastoma. Firstly, we conjugated ments, Westborough, MA). Transmission electron microscopy
RGERPPR to the surface of liposomes and tested their (TEM) (JEM-2010FEF, JEOL, Japan) was used to evaluate the
targetability to U87MG glioblastoma cells in vitro and morphology of liposomes.
2
Nanotechnology 24 (2013) 405101 Y Yang et al
The MTT assay was used to determine the cytotoxicity Statistical differences were evaluated with two-tailed cor-
of RGE-LS/DOX according to the procedure reported rected student’s t-test with p < 0.05 as significant.
previously [31]. U87MG cells were cultured in 96 well plate
for 24 h (2000/well), and treated by serial concentrations of 3. Results
LS/DOX or RGE-LS/DOX for 72 h. The MTT assay was
performed and the percentage of cell viability was determined 3.1. Characterization of RGERPPR-PEG-DSPE
by using an enzyme-labeling instrument (PowerWave XS,
Bio-TEK, USA). The IC 50 values were determined by curve Figure 1 shows the molecular structure, NMR and FTIR
analysis software (GraphPad Prism 5.02). spectra of Mal-PEG-DSPE and RGERPPR-PEG-DSPE. Both
3
Nanotechnology 24 (2013) 405101 Y Yang et al
Figure 2. LS, RGE-LS, LS/DOX and RGE-LS/DOX all exhibited uniform spherical morphology as displayed in the TEM images (scale
bar = 100 nm).
Table 1. The vesicle sizes and drug leakage rates of DOX, FAM or DiR loaded liposomes with or without RGERPAR modification.
Formulation Vesicle size (nm) Polydispersity index Drug leakage (%)
LS/DOX 88.5 ± 8.0 0.022 ± 0.004 —
RGE-LS/DOX 87.7 ± 9.1 0.026 ± 0.007 —
LS/FAM 93.1 ± 6.7 0.029 ± 0.005 1.89 ± 0.32
RGE-LS/FAM 92.3 ± 5.4 0.030 ± 0.010 1.97 ± 0.55
LS/DiR 86.6 ± 7.8 0.024 ± 0.007 N.D.
RGE-LS/DiR 88.1 ± 6.9 0.025 ± 0.008 N.D.
Data are represented with mean ± SD (n = 3); N.D. = not detectable.
NMR spectra contain the solvent peak of CDCl3 at 7.26 ppm, no obvious influence on the vesicle size and morphology
the methylene protons peaks of DSPE at 1.26 ppm and the of liposomes. The leakage of FAM from liposomes was
repeat units of PEG at 3.7–3.8 ppm. The NMR spectrum about 2% within 4 h and no leakage of DiR was
of Mal-PEG-DSPE indicates the presence of the maleimide detected from liposomes within 72 h (table 1), suggesting
group by its characteristic peak at 6.7 ppm, which disappears that the prepared fluorescein-loaded liposomes satisfied the
in that of RGERPPR-PEG-DSPE. The results suggested the requirement of the cellular uptake, in vivo targeting ability and
successful reaction of the thiol group of RGERPPR with immunofluorescence study [32].
maleimide group.
In the FTIR spectrum of Mal-PEG-DSPE, a weak C=O 3.3. Targeting property of RGE-LS
stretch band at 1666.8 cm−1 and a N–H stretch band at
3200–3600 cm−1 are found, which may be responsible for 3.3.1. In vitro cellular uptake of RGE-LS. The images of
the amide groups in Mal-PEG-DSPE. The intensity of the two cellular uptake and flow cytometry for FAM, LS/FAM and
bands in the spectrum of RGERPPR-PEG-DSPE remarkably RGE-LS/FAM were shown in figure 3. The percentages of
enhance, which should be due to the increased number of FAM-positive cells for FAM, LS/FAM and RGE-LS/FAM
were 2.99%, 14.8% and 94.1%, and the mean fluorescent
amide groups compared with Mal-PEG-DSPE. The results
intensities for them were 18.3, 18.7 and 361, respectively.
suggested that the RGERPPR peptide had been conjugated to
These data indicated that the cellular uptake of liposomes was
Mal-PEG-DSPE.
remarkably enhanced by RGERPPR modification.
4
Nanotechnology 24 (2013) 405101 Y Yang et al
Figure 3. The representative images of cellular uptake and flow cytometry for FAM (A), LS/FAM (B) and RGE-LS/FAM (C). The cellular
uptake of liposomes by U87MG cells was remarkably enhanced by RGERPPR modification.
Figure 4. Representative in vivo fluorescent images of glioblastoma-bearing nude mice following i.v. administration of LS/DiR (A) and
RGE-LS/DiR (B). RGERPPR functionalization increased the liposome delivery to glioblastoma in vivo.
3.4. The tumor-penetrating ability of RGE-LS the IC50 values were 1.445 µM for LS/DOX and 0.468 µM
for RGE-LS/DOX (figure 6), indicating that RGE-LS/DOX
The tumor-penetrating ability of RGE-LS in glioblastoma was exhibited significantly higher growth inhibitory effect to
investigated by performing the immunofluorescence analysis U87MG cells than LS/DOX.
of frozen glioblastoma sections. The immunofluorescence
images showed that LS/FAM were mainly located adjacent
to blood vessels, whereas RGE-LS/FAM were located in 3.5.2. In vivo inhibitory effect of RGE-LS/DOX to
the whole tumor tissue (figure 5). The results suggested glioblastoma. The anti-glioblastoma effect of RGE-
that RGERPPR modification enabled liposomes to penetrate LS/DOX was evaluated by the Kaplan–Meier survival curve
through tumor vessels and tumor stroma and deep into the of intracranial U87MG glioblastoma-bearing nude mice. The
whole tumor tissue. results (figure 7) showed that the median survival times of
mice injected with normal saline (NS), DOX, LS/DOX and
3.5. Growth inhibitory effect of RGE-LS/DOX to RGE-LS/DOX were 38, 39, 42 and 48 days, respectively. The
glioblastoma survival times of LS/DOX group (p < 0.05, log-rank analysis)
and RGE-LS/DOX group (p < 0.001) were significantly
3.5.1. Cytotoxicity of RGE-LS/DOX. The in vitro longer than that of NS, but there was no significant difference
cytotoxicity of RGE-LS/DOX was measured by MTT assay, between free DOX group and the NS group (p > 0.05).
compared with that of LS/DOX. After incubation for 72 h, Such superiority of LS/DOX and RGE-LS/DOX over DOX
5
Nanotechnology 24 (2013) 405101 Y Yang et al
Figure 5. Representative immunofluorescence images of frozen glioblastoma of nude mice following injection of LS/FAM and
RGE-LS/FAM. LS/FAM were mainly located adjacent to blood vessels (shown for CD31, red), whereas RGE-LS/FAM penetrated through
tumor vessels and tumor stroma and deep into the whole tumor tissue.
6
Nanotechnology 24 (2013) 405101 Y Yang et al
the tumor-penetrating peptide-modified liposomes penetrated [2] Ong B Y, Ranganath S H, Lee L Y, Lu F, Lee H S,
through tumor vessels and deep into the glioblastoma Sahinidis N V and Wang C H 2009 Paclitaxel delivery from
tissues, enabling most tumor cells to be exposed to and be PLGA foams for controlled release in post-surgical
chemotherapy against glioblastoma multiforme
effectively inhibited by drug-loaded liposomes. Furthermore, Biomaterials 30 3189–96
it is possible to decrease the drug resistance and recurrence [3] Johnson D R and Chang S M 2012 Recent medical
of the tumor by using the tumor-penetrating peptide-modified management of glioblastoma Adv. Exp. Med. Biol.
NDDS. 746 26–40
The NRP-1 is overexpressed on not only tumor cells [4] de Lange E C 2012 The physiological characteristics and
but also endothelial cells in tumor vessels. Accordingly, transcytosis mechanisms of the blood–brain barrier (BBB)
Curr. Pharm. Biotechnol. 13 2319–27
the RGE-LS should possess targetability to tumor vessels [5] Jain R K, di Tomaso E, Duda D G, Loeffler J S, Sorensen A G
in addition to tumor cells, which has two advantages. On and Batchelor T T 2007 Angiogenesis in brain tumours
the one hand, the targeting of tumor vessels can increase Nature Rev. Neurosci. 8 610–22
the distribution of NDDS into tumor tissues. Generally, the [6] Liu Y and Lu W 2012 Recent advances in brain
presence of tumor cell-targeted ligands on the nanocarrier tumor-targeted nano-drug delivery systems Expert Opin.
Drug Discov. 9 671–86
surface does not change the chances of nanocarriers to
[7] Hobbs S K, Monsky W L, Yuan F, Roberts W G, Griffith L,
enter tumor tissues, which is mainly determined by the EPR Torchilin V P and Jain R K 1998 Regulation of transport
effect [37]. In contrast, RGE peptide, a tumor vessel-targeted pathways in tumor vessels: role of tumor type and
ligand, can take more nanocarriers to the tumor vessels by microenvironment Proc. Natl Acad. Sci. USA 95 4607–12
its specific binding with NRP-1 on tumor vessels, thereby [8] Hanahan D and Coussens L M 2012 Accessories to the crime:
increasing the chances of nanocarriers to enter tumor tissues. functions of cells recruited to the tumor microenvironment
Cancer Cell 21 309–22
In addition, with the tumor-penetrating ability, RGE-LS
[9] Lu W, Xiong C, Zhang R, Shi L, Huang M, Zhang G, Song S,
adjacent to the tumor vessels can penetrate through tumor Huang Q, Liu G Y and Li C 2012 Receptor-mediated
vessels and deep into the whole tumor tissue. On the other transcytosis: a mechanism for active extravascular transport
hand, RGE-LS may be able to destroy the tumor vessels while of nanoparticles in solid tumors J. Control. Release
killing the tumor cells, which would significantly amplify the 161 959–66
antitumor effect of NDDS. Many reports have proved that [10] Hambley T W and Hait W N 2009 Is anticancer drug
development heading in the right direction? Cancer Res.
the elimination of endothelial cells can remarkably inhibit 69 1259–62
the growth of tumor cells [38]. Besides, the destruction of [11] Ruoslahti E, Bhatia S N and Sailor M J 2010 Targeting of
tumor vessels can improve the growth inhibitory effect on drugs and nanoparticles to tumors J. Cell Biol. 188 759–68
both primary tumors and metastatic solid tumors [38]. [12] Zhou Y and Kopecek J 2013 Biological rationale for the
In summary, RGE-LS exhibited an enhanced targeted design of polymeric anti-cancer nanomedicines J. Drug
therapeutic effect on glioblastoma, based on their tumor Target 21 1–26
[13] Wild J R, Staton C A, Chapple K and Corfe B M 2012
targetability and penetrating ability, indicating that tumor- Neuropilins: expression and roles in the epithelium Int. J.
penetrating peptide functionalization is an effective strategy Exp. Pathol. 93 81–103
of enhancing the anti-glioblastoma effect of doxorubicin [14] Staton C A, Kumar I, Reed M W and Brown N J 2007
liposomes. This study may provide a solution for the current Neuropilins in physiological and pathological angiogenesis
dilemma of active targeted therapy of glioblastoma and other J. Pathol. 212 237–48
[15] Osada H, Tokunaga T, Nishi M, Hatanaka H, Abe Y, Tsugu A,
tumors.
Kijima H, Yamazaki H, Ueyama Y and Nakamura M 2004
Overexpression of the neuropilin 1 (NRP1) gene correlated
Acknowledgments with poor prognosis in human glioma Anticancer Res.
24 547–52
[16] Teesalu T, Sugahara K N, Kotamraju V R and Ruoslahti E
This work was supported by the National Basic Research
2009 C-end rule peptides mediate neuropilin-1-dependent
Program of China (2013CB932500), National Natural Sci- cell, vascular, and tissue penetration Proc. Natl Acad. Sci.
ence Foundation of China (81202471, 51203024), Interna- USA 106 16157–62
tional Cooperation Project from Science and Technology [17] Haspel N, Zanuy D, Nussinov R, Teesalu T, Ruoslahti E and
Commission of Shanghai Municipality (No. 12520708000), Aleman C 2011 Binding of a C-end rule peptide to the
Nano Project from Science and Technology Commission neuropilin-1 receptor: a molecular modeling approach
Biochemistry 50 1755–62
of Shanghai Municipality (11nm0505000), Shaoxing City [18] Sugahara K N, Teesalu T, Karmali P P, Kotamraju V R,
Public Welfare Technology Applied Research Projects Agemy L, Girard O M, Hanahan D, Mattrey R F and
(2012B70050) and Open Project Program of Key Lab Ruoslahti E 2009 Tissue-penetrating delivery of compounds
of Smart Drug Delivery (Fudan University), Ministry of and nanoparticles into tumors Cancer Cell 16 510–20
Education & PLA, China (SDD2011-06). [19] Wang X, Zhen X, Wang J, Zhang J, Wu W and Jiang X 2013
Doxorubicin delivery to 3D multicellular spheroids and
tumors based on boronic acid-rich chitosan nanoparticles
References Biomaterials 34 4667–79
[20] Zhan C, Wei X, Qian J, Feng L, Zhu J and Lu W 2012
[1] Polivka J Jr, Polivka J, Rohan V, Topolcan O and Ferda J 2012 Co-delivery of TRAIL gene enhances the anti-glioblastoma
New molecularly targeted therapies for glioblastoma effect of paclitaxel in vitro and in vivo J. Control. Release
multiforme Anticancer Res. 32 2935–46 160 630–6
7
Nanotechnology 24 (2013) 405101 Y Yang et al
[21] Mulder W J M, Strijkers G J, Griffioen A W, van Bloois L, [29] Franklin K B J and Paxinos G 2007 The Mouse Brain in
Molema G, Storm G, Koning G A and Nicolay K 2004 Stereotaxic Coordinates 3rd edn (New York: Academic)
A liposomal system for contrast-enhanced magnetic [30] Zhan C, Gu B, Xie C, Li J, Liu Y and Lu W 2010 Cyclic RGD
resonance imaging of molecular targets Bioconjug. Chem. conjugated poly(ethylene glycol)-co-poly(lactic acid)
15 799–806 micelle enhances paclitaxel anti-glioblastoma effect
[22] Yan Z, Wang F, Wen Z, Zhan C, Feng L, Liu Y, Wei X, Xie C J. Control. Release 143 136–42
and Lu W 2012 LyP-1-conjugated PEGylated liposomes: a [31] Mosmann T 1983 Rapid colorimetric assay for cellular growth
carrier system for targeted therapy of lymphatic metastatic and survival: application to proliferation and cytotoxicity
tumor J. Control. Release 157 118–25 assays J. Immunol. Methods 65 55–63
[23] Yan Z, Zhan C, Wen Z, Feng L, Wang F, Liu Y, Yang X, [32] Wen Z, Yan Z, Hu K, Pang Z, Cheng X, Guo L, Zhang Q,
Dong Q, Liu M and Lu W 2011 LyP-1-conjugated Jiang X, Fang L and Lai R 2011 Odorranalectin-conjugated
doxorubicin-loaded liposomes suppress lymphatic nanoparticles: preparation, brain delivery and
metastasis by inhibiting lymph node metastases and pharmacodynamic study on Parkinson’s disease following
destroying tumor lymphatics Nanotechnology 22 415103 intranasal administration J. Control. Release 151 131–8
[24] Haran G, Cohen R, Bar L K and Barenholz Y 1993 [33] Zhan C, Qian J, Feng L, Zhong G, Zhu J and Lu W 2011
Transmembrane ammonium sulfate gradients in liposomes Cyclic RGD–poly(ethylene glycol)–polyethyleneimine is
produce efficient and stable entrapment of amphipathic more suitable for glioblastoma targeting gene transfer in
weak bases Biochim. Biophys. Acta 1151 201–15
vivo J. Drug Target 19 573–81
[25] Shen J, Zhan C, Xie C, Meng Q, Gu B, Li C, Zhang Y and
[34] Corti A and Curnis F 2011 Tumor vasculature targeting
Lu W 2011 Poly(ethylene glycol)-block-poly(D,L-lactide
through NGR peptide-based drug delivery systems Curr.
acid) micelles anchored with angiopep-2 for brain-targeting
delivery J. Drug Target 19 197–203 Pharm. Biotechnol. 12 1128–34
[26] Yu Z, Schmaltz R M, Bozeman T C, Paul R, Rishel M J, [35] Xin H, Jiang X, Gu J, Sha X, Chen L, Law K, Chen Y,
Tsosie K S and Hecht S M 2013 Selective tumor cell Wang X, Jiang Y and Fang X 2011 Angiopep-conjugated
targeting by the disaccharide moiety of bleomycin J. Am. poly(ethylene glycol)-co-poly(epsilon-caprolactone)
Chem. Soc. 135 2883–6 nanoparticles as dual-targeting drug delivery system for
[27] Li C, Wang Y, Zhang X, Deng L, Zhang Y and Chen Z 2013 brain glioma Biomaterials 32 4293–305
Tumor-targeted liposomal drug delivery mediated by a [36] Marcucci F and Corti A 2012 Improving drug penetration to
diseleno bond-stabilized cyclic peptide Int. J. Nanomed. curb tumor drug resistance Drug Discov. Today 17 1139–46
8 1051–62 [37] Kwon I K, Lee S C, Han B and Park K 2012 Analysis on the
[28] Bleau A M, Hambardzumyan D, Ozawa T, Fomchenko E I, current status of targeted drug delivery to tumors J. Control.
Huse J T, Brennan C W and Holland E C 2009 Release 164 108–14
PTEN/PI3K/Akt pathway regulates the side population [38] Wu H C and Chang D K 2010 Peptide-mediated liposomal
phenotype and ABCG2 activity in glioma tumor stem-like drug delivery system targeting tumor blood vessels in
cells Cell Stem Cell 4 226–35 anticancer therapy J. Oncol. 2010 723798