4550

Published OnlineFirst on May 18, 2010 as 10.1158/0008-5472.CAN-10-0339

Therapeutics, Targets, and Chemical Biology

Cancer

Research

Targeted Nanoparticles that Deliver a Sustained, SpecificRelease of Paclitaxel to Irradiated Tumors

Ralph J. Passarella1, Daniel E. Spratt1, Alice E. van der Ende2, John G. Phillips1, Hongmei Wu1,Vasanth Sathiyakumar2, Li Zhou1, Dennis E. Hallahan3,4, Eva Harth2, and Roberto Diaz5,6

Abstract

Authors' AUniversityUniversity,and Mallincof Medicineof Radiatio6Winship C

Note: SupResearch O

CorresponOncology,Clifton Ro404-778-17ment of CNashville,E-mail: eva

doi: 10.115

©2010 Am

Cancer R

To capitalize on the response of tumor cells to XRT, we developed a controlled-release nanoparticle drugdelivery system using a targeting peptide that recognizes a radiation-induced cell surface receptor. Phagedisplay biopanning identified Gly-Ile-Arg-Leu-Arg-Gly (GIRLRG) as a peptide that selectively recognizestumors responding to XRT. Membrane protein extracts of irradiated glioma cells identified glucose-regulatedprotein GRP78 as the receptor target for GIRLRG. Antibodies to GRP78 blocked the binding of GIRLRG in vitroand in vivo. Conjugation of GIRLRG to a sustained-release nanoparticle drug delivery system yielded increasedpaclitaxel concentration and apoptosis in irradiated breast carcinomas for up to 3 weeks. Compared withcontrols, a single administration of the GIRLRG-targeted nanoparticle drug delivery system to irradiatedtumors delayed the in vivo tumor tripling time by 55 days (P = 0.0001) in MDA-MB-231 and 12 days inGL261 (P < 0.005). This targeting agent combines a novel recombinant peptide with a paclitaxel-encapsulatingnanoparticle that specifically targets irradiated tumors, increasing apoptosis and tumor growth delay in amanner superior to known chemotherapy approaches. Cancer Res; 70(11); 4550–9. ©2010 AACR.

Introduction

In vitro, many agents are capable of killing cancer cellseffectively. These agents trigger cancer cell death throughnumerous complex pathways, such as apoptosis or preventionof further cell division. However, when these agents are trans-ferred from use in cell culture to an entire system, the effect onnormal tissue limits their use in a clinical setting. With a smalltherapeutic index between cancer destruction and toxic sideeffects, drugs are often not used in patients or discontinuedfar before they achieve a maximal effect.Thus, targeted therapy provides a means to circumvent

the toxicities and lack of treatment response of conventionalsystemic chemotherapy. With the development of targetedbiologicals, such as trastuzumab and imatinib, therapies forspecific cancer types have been developed. However, these

ffiliations: 1Department of Radiation Oncology, VanderbiltMedical Center; 2Department of Chemistry, VanderbiltNashville, Tennessee; 3Department of Radiation Oncologykrodt Institute of Radiology, Washington University School; 4Siteman Cancer Center, St. Louis, Missouri; 5Departmentn Oncology, Emory University School of Medicine; andancer Institute of Emory University, Atlanta, Georgia

plementary data for this article are available at Cancernline (http://cancerres.aacrjournals.org/).

ding Authors: Roberto Diaz, Department of RadiationEmory University School of Medicine, Suite A1300, 1365ad, NE, Atlanta, GA 30322. Phone: 404-778-3669; Fax:50; E-mail: roberto.diaz@emory.edu or Eva Harth, Depart-hemistry, Vanderbilt University, 7619 Stevenson Center,TN 37235. Phone: 615-343-3405; Fax: 615-343-1234;.harth@vanderbilt.edu.

8/0008-5472.CAN-10-0339

erican Association for Cancer Research.

es; 70(11) June 1, 2010

therapeutics are often limited to cancers expressing variousmutations and thus are limited in broad use (1, 2). Treat-ments aimed at universal solid tumor therapy, such as angio-genesis inhibitors, have had limited success thus far (3).The discovery of receptors expressed at much greater

levels on tumors than on normal tissue would provide targetsfor drug delivery. Therefore, receptor induction in tumorscould play a critical role in providing new targets. Ionizingradiation (XRT), although also therapeutic, could potentiallycause cellular stress localized to cancer cells, which maycause new receptor translocation. Beyond its cytotoxiceffects, XRT has been shown to induce gene transcription(4) and protein expression on tumor microvasculature (5).Using phage display biopanning, recombinant peptides thatbind only to treated cancers have been found (6, 7). However,combining these peptides with chemotherapeutic agents hasnot been effectively translated to clinical use.Through targeted therapeutics, nanoparticle delivery sys-

tems have the potential to overcome the normal tissuetoxicity of traditional chemotherapy. However, although pac-litaxel encapsulation in albumin nanoparticles, nab-paclitaxel,increases the efficacy and safety over paclitaxel formulated inCremophor (8–10), more nausea, diarrhea, and grade 3 sensoryneuropathy occurs in patients treated with nab-paclitaxel (8).Another remaining challenge of nanoparticle delivery systemsis the lack of control of drug release profiles. This distinct“burst effect,” in which the majority of drug is released in arapid and uncontrollable fashion, creates unpredictable phar-macokinetics, thereby making effective dose regimens difficultto predict.Our goal is to identify a novel recombinant peptide and a

radiation-inducible receptor pair in XRT-treated cancers and

Radiation-Induced GRP78 Mediates Targeted Drug Delivery

conjugate the peptide to a novel nanoparticle drug deliverysystem (DDS) for use in XRT targeted chemotherapy. In thiswork, we present three new findings: the discovery of arecombinant peptide that recognizes XRT-treated tumors,the discovery of the radiation-inducible receptor of this pep-tide, and the therapeutic benefits of a targeted recombinantpeptide/nanoparticle DDS.

Materials and Methods

Animals usedAthymic nude and C57/Bl6 mice were purchased from

Harlan Laboratories. All animal protocols were approved bythe Institutional Animal Care and Use Committee.

Tumor modelsGL261 murine glioma and MDA-MB-231 human breast

cancer cell lines were purchased from American TypeCulture Collection. Heterotopic tumor models were devel-oped by s.c. inoculating cell suspensions (6 × 106 cells) intonude or C57/Bl6 mice.

Coculture assaysHuman umbilical vein endothelial cells (HUVEC) in the

sixth passage (Lonza) and GL261 murine glioma cells werecocultured as previously described (7). The cells wereallowed to interact for 1 day before treatment with 3 GyXRT and incubated for 3 hours before they were harvested.Coverslips were blocked for 30 minutes with 5% bovineserum albumin and 1% streptavidin (ThermoScientific).Cells were incubated for 1 hour with a streptavidin-peptide-AlexaFluor594 complex (AlexaFluor594 carboxylic acid succi-nimidyl ester was purchased from Invitrogen; SupplementaryFig. S1). HUVEC nuclei were stained with 4′,6-diamidino-2-phenylindole, and images of nuclei and peptide binding weretaken by Vanderbilt Cell Imaging Shared Resource Center us-ing a Zeiss Axiophot fluorescent microscope at 40× magnifi-cation. Cell colocalization was done using Metamorph Offlinesoftware in all assays.Similarly, 3 × 105 HUVECs were layered in coculture plates

alone and treated with 3 Gy or left untreated, incubatedwith streptavidin-peptide-AlexaFluor594 complex, and im-aged as before. Positive and negative controls of XRT-treatedand untreated GL261/HUVEC cocultures with the peptideincubated on HUVECs were used. Assays were done threetimes in triplicate.

Near IR imagingTumor-bearing mice were treated with three once-daily

doses of 3 Gy XRT or sham XRT (three per group) andinjected with peptide or antibody 3 hours after the lastXRT treatment. In one experiment, labeled complexes ofbiotinylated peptide-AlexaFluor750 conjugates were in-jected (AlexaFluor750 carboxylic acid succinimidyl esterwas purchased from Invitrogen). In a second experiment,an antibody to 78-kDa glucose regulated protein (GRP78)conjugated with AlexaFluor750 was injected and tumorswere removed 7 days after labeled antibody injection; poly-

www.aacrjournals.org

clonal serum IgG antibody was used as a control. Ina third experiment, mice received an unlabeled blockingantibody to GRP78 or unlabeled polyclonal IgG serumand then injected with the labeled Gly-Ile-Arg-Leu-Arg-Gly (GIRLRG) peptide. Near IR images were taken usingthe IVIS imaging system with an ICG filter setting at var-ious time points after the injection.

Membrane protein extractionGL261 tumor samples either treated with 3 Gy XRT or

sham XRT were removed from the hind limbs of athymicnude mice 48 hours after treatment and frozen at −80°C.Forty-milligram samples of treated and untreated frozentumors were homogenized, and the protein was extractedusing the Mem-PER Eukaryotic Membrane Protein Extrac-tion Kit (ThermoScientific). The extracted protein was thenincubated overnight in the Slide-A-Lyzer Dialysis Cassettes(ThermoScientific). The protein was then incubated overnightwith NeutrAvidin-coated agarose beads (ThermoScientific)bound to biotinylated GIRLRG or scrambled peptide (RILGGR).After incubation, the beads were washed with 1× PBS, boiledat 100°C, and run on an Invitrogen NuPAGE 10% gel. The gelwas stained with Invitrogen Simply Blue SafeStain. Bandsfrom the gel were analyzed by the Vanderbilt ProteomicsCore through the liquid chromatography/tandem mass spec-trometry technique.

Western blottingNude mice implanted s.c. in the hind limb with either

GL261 and MDA-MB-231 tumors were treated with 3 GyXRT for 3 consecutive days, and tumors were removed48 hours post-XRT. Tumor samples were homogenized, andthe protein from the samples was extracted. Protein wasprobed with antibodies for GRP78 and actin (Cell Signaling)on a polyvinylidene difluoride membrane and exposed to filmthat was later developed using the Western Lightning Chemi-luminescence Plus detection system (Perkin-Elmer) accord-ing to the manufacturer's protocol.

ImmunohistochemistryParaffin-embedded tumor samples were stained using an

antibody for the von Willebrand factor (vWF; DakoCytoma-tion) at a 1:100 dilution from the original stock solution of3.1 g/L and incubated overnight. Samples were then incu-bated with a streptavidin-peptide-AlexaFluor594 complexand washed three times with PBS.In a second assay, samples were stained with vWF and in-

cubated with an antibody to GRP78 at dilutions of 1:250 and1:1,000 of original stock solution. These samples were thenincubated with streptavidin-peptide-AlexaFluor594 complexand imaged.Images were taken using a fluorescent microscope at 20×

magnification. Assays were performed in triplicate.

Nanoparticle synthesis and attachment of GIRLRGpeptide to nanoparticlesPolyester nanoparticle DDS was synthesized by the pro-

cedure described by van der Ende and colleagues (11).

Cancer Res; 70(11) June 1, 2010 4551

Passarella et al.

4552

KKCGGGGIRLRG peptide (56 mg, 3.35 μmol) in dimethylsulf-oxide (DMSO; 2 mL) was added to a solution of nanoparticlesfrom poly(valerolactoneepoxyvalerolactone-allylvalerolac-tone-oxepanedione) containing 11% epoxide and cross-linkedwith 1 equivalent of 2,2-(ethylenedioxy)bis(ethylamine) perepoxide (ref. 11; 105.6 mg, 0.78 μmol) in DMSO (1 mL). Thereaction mixture was heated for 72 hours at 34°C. Residualpeptide was removed by dialyzing with SnakeSkin Pleated Di-alysis Tubing (molecular weight cutoff 10,000) against 50/50THF/CH3CN.

Encapsulation of paclitaxel in GIRLRG-conjugatednanoparticlesPaclitaxel was encapsulated by the procedure described by

van der Ende and colleagues (12). The weight percent of pac-litaxel encapsulated in the nanoparticles was determined by

Cancer Res; 70(11) June 1, 2010

NanoDrop UV-Vis at 254 nm as mentioned in the literatureand was found to be 11.3%.

Paclitaxel antibody stainingNude mice were implanted in the hind limb with MDA-MB-

231 tumors. Once tumors reached 450 mm3 in volume, micewere treated with 3 Gy XRT once daily for 3 consecutive daysor were left untreated. On the 2nd day, mice were injected withone of either (a) systemic paclitaxel, (b) paclitaxel/nanoparti-cle with RILGGR, or (c) paclitaxel/nanoparticle with GIRLRG.The paclitaxel concentration used was 10mg/kg. Tumors wereremoved 1 and 3 weeks after treatment, embedded in paraffin,and sectioned. Tumor sections were incubated with a mono-clonal antibody to paclitaxel (Santa Cruz Biotechnology) at aconcentration of 1:500. Three mice per group were used. Allpaclitaxel antibody staining was performed in triplicate.

Figure 1. GIRLRG binds to tumors responding to XRT. A, biopanning of the T7 phage–based peptide library followed by sequencing was used in vivoto identify phage peptides binding specifically to glioma (GL261) treated with 3 Gy XRT and 40 mg/kg sunitinib. Shown here are the three aminoacid sequences isolated most frequently and the percentage of peptide-bearing phages in the sequenced clones. B, GIRLRG preferentially bindsto radiation-treated cocultures of GL261 gliomas and HUVECs. GIRLRG (top row) or a scrambled peptide (middle row) was incubated on HUVECs fromXRT-treated and untreated GL261/HUVEC cocultures. GIRLRG was also incubated on XRT-treated and untreated HUVEC cultures alone (bottom row).DAPI, 4′,6-diamidino-2-phenylindole. C and D, GIRLRG binds to tumors responding to XRT ex vivo and in vivo. C, nude mice were implanted in the righthind limb with GL261 glioma cells and were either left untreated or treated daily with radiation (3 Gy) for 5 consecutive days (three mice per treatmentgroup). Tumor sections were incubated with vWF, an endothelial marker, followed by fluorescently labeled GIRLRG, and imaged. Untreated tumor sectionsand XRT-treated tumor sections 48 hours posttreatment are shown. D, nude mice were implanted in the hind limb with GL261 glioma cells and wereeither left untreated or treated daily with radiation (3 Gy) for 3 consecutive days (three mice per treatment group). Fluorescently labeled GIRLRG peptidepreferentially binds to radiation-treated tumors in GL261 xenografts compared with untreated tumors.

Cancer Research

Radiation-Induced GRP78 Mediates Targeted Drug Delivery

Terminal deoxynucleotidyl transferase–mediated dUTPnick end labeling stainingNude mice were treated and tumors were collected as de-

scribed in the above section. Terminal deoxynucleotidyltransferase–mediated dUTP nick end labeling (TUNEL)staining was done with the DeadEnd Colorimetric TUNELSystem (Promega) following the manufacturer's instructions.Positive staining was observed by light microscopy.

Paclitaxel antibody staining and TUNEL stainingevaluationAll slides were evaluated and graded based on color inten-

sity of immunoreactions using a six-tier grading system of5 to 6 (strong), 3 to 4 (moderate), 1 to 2 (faint), and 0 (none).Assays were performed in triplicate.

Statistical analysesStudent's t test was used to perform group comparisons.

Linear correlations of peptide binding and tumor responseto treatment were developed by using the correlation coeffi-cient of tumor growth and radiance data sets (SigmaPlot).

Results

Discovery of a peptide that recognizes irradiated tumorsWe used phage display technology to identify a targeting

peptide that would discriminately bind to irradiated tumors.

www.aacrjournals.org

Using a previously characterized in vivo biopanning methodto screen the T7 phage–based random peptide library (6, 7),the novel peptide GIRLRG was identified as the predominantphage-encoded peptide recovered from irradiated GL261 gli-omas in mice (Fig. 1A). We then investigated the specificityof GIRLRG for irradiated tumors using in vitro cocultureexperiments. To model the tumor microenvironment, weused HUVECs cocultured with GL261 tumor cells. Theseexperiments revealed that the GIRLRG targeting peptidebound to tumor vasculature only when two criteria weremet: tumor cells were irradiated and tumor cells were ableto interact with HUVECs (Fig. 1B). To simulate normal tissue,HUVECs were cultured alone. There was no binding of theGIRLRG recombinant peptide to this normal tissue model,suggesting an obligate interaction between the tumor and tu-mor vasculature for the target receptor of the peptide to beavailable for binding. We also found that the GIRLRG peptidecolocalizes with an endothelial cell marker in irradiatedtumor samples ex vivo (Fig. 1C). We then used an in vivobinding model using near IR imaging with GL261 tumorsimplanted in the hind limbs of mice. The GIRLRG targetingpeptide was again shown to preferentially bind to radiation-treated tumors in vivo over untreated tumors (Fig. 1D).

GRP78 is induced by XRT in tumorsWe next sought to discover the target of GIRLRG. Agarose

beads coated with GIRLRG were incubated with membrane

Figure 2. GRP78 is induced by XRT. A, the highlighted sequences are the sequences identified by mass spectrometry as extracted and precipitatedwith GIRLRG peptide coated on agarose beads. B, GRP78 is induced in HUVECs grown in coculture with GL261 gliomas after XRT treatment. GL261 gliomacells (3 × 105) and HUVECs (1 × 104) were cocultured for 24 hours before treatment with 3 Gy XRT. Three hours posttreatment, AlexaFluor594-labeledGRP78 antibodies were added to the culture plates. C, Western blot of GRP78 expression in XRT-treated and untreated (No Tx) GL261 glioma andMDA-MB-231 breast carcinoma tumor sections showing that GRP78 is upregulated to the membrane in response to XRT. D, antibody to GRP78 bindsselectively to XRT-treated tumors. GL261 tumors were implanted in the hind limbs of nude mice and treated with 3 Gy XRT for 3 consecutive days.On the 3rd day, an antibody to GRP78 or IgG antibody serum conjugated with AlexaFluor750 was injected through the tail veins of mice. Tumors wereremoved 7 days after peptide injection and imaged using near IR imaging to assess relative levels of peptide binding.

Cancer Res; 70(11) June 1, 2010 4553

Passarella et al.

4554

protein extracts from irradiated GL261 tumors and removed48 hours postirradiation. We isolated a 78-kDa band by gelelectrophoresis. Mass spectrometry of that band revealed itto be GRP78 (Fig. 2A). Therefore, we investigated the effectsof radiation on GRP78 concentration. In vitro, we found thatGRP78 is induced in HUVECs grown in coculture with GL261gliomas after XRT treatment (Fig. 2B). Membrane protein ex-tract from GL261 gliomas and MDA-MB-231 breast carcino-mas were analyzed for GRP78 expression through Westernblot (Fig. 2C). The results revealed increased GRP78 expressionafter radiation treatment, in both tumor types, compared withcontrols. GRP78 upregulation in response to XRT in GL261 gli-omas was validated ex vivo (Supplementary Fig. S2) and in vivo(Fig. 2D). Tumors implanted in the hind limbs of nude micewere treated with XRT, then injected with either a fluores-cently labeled GRP78 antibody or labeled control polyclonalserum IgG, and imaged using near IR imaging. Tumors treatedwith XRT showed intense binding of fluorescently labeled an-tibody to GRP78 compared with IgG serum controls (Fig. 2D).

GRP78 and GIRLRG interaction studiesWe next studied the putative ligand-receptor interaction

between the GIRLRG peptide and GRP78. This was accom-plished through in vitro, ex vivo, and in vivo experiments

Cancer Res; 70(11) June 1, 2010

using blocking antibodies to GRP78. In vitro cocultures withGL261/HUVECs showed decreased binding of GIRLRG toHUVECs when blocking antibodies to GRP78 were addedto the coculture (Fig. 3A). Next, we treated implantedGL261 tumors in mice with XRT, removed and sectionedthe tumors, and treated with varying concentrations ofGRP78 blocking antibody. We found that as the GRP78 block-ing antibody concentration increased, binding of fluores-cently labeled GIRLRG decreased (Fig. 3B). An in vivoimaging study was performed to assess if fluorescentlylabeled GIRLRG peptide could still bind to XRT-treatedtumors following the addition of an antibody to GRP78.The GRP78 antibody attenuated GIRLRG signal intensity by>70% compared with control IgG serum antibody in irra-diated GL261 tumors (P < 0.05; Fig. 3C and D).

Creation of a GIRLRG-targeted nanoparticle DDSWe postulated that conjugating the GIRLRG recombinant

peptide with a nanoparticle DDS could target chemotherapeu-tics specifically to radiated tumors. After nanoparticle forma-tion (11), the peptide was conjugated using a high-yieldingthiol-ene reaction, reacting the free thiol of the cysteine nearthe NH2 terminus of the KKCGGGGIRLRG with allyl function-alities on the nanoparticle (12). Nuclear magnetic resonance

Figure 3. GRP78 is the target of GIRLRG. A, GL261 glioma cells and HUVECs were cocultured for 24 hours before treatment with 3 Gy XRT followed byblocking antibody to GRP78 or a control IgG antibody serum. Cells were then incubated with AlexaFluor594-labeled GIRLRG peptide and imaged forpeptide binding. B, GL261 gliomas implanted in the hind limbs of nude mice were treated with 3 Gy XRT daily for 3 days, sectioned and incubatedwith a polyclonal antibody to GRP78 at antibody concentrations of 1:1,000 and 1:250. C and D, GL261 gliomas implanted in the hind limbs of nude mice(three mice per treatment group) were treated with 3 Gy XRT daily for 3 days. One hour after the final XRT treatment, mice were injected with either apolyclonal antibody to GRP78 or IgG antibody serum. Three hours after antibody administration, mice were injected with GIRLRG peptide labeled withAlexaFluor750. C, mice were imaged 24 hours after the final XRT treatment using near IR imaging. D, graph of relative peptide binding in vivo in micetreated with XRT and injected with either an antibody to GRP78 or an IgG control antibody serum. Differential peptide binding is shown as the percentage oftumor fluorescence increase relative to an untreated tumor-bearing mouse injected with an IgG control antibody serum.

Cancer Research

Radiation-Induced GRP78 Mediates Targeted Drug Delivery

spectroscopy methods could determine the conjugation of 37peptides (Fig. 4). In the final step, paclitaxel was incorporated,resulting in a DDS that is well dispersed in a Cremophor-freesolution. The biocompatibility of peptide-targeted particles inconcentrations applied for in vivo studies was confirmed incytotoxicity assays (Supplementary Fig. S3).

GIRLRG-targeted nanoparticle DDS increasespaclitaxel concentration and apoptosis inirradiated tumorsWe next investigated the effects of the GIRLRG-targeted

nanoparticle DDS on paclitaxel concentration and apoptosisin tumors compared with controls. MDA-MB-231 breastcarcinomas were implanted in the hind limbs of nude miceand treated as described in Fig. 5. Tumors were harvestedat 1 and 3 weeks posttreatment, and the levels of paclitaxel(Fig. 5A and B) and apoptosis (Fig. 5C and D) were deter-mined with the respective cell staining assays and quantified.Paclitaxel was found in significantly greater concentrationsin the targeted nanoparticle group with the use of irradiationover all other treatment groups at 1 and 3 weeks (P < 0.05;Fig. 5A and B). Similarly, TUNEL staining of these tumor sec-tions showed that at 1 and 3 weeks, the nanoparticle-GIRLRG

www.aacrjournals.org

DDS was superior to radiation and systemic paclitaxel in main-taining persistent cytotoxicity (P < 0.05; Fig. 5C and D). In fact,staining for paclitaxel and apoptosis significantly persisted for3 weeks after just a single administration of the nanoparticleover the other control groups (P < 0.05), indicating that thenanoparticle-GIRLRG peptide complex provides a prolongedand sustained release of paclitaxel when properly targeted tothe tumor with XRT.

Treatment with targeted nanoparticle DDS producesin vivo tumor growth delayOur primary outcome to determine the overall efficacy of

our novel targeting nanoparticle DDS was to assess tumorvolume tripling time in human tumor cell lines and in synge-neic mouse tumors. Therefore, we implanted MDA-MB-231breast carcinomas in nude mice and GL261 gliomas in C57/B16 mice and performed a tumor growth delay study aftertreating the mice as shown in Fig. 6. Our results showed thatMDA-MB-231 tumor tripling time was delayed 55 days withthe nanoparticle-targeted peptide with XRT (P = 0.0001), com-pared with 11 to 14 days by the three other XRT treatmentgroups (P < 0.05; Fig. 6A). Both unirradiated nanoparticlegroups provided no significant tumor growth delay when

Figure 4. Synthesis of a GIRLRG-targeted nanoparticle DDS. A, polyester nanoparticle formation resulting from an intermolecular cross-linking of diamine[2,2′-(ethylenedioxy)bis(ethylamine); red] and linear polymer precursor [poly(vl-avl-evl-opd); blue] containing epoxide units as cross-linking partners toform well-defined spherical nanoparticles (31). B, the nanoparticle-GIRLRG conjugate was synthesized through thiol-ene chemistry through the free allylmoieties on the nanoparticle (11) and the free cysteine at the NH2 terminus of the targeting GIRLRG peptide (orange). Paclitaxel (yellow spheres) wasincorporated into the cavities of the three-dimensional nano-network of the particles by a vitamin E D-α-tocopherol polyethylene glycol 1000 succinateformulation technique (12) to lead to the nanoparticle DDS. C, illustration of an opened fully synthesized nanoparticle DDS, showing the linear polymers(blue), cross-linking (red), paclitaxel (yellow spheres), and GIRLRG targeting peptide (orange).

Cancer Res; 70(11) June 1, 2010 4555

Passarella et al.

4556

compared with the untreated control, suggesting that evennanoparticle-GIRLRG is not adequately targeted in unirradi-ated tumors. The administration of radiation with systemicpaclitaxel or with untargeted nanoparticle (nanoparticle-RILGGR) provided no significant tumor growth delay whencompared with radiation alone (Fig. 6A). Similarly, in theGL261 group, tumor tripling time was significantly delayedby 12 days by nanoparticle-targeted peptide with XRT treat-ment (P < 0.005); however, all other treatment groups failed tosignificantly delay tumor tripling time compared with un-treated controls (Fig. 6B).

Discussion

We began designing our targeted DDS by seeking peptidescapable of recognizing irradiated cancer cells. Using a previ-

Cancer Res; 70(11) June 1, 2010

ously characterized in vivo biopanning method (6, 7, 13, 14),we discovered several candidate peptides (Fig. 1A). One ofthe candidates, GIRLRG, proved to be specific to irradiatedcancer cells capable of interacting with tumor vascular endo-thelial cells in vitro (Fig. 1B) and tumors ex vivo and in vivo(Fig. 1C and D). For the GIRLRG peptide to be of clinical util-ity for targeted therapy, its receptor must not be a ubiqui-tously expressed surface protein. XRT provides an intensecellular stress, causing activation of DNA damage repair cas-cades and endoplasmic reticulum stress pathways (15).Therefore, we hypothesized that the receptor for GIRLRGcould potentially be involved in one of these stress pathwaysand be induced by XRT.We sought to find this receptor using GL261 tumor

membrane protein affinity purification with GIRLRG. Thereceptor identified for GIRLRG is GRP78 (Fig. 2A). GRP78 is

Figure 5. Nanoparticle DDS increases paclitaxel concentration and apoptosis in irradiated tumors. A and B, analysis of paclitaxel levels shows thatGIRLRG-targeted nanoparticle DDS increases paclitaxel concentration in irradiated breast cancers. MDA-MB-231 breast carcinomas were implantedin the hind limbs of nude mice and treated with 3 Gy XRT daily for 3 days, and systemic paclitaxel, targeted nanoparticle (NP) DDS, or random peptideDDS on the 2nd day once they reached 450 mm3 (three mice per treatment group). C and D, TUNEL staining was used to assess relative levels ofapoptosis between treatment groups in tumor sections removed at 1 and 3 weeks posttreatment. Otherwise, mice were treated in the same fashion as in A.*, P < 0.05 when compared with the XRT + paclitaxel/nanoparticle with GIRLRG group (B and D).

Cancer Research

Radiation-Induced GRP78 Mediates Targeted Drug Delivery

known as an endoplasmic reticulum chaperone involvedin suppression of stress-induced apoptosis (16) but can existas a cell surface protein to transduce extracellular stimulito intracellular signals to promote tumorigenesis (17–19).Signaling through cell surface GRP78 increases cytosoliccalcium concentration, Akt phosphorylation, IP3, andNF-κB, leading to an increase in DNA and protein synthesisas well as cellular proliferation (19). A breadth of researchsupports the correlation of GRP78 to higher pathologicgrade, metastasis, chemotherapeutic response, cancer prog-nosis, and patient survival in gliomas and breast carcinomas(16, 18, 20–23). Importantly for clinical translation, GRP78is expressed at much higher levels in a variety of tumorsand tumor vasculature compared with much lower levelsin normal tissues and non–tumor-bearing vasculaturewhere expression potentially increases during tissueinflammation (17, 18, 20, 21, 23–25). This “natural” gradient

www.aacrjournals.org

of GRP78 expression between tumor vasculature and non–tumor-bearing vasculature is consistent with our in vitrodata of GIRLRG selectively binding only to irradiated HU-VECs incubated with cancer cells, not to HUVECs alone(Fig. 1B).The discovery that GRP78 is upregulated at the cell surface

in XRT-treated tumors and tumor vasculature (Fig. 2) maybe a further indicator of its role in the cellular stress re-sponse, and in the ability of a cancer cell to escape stressorsthat would lead a normal cell to apoptosis (16, 20–23). Themechanism by which GRP78 translocates to the cell surfaceis not fully understood, but hypotheses include particularmechanisms adapted by cancer cells, oversaturation of theendoplasmic reticulum retention system, transmembrane cy-cling of endoplasmic reticulum GRP78 to the cell surface, andcotrafficking with cell surface client proteins (21). BecauseGRP78 is expressed at the cell surface of tumors but not

Figure 6. GIRLRG-targeted nanoparticleDDS causes tumor growth delay in vivo.MDA-MB-231 (A) or GL261 (B) tumors wereimplanted in the hind limbs of nude mice orC57/Bl6, respectively. Once tumors reached300 mm3 in volume, mice were treated with3 Gy XRT daily for 3 days, or were left asuntreated controls. On the 2nd day, micewere injected with either systemic paclitaxel,nanoparticle-RILGGR scrambled peptide, ornanoparticle-GIRLRG targeted peptide at aconcentration of 10 mg/kg (five mice pertreatment group) (A) or 20 mg/kg (three mice pertreatment group) (B). Tumor volumes weremonitored throughout using calipers.

Cancer Res; 70(11) June 1, 2010 4557

Passarella et al.

4558

normal organs, cell surface GRP78 has become an attractivestrategy for targeted therapy (21). Ligand peptides for GRP78are rapidly internalized through clathrin-mediated endocyto-sis (26). Previously, GRP78 targeting peptides linked withpaclitaxel (27, 28), doxorubicin (28), or proapoptotic peptides(26) have been shown to induce melanoma cell death in vitro.Once weekly systemic administration over 4 weeks of pro-apoptotic chimeric peptides fused to GRP78 binding motifssuppressed tumor growth in xenograft models without af-fecting normal organs (17). Nonetheless, these observationshave not been approved for clinical use.Having discovered a receptor induced by XRT, we sought

to use nanoparticle technology to deliver a paclitaxel drugpayload using GIRLRG as a targeting molecule for GRP78.We postulated that conjugating the GIRLRG recombinantpeptide with a nanoparticle DDS capable of controlled phar-macokinetics could target chemotherapeutics specifically toradiated tumors. In addition, the control over particle sizeshas been recognized to be crucial to predict the interactionwith cells and other biological barriers (29) and reduce therisk of undesired clearance from the body through the liveror spleen (30). Therefore, the nanoparticle we used appliesan intermolecular cross-linking technique (31) that not onlyallows for predetermined nanoparticle dimensions with SDsof 10% but also provides adjustable cross-linking densities tocontrol the degradation of the particles and allows for post-modification reactions with bioactive groups such as thetargeting peptide. The adjustable cross-linking densities ofthe nanoparticle-targeting peptide complex can be appliedtoward the controlled and sustained release of paclitaxel.Consistent with the nanoparticle biodegradation profile(12, 31), paclitaxel was found in significantly greater concen-trations in our MDA-MB-231 breast cancer xenografted micein the targeted nanoparticle group with the use of XRT overall other treatment groups at 1 and 3 weeks (P < 0.05; Fig. 5Aand B). Similarly, TUNEL staining for apoptosis at both1 and 3 weeks were greatly increased, indicating that thenanoparticle-GIRLRG DDS was superior to radiation andsystemic paclitaxel in maintaining persistent cytotoxicity(P < 0.05; Fig. 5C and D). Our data support the model thatthe GIRLRG peptide is able to achieve significant targetingof paclitaxel to the tumor (Fig. 5A and B) when there is highexpression of GRP78 at the surface, which can be induced intumors with XRT (Fig. 2).Our primary outcome to determine the overall efficacy of

our novel targeting nanoparticle DDS was to assess tumorvolume tripling time in both human tumor cell lines andin syngeneic mouse tumors. Our results showed thatMDA-MB-231 tumor tripling time was delayed 55 days withthe nanoparticle-targeted peptide with XRT (P = 0.0001),compared with 11 to 14 days by the three other XRTtreatment groups (P < 0.05; Fig. 6A). Both unirradiated nano-particle groups provided no significant tumor growthdelay when compared with the untreated control, suggestingthat even the nanoparticle-GIRLRG complex itself is notadequately targeted in unirradiated tumors. The adminis-tration of radiation with systemic paclitaxel or with untar-geted nanoparticle (nanoparticle-RILGGR) provided no

Cancer Res; 70(11) June 1, 2010

significant tumor growth delay when compared with radia-tion alone (Fig. 6A), suggesting that the nanoparticle itself(nanoparticle-RILGGR) does not target irradiated tumors.Similarly, in the GL261 group, tumor tripling time was signif-icantly delayed by 12 days by nanoparticle-targeted peptidewith XRT treatment (P < 0.005); however, all other treatmentgroups failed to significantly delay tumor tripling timecompared with untreated controls (Fig. 6B). Thus, a singleadministration of the targeted nanoparticle DDS achievedtumor growth delay in irradiated tumors that was signifi-cantly greater than conventional systemic chemotherapyand radiation.In conclusion, our results indicate that administration of

XRT to tumors and tumor vasculature causes migration ofGRP78 to the cell surface where the nanoparticle-GIRLRGDDS specifically delivers paclitaxel to the radiated site. Bycombining the controllable, sustained drug release of the na-noparticle with the newly identified GIRLRG targeting pep-tide, we were able to specifically target chemotherapeuticsdirectly to an XRT-inducible receptor causing significant tu-mor cell death. The receptor identified for the peptide,GRP78, is an ideal target for our nanoparticle-peptide DDSbecause it is inducible by XRT (Fig. 2). Even after a singleadministration of the nanoparticle-GIRLRG complex, pacli-taxel can be detected in radiated tumors after 3 weeks, whichtranslates into significantly increased levels of apoptosis andtumor growth delay (Figs. 5 and 6). Thus, we have used novelnanotechnology in vivo to produce a significant increase inthe efficacy of cancer treatment over current clinical models.We expect our targeted nanoparticle DDS to have clinicalutility, and with further investigation hope to implementour system into clinical trials.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

We thank Samuel Spratt for figure illustrations; Erkki Ruoslahti (BurnhamInstitute) for the gift of T7 phage-based random peptide library; VanderbiltUniversity Medical Center Cell Imaging Shared Resource Center, VanderbiltAcademic Fund Venture Capital for Proteomics, and Vanderbilt MassSpectrometry Core for experimental support; and Jessica Huamani and AllieFu for technical support. Roberto Diaz is a recipient of the Leonard B.Holman Research Pathway fellowship.

Grant Support

Department of Defense Breast Cancer Research Program grant BC061828(R. Diaz); National Science Foundation Career Award CHE-0645737 (E. Harth);start-up funds from Vanderbilt University (E. Harth) and Emory University(R. Diaz); Resident Research Seed Grant from the American Society forRadiation Oncology (R. Diaz); NIH grant R01-CA112385 (D. Hallahan);Vanderbilt In Vivo Cellular and Molecular Imaging Center grantP50CA128323 (D. Hallahan); StarBRITE microgrant from Vanderbilt University(R. Passarella, D. Spratt, and R. Diaz).

The costs of publication of this article were defrayed in part by the paymentof page charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Received 01/28/2010; revised 03/17/2010; accepted 03/30/2010; publishedOnlineFirst 05/18/2010.

Cancer Research

Radiation-Induced GRP78 Mediates Targeted Drug Delivery

References

1. Varmus H. The new era in cancer research. Science 2006;312:1162–5.2. Baselga J. Targeting tyrosine kinases in cancer: the second wave.

Science 2006;312:1175–8.3. Ebos JM, Lee CR, Kerbel RS. Tumor and host-mediated pathways of

resistance and disease progression in response to antiangiogenictherapy. Clin Cancer Res 2009;15:5020–5.

4. Hallahan DE. Radiation-mediated gene expression in the pathogen-esis of the clinical radiation response. Semin Radiat Oncol 1996;6:250–67.

5. Hallahan D, Geng L, Qu SM, et al. Integrin-mediated targeting of drugdelivery to irradiated tumor blood vessels. Cancer Cell 2003;3:63–74.

6. Han Z, Fu A, Wang H, et al. Noninvasive assessment of cancer re-sponse to therapy. Nat Med 2008;14:343–9.

7. Passarella RJ, Zhou L, Phillips JG, Wu H, Hallahan DE, Diaz R.Recombinant peptides as biomarkers for tumor response to mole-cular targeted therapy. Clin Cancer Res 2009;15:6421–9.

8. Gradishar WJ, Tjulandin S, Davidson N, et al. Phase III trial of nano-particle albumin-bound paclitaxel compared with polyethylatedcastor oil-based paclitaxel in women with breast cancer. J Clin Oncol2005;23:7794–803.

9. Henderson IC, Bhatia V. Nab-paclitaxel for breast cancer: a newformulation with an improved safety profile and greater efficacy.Expert Rev Anticancer Ther 2007;7:919–43.

10. Morris PG, Fornier MN. Novel anti-tubulin cytotoxic agents for breastcancer. Expert Rev Anticancer Ther 2009;9:175–85.

11. van der Ende A, Croce T, Hamilton S, Sathiyakumar V, Harth E.Tailored polyester nanoparticles: post-modification with dendritictransporter and targeting units via reductive amination and thiol-ene chemistry. Soft Matter 2009;5:1417–25.

12. van der Ende AE, Sathiyakumar V, Diaz R, Hallahan DE, Harth E. Lin-ear release nanoparticle devices for advanced targeted cancer ther-apies with increased efficacy. Polym Chem 2010;1:93–6.

13. Arap W, Pasqualini R, Ruoslahti E. Cancer treatment by targeteddrug delivery to tumor vasculature in a mouse model. Science1998;279:377–80.

14. Diaz R, Passarella RJ, Hallahan DE. Determining glioma response toradiation therapy using recombinant peptides. Expert Rev AnticancerTher 2008;8:1787–96.

15. He L, Kim SO, Kwon O, et al. ATM blocks tunicamycin-inducedendoplasmic reticulum stress. FEBS Lett 2009;583:903–8.

16. Lee HK, Xiang C, Cazacu S, et al. GRP78 is overexpressed inglioblastomas and regulates glioma cell growth and apoptosis.Neuro-oncol 2008;10:236–43.

www.aacrjournals.org

17. Arap MA, Lahdenranta J, Mintz PJ, et al. Cell surface expression ofthe stress response chaperone GRP78 enables tumor targeting bycirculating ligands. Cancer Cell 2004;6:275–84.

18. Fu Y, Lee AS. Glucose regulated proteins in cancer progression, drugresistance and immunotherapy. Cancer Biol Ther 2006;5:741–4.

19. Gonzalez-Gronow M, Selim MA, Papalas J, Pizzo SV. GRP78: a mul-tifunctional receptor on the cell surface. Antioxid Redox Signal 2009;11:2299–306.

20. Dong D, Ni M, Li J, et al. Critical role of the stress chaperone GRP78/BiP in tumor proliferation, survival, and tumor angiogenesis in trans-gene-induced mammary tumor development. Cancer Res 2008;68:498–505.

21. Lee AS. GRP78 induction in cancer: therapeutic and prognosticimplications. Cancer Res 2007;67:3496–9.

22. Lee E, Nichols P, Spicer D, Groshen S, Yu MC, Lee AS. GRP78 as anovel predictor of responsiveness to chemotherapy in breast cancer.Cancer Res 2006;66:7849–53.

23. Pyrko P, Schonthal AH, Hofman FM, Chen TC, Lee AS. The unfoldedprotein response regulator GRP78/BiP as a novel target for increas-ing chemosensitivity in malignant gliomas. Cancer Res 2007;67:9809–16.

24. Mintz PJ, Kim J, Do KA, et al. Fingerprinting the circulating repertoireof antibodies from cancer patients. Nat Biotechnol 2003;21:57–63.

25. Quinones QJ, de Ridder GG, Pizzo SV. GRP78: a chaperone withdiverse roles beyond the endoplasmic reticulum. Histol Histopathol2008;23:1409–16.

26. Liu Y, Steiniger SC, Kim Y, Kaufmann GF, Felding-Habermann B,Janda KD. Mechanistic studies of a peptidic GRP78 ligand for cancercell-specific drug delivery. Mol Pharm 2007;4:435–47.

27. Kim Y, Lillo AM, Steiniger SC, et al. Targeting heat shock proteins oncancer cells: selection, characterization, and cell-penetrating proper-ties of a peptidic GRP78 ligand. Biochemistry 2006;45:9434–44.

28. Yoneda Y, Steiniger SC, Capkova K, et al. A cell-penetrating peptidicGRP78 ligand for tumor cell-specific prodrug therapy. Bioorg MedChem Lett 2008;18:1632–6.

29. Li Y, Pan S, Zhang W, Du Z. Novel thermo-sensitive core-shell nano-particles for targeted paclitaxel delivery. Nanotechnology 2009;20:65104.

30. Svenson S, Tomalia DA. Dendrimers in biomedical applications—reflections on the field. Adv Drug Deliv Rev 2005;57:2106–29.

31. van der Ende AE, Kravitz EJ, Harth E. Approach to formation of mul-tifunctional polyester particles in controlled nanoscopic dimensions.J Am Chem Soc 2008;130:8706–13.

Cancer Res; 70(11) June 1, 2010 4559