Stimuli-responsive clustered nanoparticles forimproved tumor penetration and therapeutic efficacyHong-Jun Lia,1, Jin-Zhi Dua,b,1, Xiao-Jiao Dua,1, Cong-Fei Xuc, Chun-Yang Suna, Hong-Xia Wanga, Zhi-Ting Caoc,Xian-Zhu Yanga, Yan-Hua Zhua, Shuming Nieb,2, and Jun Wanga,c,d,2

aCAS Center for Excellence in Nanoscience, School of Life Sciences and Medical Center, University of Science and Technology of China, Hefei, Anhui230027, People’s Republic of China; bDepartment of Biomedical Engineering, Emory University and Georgia Institute of Technology, Atlanta, GA 30322;cHefei National Laboratory for Physical Sciences at the Microscale, University of Science and Technology of China, Hefei, Anhui 230027, People’sRepublic of China; and dInnovation Center for Cell Signaling Network, University of Science and Technology of China, Hefei, Anhui 230027, People’sRepublic of China

Edited by Mark E. Davis, California Institute of Technology, Pasadena, CA, and approved March 3, 2016 (received for review November 8, 2015)

A principal goal of cancer nanomedicine is to deliver therapeuticseffectively to cancer cells within solid tumors. However, there area series of biological barriers that impede nanomedicine fromreaching target cells. Here, we report a stimuli-responsive clusterednanoparticle to systematically overcome these multiple barriers bysequentially responding to the endogenous attributes of the tumormicroenvironment. The smart polymeric clustered nanoparticle(iCluster) has an initial size of ∼100 nm, which is favorable for longblood circulation and high propensity of extravasation through tu-mor vascular fenestrations. Once iCluster accumulates at tumor sites,the intrinsic tumor extracellular acidity would trigger the dischargeof platinum prodrug-conjugated poly(amidoamine) dendrimers (di-ameter ∼5 nm). Such a structural alteration greatly facilitates tumorpenetration and cell internalization of the therapeutics. The internal-ized dendrimer prodrugs are further reduced intracellularly to re-lease cisplatin to kill cancer cells. The superior in vivo antitumoractivities of iCluster are validated in varying intractable tumor mod-els including poorly permeable pancreatic cancer, drug-resistant can-cer, and metastatic cancer, demonstrating its versatility and broadapplicability.

nanomedicine | particle size | tumor penetration | tumor extracellular pH |stimuli responsive

Over the past few decades, nanomedicine has emerged as apromising means to deliver anticancer therapeutics to tumors

as a result of its preferential and selective accumulation at tumorsites via the enhanced permeability and retention (EPR) effect(1–4). However, cancer nanomedicine encounters a series ofbiological barriers from the site of i.v. injection to the site ofaction (5). These challenges can be briefly summarized as circu-lation in the blood stream, extravasation from blood vessels andaccumulation at tumor sites, deep penetration into the tumorinterstitium, internalization by cancer cells, and intracellular drugrelease (6). These obstacles existing in the body could considerablyprevent nanomedicine from reaching its targets in a sufficient drugconcentration (7, 8). To overcome these barriers, a variety of strat-egies have been envisioned (9–18). Despite great advances, thesestrategies have mainly focused on one or a few biological barriersand led to suboptimal therapeutic effect.It is known that physical properties of nanoparticles such as

size, shape, and surface charge have profound effects on systemictransport of the nanoparticles in solid tumors (19–21). Due toaberrant vasculature, elevated interstitial fluid pressure, and denseextracellular matrix in the tumor microenvironment, nanoparticleshave to overcome considerable interstitial transport hindrance toachieve deep and uniform tumor penetration (7). Such a transportprocess highly relies on much slower diffusion rather than fasterconvective transport because of the inherently large sizes ofnanoparticles compared with small therapeutics (22). Particlesize plays a vital role in dominating the penetration of nanoparticlesinto tumor tissue as diffusion scales inversely with particle size(23–25). Larger nanoparticles, despite being more advantageous

for improved pharmacokinetics and high propensity of vascularextravasation (26), are inherently unfavorable for tumor penetra-tion due to their huge diffusional hindrance in the tumor inter-stitial space (23, 24). This is also recognized as a reason for theclinically approved Doxil (∼90 nm) only showing modest thera-peutic benefits (27). In contrast, smaller nanoparticles show muchbetter tumor penetration (27–30), but very small particles typicallysuffer from short half-life time and insufficient tumor accumula-tion because of their rapid clearance (31, 32). Therefore, an idealdelivery system should be relatively larger in its initial size toachieve longer circulation and selective extravasation (33), but once“docking” at tumor sites, it should be switchable to small particlesto facilitate tumor penetration. Such a requirement has promotedthe recent development of stimuli-responsive nanoparticles thatare able to shrink their sizes by responding to enzymes or UVlight (15, 34). Although conceptually impressive, those systemsare still at an early stage of development. Developing moresophisticated nanomedicines that are capable of systematicallyovercoming the sequential barriers in a concerted fashion, andare also applicable to a broad range of cancer types, remainsformidably challenging.Here we report the development of stimuli-responsive clus-

tered nanoparticles to systematically overcome these multiple

Significance

Successively overcoming a series of biological barriers thatcancer nanotherapeutics would encounter upon intravenousadministration is required for achieving positive treatmentoutcomes. A hurdle to this goal is the inherently unfavorabletumor penetration of nanoparticles due to their relatively largesizes. We developed a stimuli-responsive clustered nanoparticle(iCluster) and justified that its adaptive alterations of physico-chemical properties (e.g. size, zeta potential, and drug releaserate) in accordance with the endogenous stimuli of the tumormicroenvironment made possible the ultimate overcoming ofthese barriers, especially the bottleneck of tumor penetration.Results in varying intractable tumor models demonstrated sig-nificantly improved antitumor efficacy of iCluster than its controlgroups, demonstrating that overcoming these delivery barrierscan be achieved by innovative nanoparticle design.

Author contributions: H.-J.L., J.-Z.D., X.-J.D., S.N., and J.W. designed research; H.-J.L., J.-Z.D.,X.-J.D., C.-F.X., C.-Y.S., H.-X.W., Z.-T.C., X.-Z.Y., and Y.-H.Z. performed research; C.-Y.S.contributed new reagents/analytic tools; H.-J.L., J.-Z.D., X.-J.D., S.N., and J.W. analyzeddata; and H.-J.L., J.-Z.D., S.N., and J.W. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1H.-J.L., J.-Z.D., and X.-J.D. contributed equally to this work.2To whom correspondence may be addressed. Email: jwang699@ustc.edu.cn or snie@emory.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1522080113/-/DCSupplemental.

4164–4169 | PNAS | April 12, 2016 | vol. 113 | no. 15 www.pnas.org/cgi/doi/10.1073/pnas.1522080113

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

20, 2

020

barriers to cancer chemotherapy. The nanoparticles were con-structed through molecular assembly of platinum (Pt) prodrug-conjugated poly(amidoamine)-graft-polycaprolactone (PCL-CDM-PAMAM/Pt) with PCL homopolymer and poly(ethyleneglycol)-b-poly(e-caprolactone) (PEG-b-PCL) copolymer (Fig. 1A and B). PEG-b-PCL is used to offer the stealth layer, whereasPCL is chosen to control the size and stability. The nanoparticles areable to function adaptively in the body through precisely respondingto the physiological pH, tumor extracellular acidity, and intracellularreductive environment, respectively (Fig. 1B). At physiological pH,the clustered nanoparticles hold the size around 100 nm and havehigh propensity for long blood circulation and enhanced tumor ac-cumulation through the EPR effect. Then, the acidic tumor extra-cellular pH (pHe ∼6.5–7.2) (35, 36) triggers the release of smallPAMAM prodrugs (∼5 nm) that enable deep and uniform tumorpenetration to reach more cancer cells. Finally, the PAMAMprodrugs can be rapidly reduced in the reductive cytosol to re-lease active and potent cisplatin to kill cancer cells and lead torobust antitumor efficacy (37).

ResultsPreparation and Characterization of the Clustered Nanoparticles. Toprepare the clustered nanoparticles, the polymer components PCL,PEG-b-PCL, and PCL-CDM-PAMAM/Pt were synthesized. Thestructures and molecular weights of PEG-b-PCL and PCL were char-acterized (SI Appendix, Figs. S1 and S2). For PCL-CDM-PAMAM/Ptsynthesis, PCL was first reacted with 2-propionic-3-methylmaleicanhydride (CDM) to produce PCL-CDM (SI Appendix, Fig. S3).A Pt prodrug c,c,t-[Pt(NH3)2Cl2(OH)(O2CCH2CH2CO2H)] wasconjugated to PAMAM to afford PAMAM/Pt, which was fur-ther coupled to PCL-CDM through the reaction of the aminogroups of PAMAM with the CDM anhydride residue (SI Ap-pendix, Scheme S1). The resultant amide bond is acid labile(38) and will be cleaved at pHe to release PAMAM attumor sites.The Pt-containing pH-instable clustered nanoparticle (iCluster/Pt)

was prepared from coassembly of PCL-CDM-PAMAM/Pt, PEG-

b-PCL, and PCL by nanoprecipitation method. The weight ratio ofPCL-CDM-PAMAM/Pt:PEG-b-PCL:PCL was optimized as 1:1:1based on size and size distribution (SI Appendix, Table S1 and Fig.S4). Dynamic light scattering (DLS) indicated that the diameter ofiCluster/Pt was around 104.1 nm (Fig. 1C), which was consistentwith the transmission electron microscopy (TEM) observation(Fig. 1D). TEM images showed clearly that iCluster/Pt had araspberry-like structure, presumably due to the presence ofPAMAM dendrimers surrounding the hydrophobic core. EachiCluster/Pt contained 108 PAMAM and 719 platinum drugs, asestimated by static light scattering. For comparison, we prepared apH-stable clustered nanoparticle (Cluster/Pt) by replacing PCL-CDM-PAMAM/Pt with its nonresponsive counterpart PCL-PAMAM/Pt (SI Appendix, Scheme S2). Each Cluster/Pt contained90 PAMAM and 573 platinum drugs and showed similar mor-phology, size, and zeta potential as iCluster/Pt (Fig. 1 E and F andSI Appendix, Table S2).One key design of iCluster/Pt is its sensitivity to biological

stimuli such as the acidic pHe and elevated intracellular redoxmilieu. To test its response to pHe, we incubated iCluster/Pt in atumor-acidity mimic phosphate buffer (PB) solution at pH 6.8over predetermined time and observed its morphological evo-lution under TEM. After a 4-h incubation, the raspberry-likemorphology of iCluster/Pt was partially deformed and smallparticles appeared in the solution (Fig. 1D, Middle). After anadditional incubation for 24 h, the raspberry structure was almostcompletely disintegrated (Fig. 1D, Right and SI Appendix, Fig.S5) and changed to a smooth surface that was analogous tonanoparticles formed by PEG-PCL and PCL (SI Appendix, Fig.S6A). Meanwhile, plenty of small nanoparticles were observed inthe surroundings, with size comparable to PAMAM (SI Appen-dix, Fig. S6B), suggesting that small PAMAM dendrimers werereleased upon cleavage of the amide bond at pH 6.8. In contrast,Cluster/Pt maintained its initial morphology and size underidentical conditions (Fig. 1F).Next, high-performance liquid chromatography (HPLC) was

used to monitor and compare the release kinetics of PAMAM

PCL-CDM-PAMAM/Pt

PCL

PEG-b-PCL

Self-assembly Tumor acidityIntracellular Reduction

4 h 24 h

4 h 24 h

Size (nm)10 100 1000

048121620

)%(

ytisnetnI

10 100 1000048

121620

Size (nm)

)%(

ytisnet nI

B

F

D G

HE

C

0 4 8 12 160

20406080

100

Pt drugPAMAM

Time (h)Cum

ulat

ive

r ele

ase

(%)

PBpH 7.4

PBpH 6.8

Ascorbic acid 5 mM, pH 7.4

0 4 8 12 160

20406080

100

Pt drugPAMAMPB

pH 7.4

PBpH 6.8

Ascorbic acid 5 mM, pH 7.4

Cum

ulat

ive

rele

ase

(%)

Time (h)

0 h

0 h

PCL-CDM-PAMAM/Pt

A

Fig. 1. Preparation and physicochemical propertiesof the clustered nanoparticles. (A) Chemical structureof PCL-CDM-PAMAM/Pt. (B) Self-assembly and struc-tural change of iCluster/Pt in response to tumoracidity and intracellular reductive environment.(C and E) DLS distributions of iCluster/Pt (C) andCluster/Pt (E). (D and F) TEM images of iCluster/Pt(D) and Cluster/Pt (F) treated in PB at pH 6.8 for 0, 4,and 24 h, respectively. (Scale bar, 100 nm and for theInset images, 50 nm.) (G and H) PAMAM (green line)and platinum drug (red line) release from iCluster (G)and Cluster (H) under three different conditions,which include PB at pH 7.4 to mimic a neutral envi-ronment, PB at pH 6.8 to mimic a tumor extracellularenvironment, and ascorbic acid solution (5 mM, pH 7.4)to mimic an intracellular redox environment. PAMAMrelease was quantified by HPLC, whereas platinumrelease was determined by ICP-MS.

Li et al. PNAS | April 12, 2016 | vol. 113 | no. 15 | 4165

MED

ICALSC

IENCE

S

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

20, 2

020

from fluorescein (Flu)-labeled clustered nanoparticles (iClusterFluand ClusterFlu). At pH 7.4, iClusterFlu could generally maintainits stability with ∼12% PAMAM release after a 4-h incubation.However, incubation at pH 6.8 resulted in ∼60% cumulativerelease by another 4 h (Fig. 1G) and ∼90% release by 24 h (SIAppendix, Fig. S6C), indicating much faster release of PAMAMat acidic pH. In contrast, ClusterFlu showed minimal release ateither pH 7.4 or 6.8 (Fig. 1H). The nanoparticles are designed torelease cisplatin specifically in the redox environment (39). Totest the release of cisplatin, we incubated the nanoparticles in PBsolutions (pH 7.4 and 6.8), and ascorbic acid solution (5 mM, pH7.4), an intracellular redox environment mimic solution (40),respectively, and quantified Pt drug release via inductively cou-pled plasma mass spectrometry (ICP-MS). Both nanoparticlesshowed minimal drug release in PB solutions regardless of pH,but rapid release in the redox environment (Fig. 1 G and H).Blood plasma showed minimal effect on the stability and pHresponsiveness of iCluster (SI Appendix, Fig. S7).

Penetration and Cell Apoptosis in Multicellular Spheroids. We chosemulticellular spheroids (MCSs) derived from BxPC-3 humanpancreatic cancer cells as an in vitro tumor model to evaluatethe penetration and cell killing efficacy of the clustered nano-particles. Compared with single-layer adherent cells, MCSs havebeen proposed as a versatile 3D model to study tumor biologyand to screen therapeutic agents (41). iCluster and Clusterwere dual labeled with two dyes (denoted as RhBiClusterFlu andRhBClusterFlu), in which PAMAM was labeled with fluorescein(Flu, green), whereas the PCL component of the hydrophobiccore was labeled with rhodamine B (RhB, red). MCSs were in-cubated with RhBiClusterFlu or

RhBClusterFlu for 15 min, 4 h, and24 h at pH 6.8, then washed and observed under confocal laserscanning microscopy (CLSM, Fig. 2A and SI Appendix, Fig. S8).For RhBiClusterFlu treatment, the red fluorescence from the hy-drophobic core attached to the periphery of MCSs, and no no-ticeable fluorescence was detected in the internal area by 24 h.Instead, the green signals from PAMAM inside MCSs clearlyincreased over time, demonstrating the improved penetration ofPAMAM. By contrast, both red and green fluorescence signalsfrom nonresponsive RhBClusterFlu were on the peripheral regionof the MCSs by 24 h, revealing limited penetration of largerparticles. Quantitative analysis indicated that more than 10-foldhigher green fluorescence was detected in RhBiClusterFlu-treatedMCSs than in RhBClusterFlu-treated ones (Fig. 2B). These results

provide clear evidence that the size-changeable iCluster hasbetter tumor penetration.Next, cell internalization of both nanoparticles was studied by

flow cytometry (FACS) and ICP-MS, respectively. In FACS anal-ysis, MCSs were treated with iClusterFlu, ClusterFlu, or PAMAMFluat pH 6.8 for 4 h and 24 h. Compared with ClusterFlu treatment,the population of positive cells treated with iClusterFlu was 1.7-foldhigher and 3-fold higher at 4 h (P < 0.05) and 24 h (P < 0.001),respectively (Fig. 2C and SI Appendix, Fig. S9). For ICP-MSanalysis, Fig. 2D shows that iCluster/Pt treatment resulted in sig-nificantly higher intracellular accumulation of Pt than Cluster/Pttreatment (2.6-fold, P < 0.001). Moreover, DNA of MCS cells wereisolated after a 24-h incubation, and the amount of Pt binding tothese DNA in MCS cells receiving iCluster/Pt treatment was sig-nificantly higher than that receiving Cluster/Pt treatment (3.6-fold,P < 0.001, Fig. 2E). Apoptosis results indicated that iCluster/Pttreatment showed 38.6% total cell apoptosis, which was signifi-cantly higher than Cluster/Pt treatment (14.3%, P < 0.001, Fig. 2Fand SI Appendix, Fig. S10). Of note, in all experiments, iClustertreatment showed comparable efficacy to PAMAM treatment.

Antitumor Activity and Tumor Penetration in a BxPC-3 PancreaticTumor Model. The superior performance of iCluster/Pt in MCSsfurther compelled us to study its in vivo activities. We first studiedthe pharmacokinetic profiles of the clustered nanoparticles. Asshown in Fig. 3A, both iCluster/Pt and Cluster/Pt exhibited pro-longed half-life time (T1/2 > 10 h) in the bloodstream comparedwith PAMAM/Pt and cisplatin, which is reasonable becauseiCluster/Pt and Cluster/Pt are PEGylated nanoparticles. Otherpharmacokinetic parameters also revealed better performance ofthe two PEGylated nanoparticles (SI Appendix, Table S3). Next,we studied the antitumor activity of iCluster/Pt in nude micebearing BxPC-3 human pancreatic tumors, which is recognized asa notoriously poorly permeable and intractable tumor model (27).As shown in Fig. 3B, free cisplatin and PAMAM/Pt displayedmodest therapeutic efficacy, with 38% and 45% tumor inhibitionversus PBS control, whereas Cluster/Pt showed 57% tumor in-hibition. In contrast, iCluster/Pt exhibited significant suppressionof tumor growth, reaching 88% tumor suppression (P < 0.001,versus Cluster/Pt treatment). The average weight of the tumormass excised at the end of treatment also demonstrated the sametrend (SI Appendix, Fig. S11A). Histological analyses, includinghematoxylin and eosin (H&E), proliferating cell nuclear antigen(PCNA), and terminal deoxynucleotidyl transferase-mediateddeoxyuridine triphosphate nick end (TUNEL) staining, also

Fig. 2. In vitro penetration and cell killing efficacyof clustered nanoparticles in BxPC-3 MCSs. (A) Pen-etration of RhBiClusterFlu and RhBClusterFlu in MCSs atpH 6.8 after a 4-h or 24-h incubation. The areamarked with white circles was considered the insidearea. (Scale bar, 200 μm.) (B) Mean fluorescence in-tensity (MFI) of green signals in the inside area ofMCSs. (C) Quantification of fluorecein-positive MCScells after incubation with different formulations atpH 6.8 for 4 h or 24 h. (D) Quantification of Pt con-tent in MCSs after incubation with various formula-tions at pH 6.8 for 24 h. (E) Quantification of Pt contentin DNA of MCS cells after various treatments at pH 6.8for 24 h. (F) Apoptotic cells induced by differenttreatments at pH 6.8 for 24 h. All data are presented asmean ± SD (n = 3). *P < 0.05, ***P < 0.001.

4166 | www.pnas.org/cgi/doi/10.1073/pnas.1522080113 Li et al.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

20, 2

020

indicated that iCluster/Pt treatment markedly reduced prolif-erative cells while increasing apoptotic cells (SI Appendix,Fig. S12). To investigate dose effect of iCluster/Pt on theinhibition of BxPC-3 tumors, a separate set of in vivo studies withinjection dose varying from 1.5 to 6.0 mg/kg was performed. Asshown in Fig. 3C, compared with PBS control, all of the iCluster/Pt treatments showed dramatic tumor growth suppression, withinhibition rate of 71%, 92%, and 107% for the dose of 1.5, 4.5,and 6.0 mg/kg, respectively. It is noteworthy that tumor shrinkagewas observed for the 6.0 mg/kg treatment. Slight weight loss wasobserved for mice receiving free cisplatin and the 6.0 mg/kgtreatments, whereas no obvious weight loss from other treatmentswas observed (SI Appendix, Fig. S11 B and C). Additionally, tissuestaining and blood tests revealed that the clustered nanoparticlesdid not cause detectable toxicities to the mice (SI Appendix, Figs.S13 and S14).The distribution of the Pt drug in other major organs was not

significantly different between iCluster/Pt and Cluster/Pt treat-ments (SI Appendix, Fig. S15). However, iCluster/Pt was able tosignificantly enhance Pt drug accumulation in tumor tissuescompared with other treatments, with ∼2-fold higher drug con-centration than Cluster/Pt (P < 0.01 for 12 h, and P < 0.05 for24 h) and at least 7-fold higher than free cisplatin and PAMAM/Pt(Fig. 3D). To get more details, we digested the tumor mass into

individual cells and quantified Pt content in these cells by ICP-MS.iCluster/Pt showed two to three times higher internalization ofPt than Cluster/Pt (P < 0.05 for 12 h and P < 0.001 for 24 h; Fig. 3E).To further distinguish tumor cells from other stromal cells in theheterogeneous tumor tissue, we established tumor models by sub-cutaneously injecting green fluorescent protein (GFP)-expressingBxPC-3 cells in nude mice. After injecting these drug-containingformulations, GFP-positive tumor cells were isolated and sortedby FACS and subjected to ICP-MS to determine Pt content. Inthese GFP-positive tumor cells, iCluster/Pt still showed signifi-cantly higher Pt content than Cluster/Pt at 12 h (2.7-fold, P < 0.05)and 24 h (3-fold, P < 0.001) (Fig. 3F).iCluster/Pt and Cluster/Pt have similar size, surface property,

and pharmacokinetics in the bloodstream. The enhanced depositof iCluster/Pt in tumor tissues and tumor cells is presumed to beassociated with its superior tumor penetration. To test ourhypothesis, we used immunofluorescence staining and real-timeCLSM observation to study the intratumoral microdistribution ofRhBiClusterFlu and RhBClusterFlu. In the immunofluorescencestaining study, for RhBiClusterFlu treatment, the green fluorescencefrom PAMAM showed a uniform perfusion in the tumor inter-stitium, whereas the red fluorescence was highly colocalized withthe blood vessels (yellow). This suggests that RhBiClusterFlu canrelease PAMAM at tumor sites and the released PAMAM enablesefficient extravasation and penetration into deep tumor space,whereas the larger residual nanoparticles are mainly restricted inthe blood vessels or the peripheral areas (Fig. 4A, Upper). For thenonresponsive RhBClusterFlu, both green and red signals werecolocalized with the blood vessels, indicating its inability to diffuseinto tumor space (Fig. 4A, Lower). The quantitative analysis ofoverlap coefficient of red and yellow, as well as green and yellowalso showed the same trend (SI Appendix, Fig. S16). These resultssubstantiate previous observations that smaller nanoparticles aremore advantageous for deep tumor penetration than larger onesbecause of their reduced diffusional hindrance (23, 42). To furthervisualize real-time extravasation and tumor penetration of thesenanoparticles, intravital CLSM was used. For RhBiClusterFlu in-jection (Fig. 4B), strong green and red fluorescence signals wereconfined in the blood vessels at 10 min postinjection. By 90 min,the green fluorescence in the blood vessels weakened, whereasmore green signals extravasated from the blood vessels and dis-tributed dispersedly in the surroundings. In contrast, the red fluo-rescence still remained within the blood vessels. To quantitativelyanalyze the spatiotemporal evolution of the fluorescence, we nor-malized the fluorescence intensity of each color to its initial in-tensity at 10 min to afford the time and penetration depth-dependent profiles (Fig. 4C). At 90 min postinjection, nearly 50%of the original intensity of green fluorescence from PAMAM couldbe detected until 70 μm from the blood vessels, whereas red signalsfrom the larger residual nanoparticles were not detectable beyondthe blood vessels. This trend became even more evident by 240 minpostinjection. The green fluorescence inside blood vessels di-minished markedly while spreading more uniformly in the tumorinterstitial space. The quantitative data showed that 25% of theoriginal intensity was still detectable at 160 μm from the blood ves-sels. The red fluorescence inside the blood vessels weakened as well,but no red fluorescence was observed beyond the vascular wall. Incontrast, both the green and red signals from the RhBClusterFluwere confined to the blood vessels over 240 min, suggestingpoor tumor penetration of large RhBClusterFlu nanoparticles (SIAppendix, Fig. S17).

Antitumor Activity in Drug-Resistant and Metastatic Tumor Models.To verify the broad applicability of iCluster/Pt in cancer chemo-therapy, we further investigated its antitumor activities in cisplatin-resistant A549R human lung cancer and 4T1 metastatic murinebreast cancer models. In the A549R model, different formulationsshowed remarkable differences in antitumor activities. Compared

F

D

E

A B

C

0 5 10 15 20 250.1

1

10

100Cluster/Pt iCluster/PtCisplatin PAMAM/Pt

Time after injection (h)

%of

inje

ctio

ndo

se

***

0 4 8 12 160

100

200300

400500

PBS

CisplatiniCluster

Cluster/PtiCluster/Pt

PAMAM/Pt

Days after first injection

Tum

or v

olum

e (m

m3 )

0 4 8 12 160

100

200

300

400

500 1.5 mg/kgPBS

4.5 mg/kg6.0 mg/kg

Days after first injection

Tum

or v

olum

e (m

m3 )

0

400

800

1200

Pt c

onte

nt in

tum

ortis

sue

(ng/

g tis

sue)

12 h 24 h

***

CisplatinPAMAM/Pt

iCluster/PtCluster/Pt

12 h 24 h0

200

400

600

*****

Pt c

onte

nt in

tota

l tum

ortis

sue

cells

(ng/

108

cells

)

*****

12 h 24 h0

200

400

600

Pt c

onte

nt in

GFP

+

tum

or c

ells

(ng/

108

cells

)

Cluster/PtiCluster/Pt

Cluster/PtiCluster/Pt

Fig. 3. In vivo antitumor activity of iCluster/Pt in a BxPC-3 human pancreatictumor model. (A) Pharmacokinetics of the formulations. These formulationswere i.v. injected into ICR mice at a Pt dose of 60 μg per mouse. (B) Growthinhibition of BxPC-3 tumors by different treatments. Mice were i.v. admin-istered at a Pt dose of 3 mg/kg on days 0, 2, and 4. ***P < 0.001. (C) Doseeffect of iCluster/Pt on tumor growth inhibition. Mice were i.v. administeredon days 0, 2, and 4. (D) Quantification of Pt content in tumor tissue. Formu-lations were administered i.v. at a Pt dose of 60 μg per mouse. Mice were killedat 12 h and 24 h postinjection, and tumors were excised. *P < 0.05, **P < 0.01.(E and F) Quantification of Pt content in tumor tissue cells (E) and GFP-positivetumor cells (F). For E and F, the tumor was established by s.c. injecting greenfluorescent protein (GFP)-expressing BxPC-3 cells. At 12 h and 24 h after in-jection of the formulations, the tumors were excised, digested, and subjected toFACS to sort the total tumor cells and the GFP+ tumor cells. **P < 0.01, ***P <0.001. Data are presented as mean ± SD n = 3 for A and D–F; n = 5 for B and C.

Li et al. PNAS | April 12, 2016 | vol. 113 | no. 15 | 4167

MED

ICALSC

IENCE

S

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

20, 2

020

with PBS control, free cisplatin exhibited minimal tumor growth in-hibition (only 10% inhibition), whereas PAMAM/Pt showed a slightlybetter effect (∼20% inhibition). In comparison, the long-circulatingCluster/Pt showed considerable enhancement in tumor growthinhibition (60% inhibition). However, the most effective antitu-mor effect was achieved by iCluster/Pt treatment, reaching 95%inhibition (Fig. 5A, P < 0.001). Of note, the significance betweeniCluster/Pt and Cluster/Pt appeared as early as day 9 post-injection. No obvious body weight loss was observed for thenanoparticle formulations (SI Appendix, Fig. S18), and histo-logical staining also confirmed the improved therapeutic effectof iCluster/Pt, showing reduction in proliferation while in-creasing the apoptosis of tumor cells (SI Appendix, Fig. S19).To further extend the applicability of our strategy to combat

metastatic cancer, we established a highly invasive and metastatic4T1 orthotopic tumor model, which is known to be more aggres-sive and more refractory to chemotherapy than a s.c. tumor model(43). The mice were treated with varying Pt-containing formulations,and their survival curves were recorded. Compared with PBS andblank iCluster control groups, all other treatments showed improvedmedian survival time. In particular, the iCluster/Pt treatment im-proved survival time by 74.2%, with significantly longer time to endpoint than that of Cluster/Pt (Fig. 5B and SI Appendix, Table S4). Thecomparison of intratumoral microdistribution of RhBiClusterFlu andRhBClusterFlu in A549R and 4T1 tumor tissues also demonstratedthat iCluster showed much better tumor penetration than Clusterbecause of their pHe-activated PAMAM release at the tumor site (SIAppendix, Figs. S20 and S21), once again indicating that the improved

antitumor activities of iCluster are highly associated with enhancedtumor penetration.

DiscussionDespite the fact that nanoparticle-based therapeutics are amena-ble to preferential accumulation in solid tumors by taking ad-vantage of the EPR effect, they encounter a series of sequentialbiological barriers upon i.v. administration, which severely impedethe achievement of optimal therapeutic outcomes. To adequatelyaddress these barriers and achieve effective therapy, nanoparticlesmust be rationally designed to overcome substantial interstitialtransport hindrance brought about by their inherently large sizesto realize deep and uniform tumor penetration (7). In this study,our iCluster system enables its basic physicochemical properties toadaptively change in response to the endogenous stimuli of thetumor microenvironment to accomplish improved therapeutic ef-ficacy by successively increasing blood circulation and tumor vas-cular extravasation, improving tumor penetration, facilitating cellinternalization, and accelerating intracellular drug release.Our results demonstrate that the decisive step for the effec-

tiveness of iCluster is its robust tumor penetration achievedthrough pHe-triggered shattering of small PAMAM dendrimersat tumor sites (Figs. 3 and 4). It has been validated that thepenetration of nanoparticles in tumor space relies heavily onparticle size, with the consensus that smaller particles have im-proved tissue penetration (26, 33, 44). Such progress has recentlyinspired interest in developing size-shrinkable anticancer drugdelivery systems (15, 34, 45, 46). Compared with previous stud-ies, our strategy has several unique features. First, previous de-livery systems simply focused on size-shrinkage medicated tumorpenetration, whereas our system is devised to systematicallyovercome a series of barriers including tumor penetration.Achieving this goal is vitally important because these barriers areinterconnected, and simply overcoming one individual barrier isnot adequate to produce proper therapeutic outcomes (42, 47).

Fig. 4. Microdistribution of iCluster and Cluster in BxPC-3 xenograft tumorafter i.v. injection. (A) CLSM images of immunofluorescence showing themicrodistribution of RhBiClusterFlu and RhBClusterFlu in tumor tissue at 4 hpostinjection. PAMAM was labeled with Flu (green), whereas the core of thenanoparticles was labeled with RhB (red), and blood vessels were markedwith platelet endothelial cell adhesion molecule 1 (PECAM-1) and CFL-647secondary antibody (yellow). (Scale bar, 50 μm.) (B) Real-time microdistributionof RhBiClusterFlu in BxPC-3 tumor at 10, 90, and 240 min postinjection. (Scalebar, 100 μm.) (C ) Time and penetration depth-dependent distribution ofRhBiClusterFlu. A region marked with the rectangular frame was selected forthe analysis. The intensity profiles were obtained by normalizing the fluo-rescence intensity of each color to its initial intenstiy at 10 min.

A

B

**** **** ****** ******

00 33 66 99 1212 1515 181800

200200

400400

600600

800800

10001000

PBSPBSiClusteriClusterCisplatinCisplatinPAMAM/PtPAMAM/PtCluster/PtCluster/PtiCluster/PtiCluster/Pt

Days after first injectionDays after first injection

Tum

or v

olum

e (m

mTu

mor

vol

ume

(mm

33 ))

00 101000

2020404060608080

100100

2020 3030 4040 5050 6060Days after Days after tumor tumor inoculationinoculation

% S

urvi

val

% S

urvi

val

Fig. 5. In vivo antitumor activity in drug-resistant and metastatic tumormodels. (A) Inhibition of tumor growth of a A549R cisplatin-resistant humanlung cancer model. Mice were i.v. administered an equivalent platinum doseof 1.5 mg/kg on days 0, 3, and 6. Data are presented as mean ± SD (n = 5).**P < 0.05, ***P < 0.001. (B) Kaplan–Meier plots of the animal survival in 4T1tumor models (n = 10). Mice were treated at a platinum dose of 3 mg/kg viai.v. administration on days 10, 15, and 20 after tumor inoculation.

4168 | www.pnas.org/cgi/doi/10.1073/pnas.1522080113 Li et al.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

20, 2

020

Second, the stimuli that were used to trigger size shrinkagepreviously were either by enzyme or UV light, whose applica-bility, to a certain extent, would be restricted to only a subset ofcancer types owing to either the heterogeneous expression levelsof target enzymes in a specific cancer type or the superficial pen-etration depth of UV light (14). In contrast, we used acidic ex-tracellular pH, a more general hallmark of the microenvironmentof most solid tumors (36, 48), as the stimulus to activate the releaseof small particles at tumor sites. This endows our strategy withmuch broader specificity and applicability, which has been partiallyjustified by the superior antitumor activities in multiple tumormodels. Third, it should be noted that it is the postshrinkage sizerather than the property of size shrinkage that is the pivotal de-terminant of therapeutic efficacy. This is particularly important fortreating poorly permeable solid tumors, because only the sub–30-nmmicelles could penetrate the poorly permeable pancreatic tumorsto achieve an antitumor effect (27). For previous systems, theirfinal sizes after shrinkage were still as large as 40–70 nm (34, 45,46), which is likely to offset the usefulness of size shrinkage fortreating poorly permeable tumors. In our system, the releasedsmall PAMAM dendrimer prodrugs have a size of around 5 nm,

which is highly potent in penetrating the intractable tumors toreach cancer cells that are far away from the blood vessels. In es-sence, our design strategy may open up a new avenue for thecreation of the next generation of nanotherapeutics, representing aparadigm shift in nanoparticle-based drug delivery.

Materials and MethodsDetailed materials and methods are provided in SI Appendix, including thesynthesis of polymers, preparation and characterization of the clusterednanoparticles, stimuli-responsive PAMAM and platinum drug release fromnanoparticles, in vitro penetration, cell internalization, apoptosis in multi-cellular spheroids, in vivo real-time tumor penetration, immunofluorescentstaining, antitumor activities in varying tumor models, histological studies,and statistics. The animal study procedures were approved by the AnimalCare and Use Committee of University of Science and Technology of China.

ACKNOWLEDGMENTS. The authors thank Dr. Xiao-Dong Ye and Miss Jin-Xian Yang for their assistance with static light scattering measurement. Thiswork was supported by the National Basic Research Program of China (973Programs, 2012CB932500, 2015CB932100, and 2013CB933900) and theNational Natural Science Foundation of China (51125012, 51390482, and51503195).

1. Matsumura Y, Maeda H (1986) A new concept for macromolecular therapeutics incancer chemotherapy: Mechanism of tumoritropic accumulation of proteins and theantitumor agent smancs. Cancer Res 46(12 Pt 1):6387–6392.

2. Peer D, et al. (2007) Nanocarriers as an emerging platform for cancer therapy. NatNanotechnol 2(12):751–760.

3. Davis ME, Chen ZG, Shin DM (2008) Nanoparticle therapeutics: An emerging treat-ment modality for cancer. Nat Rev Drug Discov 7(9):771–782.

4. Torchilin VP (2014) Multifunctional, stimuli-sensitive nanoparticulate systems for drugdelivery. Nat Rev Drug Discov 13(11):813–827.

5. Godin B, Tasciotti E, Liu X, Serda RE, Ferrari M (2011) Multistage nanovectors: Fromconcept to novel imaging contrast agents and therapeutics. Acc Chem Res 44(10):979–989.

6. Sanhai WR, Sakamoto JH, Canady R, Ferrari M (2008) Seven challenges for nano-medicine. Nat Nanotechnol 3(5):242–244.

7. Chauhan VP, Jain RK (2013) Strategies for advancing cancer nanomedicine. Nat Mater12(11):958–962.

8. Petros RA, DeSimone JM (2010) Strategies in the design of nanoparticles for thera-peutic applications. Nat Rev Drug Discov 9(8):615–627.

9. Kamaly N, Xiao Z, Valencia PM, Radovic-Moreno AF, Farokhzad OC (2012) Targetedpolymeric therapeutic nanoparticles: Design, development and clinical translation.Chem Soc Rev 41(7):2971–3010.

10. Duncan R (2006) Polymer conjugates as anticancer nanomedicines. Nat Rev Cancer6(9):688–701.

11. Zhu L, Wang T, Perche F, Taigind A, Torchilin VP (2013) Enhanced anticancer activityof nanopreparation containing an MMP2-sensitive PEG-drug conjugate and cell-penetrating moiety. Proc Natl Acad Sci USA 110(42):17047–17052.

12. Murakami M, et al. (2011) Improving drug potency and efficacy by nanocarrier-mediated subcellular targeting. Sci Transl Med 3(64):64ra2.

13. Poon Z, Chang D, Zhao X, Hammond PT (2011) Layer-by-layer nanoparticles with a pH-sheddable layer for in vivo targeting of tumor hypoxia. ACS Nano 5(6):4284–4292.

14. Mura S, Nicolas J, Couvreur P (2013) Stimuli-responsive nanocarriers for drug delivery.Nat Mater 12(11):991–1003.

15. Wong C, et al. (2011) Multistage nanoparticle delivery system for deep penetrationinto tumor tissue. Proc Natl Acad Sci USA 108(6):2426–2431.

16. Du JZ, Du XJ, Mao CQ, Wang J (2011) Tailor-made dual pH-sensitive polymer-doxo-rubicin nanoparticles for efficient anticancer drug delivery. J Am Chem Soc 133(44):17560–17563.

17. Ling D, et al. (2014) Multifunctional tumor pH-sensitive self-assembled nanoparticlesfor bimodal imaging and treatment of resistant heterogeneous tumors. J Am ChemSoc 136(15):5647–5655.

18. Li Y, et al. (2014) A smart and versatile theranostic nanomedicine platform based onnanoporphyrin. Nat Commun 5:4712.

19. Gratton SE, et al. (2008) The effect of particle design on cellular internalizationpathways. Proc Natl Acad Sci USA 105(33):11613–11618.

20. Mitragotri S, Lahann J (2009) Physical approaches to biomaterial design. Nat Mater8(1):15–23.

21. Decuzzi P, et al. (2010) Size and shape effects in the biodistribution of intravascularlyinjected particles. J Control Release 141(3):320–327.

22. Lane LA, Qian X, Smith AM, Nie S (2015) Physical chemistry of nanomedicine: Un-derstanding the complex behaviors of nanoparticles in vivo. Annu Rev Phys Chem 66:521–547.

23. Tang L, et al. (2014) Investigating the optimal size of anticancer nanomedicine. ProcNatl Acad Sci USA 111(43):15344–15349.

24. Dreher MR, et al. (2006) Tumor vascular permeability, accumulation, and penetrationof macromolecular drug carriers. J Natl Cancer Inst 98(5):335–344.

25. Albanese A, Lam AK, Sykes EA, Rocheleau JV, Chan WC (2013) Tumour-on-a-chipprovides an optical window into nanoparticle tissue transport. Nat Commun 4:2718.

26. Perrault SD, Walkey C, Jennings T, Fischer HC, Chan WCW (2009) Mediating tumortargeting efficiency of nanoparticles through design. Nano Lett 9(5):1909–1915.

27. Cabral H, et al. (2011) Accumulation of sub-100 nm polymeric micelles in poorlypermeable tumours depends on size. Nat Nanotechnol 6(12):815–823.

28. Chauhan VP, et al. (2012) Normalization of tumour blood vessels improves the deliveryof nanomedicines in a size-dependent manner. Nat Nanotechnol 7(6):383–388.

29. Huo S, et al. (2013) Superior penetration and retention behavior of 50 nm goldnanoparticles in tumors. Cancer Res 73(1):319–330.

30. Popovi�c Z, et al. (2010) A nanoparticle size series for in vivo fluorescence imaging.Angew Chem Int Ed Engl 49(46):8649–8652.

31. Choi HS, et al. (2007) Renal clearance of quantum dots. Nat Biotechnol 25(10):1165–1170.

32. Choi HS, et al. (2010) Design considerations for tumour-targeted nanoparticles. NatNanotechnol 5(1):42–47.

33. Wang J, et al. (2015) The role of micelle size in tumor accumulation, penetration, andtreatment. ACS Nano 9(7):7195–7206.

34. Tong R, Chiang HH, Kohane DS (2013) Photoswitchable nanoparticles for in vivocancer chemotherapy. Proc Natl Acad Sci USA 110(47):19048–19053.

35. Helmlinger G, Yuan F, Dellian M, Jain RK (1997) Interstitial pH and pO2 gradients insolid tumors in vivo: high-resolution measurements reveal a lack of correlation. NatMed 3(2):177–182.

36. Wang Y, et al. (2014) A nanoparticle-based strategy for the imaging of a broad rangeof tumours by nonlinear amplification of microenvironment signals. Nat Mater 13(2):204–212.

37. Dhar S, Daniel WL, Giljohann DA, Mirkin CA, Lippard SJ (2009) Polyvalent oligonu-cleotide gold nanoparticle conjugates as delivery vehicles for platinum(IV) warheads.J Am Chem Soc 131(41):14652–14653.

38. Takemoto H, et al. (2013) Acidic pH-responsive siRNA conjugate for reversible carrierstability and accelerated endosomal escape with reduced IFNα-associated immuneresponse. Angew Chem Int Ed Engl 52(24):6218–6221.

39. Dhar S, Gu FX, Langer R, Farokhzad OC, Lippard SJ (2008) Targeted delivery of cis-platin to prostate cancer cells by aptamer functionalized Pt(IV) prodrug-PLGA-PEGnanoparticles. Proc Natl Acad Sci USA 105(45):17356–17361.

40. Zheng YR, Suntharalingam K, Johnstone TC, Lippard SJ (2015) Encapsulation of Pt(IV)prodrugs within a Pt(II) cage for drug delivery. Chem Sci (Camb) 6(2):1189–1193.

41. Friedrich J, Seidel C, Ebner R, Kunz-Schughart LA (2009) Spheroid-based drug screen:Considerations and practical approach. Nat Protoc 4(3):309–324.

42. Sun Q, et al. (2014) Integration of nanoassembly functions for an effective deliverycascade for cancer drugs. Adv Mater 26(45):7615–7621.

43. Bibby MC (2004) Orthotopic models of cancer for preclinical drug evaluation: Ad-vantages and disadvantages. Eur J Cancer 40(6):852–857.

44. Tang L, Fan TM, Borst LB, Cheng J (2012) Synthesis and biological response of size-specific, monodisperse drug-silica nanoconjugates. ACS Nano 6(5):3954–3966.

45. Ruan S, et al. (2015) Matrix metalloproteinase-sensitive size-shrinkable nanoparticlesfor deep tumor penetration and pH triggered doxorubicin release. Biomaterials 60:100–110.

46. Tong R, Hemmati HD, Langer R, Kohane DS (2012) Photoswitchable nanoparticles fortriggered tissue penetration and drug delivery. J Am Chem Soc 134(21):8848–8855.

47. Blanco E, Shen H, Ferrari M (2015) Principles of nanoparticle design for overcomingbiological barriers to drug delivery. Nat Biotechnol 33(9):941–951.

48. Webb BA, Chimenti M, Jacobson MP, Barber DL (2011) Dysregulated pH: A perfectstorm for cancer progression. Nat Rev Cancer 11(9):671–677.

Li et al. PNAS | April 12, 2016 | vol. 113 | no. 15 | 4169

MED

ICALSC

IENCE

S

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

20, 2

020