An exosome-like programmable-bioactivating paclitaxel prodrug nanoplatform for enhanced breast cancer metastasis inhibition

Article abstract of DOI:10.1016/j.biomaterials.2020.120224

Metastasis is closely associated with high breast cancer mortality. Although nanotechnology-based anti-metastatic treatments have developed rapidly, the anti-metastasis efficiency is still far from satisfactory, mainly due to the poor recognition of circulating tumor cells (CTCs) in blood. Herein, we developed an exosome-like sequential-bioactivating prodrug nanoplatform (EMPCs) to overcome the obstacle. Specifically, the reactive oxygen species (ROS)-responsive thioether-linked paclitaxel-linoleic acid conjugates (PTX-S-LA) and cucurbitacin B (CuB) are co-encapsulated into polymeric micelles, and the nanoparticles are further decorated with exosome membrane (EM). The resulting EMPCs could specifically capture and neutralize CTCs during blood circulation through the high-affinity interaction between cancer cell membrane and homotypic EM. Following cellular uptake, EMPCs first release CuB, remarkably blocking tumor metastasis via downregulation of the FAK/MMP signaling pathway. Moreover, CuB obviously elevates the intracellular oxidative level to induce a sequential bioactivation of ROS-responsive PTX-S-LA. In vitro and in vivo results demonstrate that EMPCs not only exhibit amplified prodrug bioactivation, prolonged blood circulation, selective targeting of homotypic tumor cells, and enhanced tumor penetration, but also suppress tumor metastasis through CTCs clearance and FAK/MMP signaling pathway regulation. This study proposes an integrated approach for mechanism-based inhibition of tumor metastasis and manifests a promising potential of programmable-bioactivating prodrug nanoplatform for cancer metastasis inhibition.

Full text of DOI:10.1016/j.biomaterials.2020.120224

Journal Pre-proof
An exosome-like programmable-bioactivating paclitaxel prodrug nanoplatform for
enhanced breast cancer metastasis inhibition
Kaiyuan Wang, Hao Ye, Xuanbo Zhang, Xia Wang, Bin Yang, Cong Luo, Zhiqiang
Zhao, Jian Zhao, Qi Lu, Haotian Zhang, Qiming Kan, Yongjun Wang, Zhonggui He,
Jin Sun
PII:
S0142-9612(20)30470-1
DOI:
https://doi.org/10.1016/j.biomaterials.2020.120224
JBMT 120224
Reference:
To appear in: Biomaterials
Received Date: 23 December 2019
Revised Date: 29 June 2020
Accepted Date: 3 July 2020
Please cite this article as: Wang K, Ye H, Zhang X, Wang X, Yang B, Luo C, Zhao Z, Zhao J, Lu
Q, Zhang H, Kan Q, Wang Y, He Z, Sun J, An exosome-like programmable-bioactivating paclitaxel
prodrug nanoplatform for enhanced breast cancer metastasis inhibition, Biomaterials (2020), doi: https://
doi.org/10.1016/j.biomaterials.2020.120224.
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. Please note that,
during the production process, errors may be discovered which could affect the content, and all legal
disclaimers that apply to the journal pertain.
© 2020 Published by Elsevier Ltd.

An Exosome-like Programmable-Bioactivating
Paclitaxel Prodrug Nanoplatform for Enhanced
Breast Cancer Metastasis Inhibition
Kaiyuan Wang†,#, Hao Ye†,#, Xuanbo Zhang†, Xia Wang‡, Bin Yang§, Cong Luo†, Zhiqiang
Zhao†, Jian Zhao‡, Qi Lu†, Haotian Zhang○, Qiming Kan○, Yongjun Wang†, Zhonggui He†, Jin
†
Sun*,
†Department of Pharmaceutics, Wuya College of Innovation, Shenyang Pharmaceutical
University, Shenyang, Liaoning 110016, P. R. China
‡School of Pharmacy, Shenyang Pharmaceutical University, Shenyang, Liaoning 110016, P.
R. China
§State Key Laboratory of Drug Research & Center of Pharmaceutics, Shanghai Institute of
Materia Medica, Chinese Academy of Sciences, Shanghai 201203, P. R. China
○School of Life Science and Biopharmaceutics, Shenyang Pharmaceutical University,
Shenyang, Liaoning 110016, P. R. China
*Correspondence and requests for materials should be addressed to sunjin@syphu.edu.cn; Tel:
+86-024-23986325; Fax: +86-024-23986325

ABSTRACT: Metastasis is closely associated with high breast cancer mortality. Although
nanotechnology-based anti-metastatic treatments have developed rapidly, the anti-metastasis
efficiency is still far from satisfactory, mainly due to the poor recognition of circulating
tumor cells (CTCs) in blood. Herein, we developed an exosome-like sequential-bioactivating
prodrug nanoplatform (EMPCs) to overcome the obstacle. Specifically, the reactive oxygen
species (ROS)-responsive thioether-linked paclitaxel-linoleic acid conjugates (PTX-S-LA)
and cucurbitacin B (CuB) are co-encapsulated into polymeric micelles, and the nanoparticles
are further decorated with exosome membrane (EM). The resulting EMPCs could specifically
capture and neutralize CTCs during blood circulation through the high-affinity interaction
between cancer cell membrane and homotypic EM. Following cellular uptake, EMPCs first
release CuB, remarkably blocking tumor metastasis via downregulation of the FAK/MMP
signaling pathway. Moreover, CuB obviously elevates the intracellular oxidative level to
induce a sequential bioactivation of ROS-responsive PTX-S-LA. In vitro and in vivo results
demonstrate that EMPCs not only exhibit amplified prodrug bioactivation, prolonged blood
circulation, selective targeting of homotypic tumor cells, and enhanced tumor penetration, but
also suppress tumor metastasis through CTCs clearance and FAK/MMP signaling pathway
regulation. This study proposes an integrated approach for mechanism-based inhibition of
tumor metastasis and manifests a promising potential of programmable-bioactivating prodrug
nanoplatform for cancer metastasis inhibition.
KEYWORDS: exosome membrane; cucurbitacin B; paclitaxel; breast cancer metastasis;
programmable bioactivation

1. Introduction
Metastasis is an overwhelming challenge for tumor treatment, accounting for 90 % of
tumor-caused deaths.[1] Although chemotherapy is a criterion of cancer therapy,
small-molecule chemotherapeutic agents exhibit a marginal effect on metastasis inhibition
due to the negligible efficiency in eliminating circulating tumor cells (CTCs).[2, 3] CTCs have
been proven to aggregate in blood flow to prolong the survival time and construct metastatic
lesions.[4, 5] The homologous aggregation is mediated by high-affinity interaction between
binding proteins overexpressed on the membrane of CTCs.[6, 7]
Exosomes are 40-180 nm sized extracellular vesicles (EVs) extensively secreted by
different kinds of cells, including cancer cells, immune cells, stem cells, and endothelial cells.
They play a critical role in long- and short-distance cell-cell communication.[8, 9] Various
specific surface adhesion molecules are found in the membrane of cancer-derived exosomes,
such as CD44, which mediates the homotypic targeting[10] and integrin-associated protein
CD47 to escape from macrophage phagocytosis.[11] To the best of our knowledge,
exosome-biomimetic nanoparticles have not been reported to eliminate CTCs and for
inhibition of tumor metastasis. The strategy of decoration with exosome membrane (EM)
could assist nanoparticles in evading macrophage phagocytosis and recognizing both CTCs
and primary tumor.
Due to the overproduction of reactive oxygen species (ROS) in cancer cells,
ROS-responsive prodrug strategies aiming to realize tumor-specific bioactivation have
developed rapidly in recent years.[12, 13] However, the insufficient intrinsic ROS levels within
heterogeneous cancer cells hinder the rapid and effective release of parent drug. Therefore, a
variety of ROS amplification strategies, including synergistic chemo-photodynamic therapy,
Fenton reaction, and glucose oxidase (GOD) co-delivery, have been applied to bioactivate

ROS-responsive prodrug.[14-17] In this study, we hypothesize that the sequential-bioactivating
prodrug nanoparticles could facilitate ROS-triggered prodrug activation through cucurbitacin
B (CuB)-mediated ROS cascade amplification.
CuB is a kind of tetracyclic triterpenoid molecule, extracted from the plant of the
Cucurbitaceae family. CuB greatly inhibits the adhesion, migration, and invasion of breast
cancer through downregulation of the FAK/MMP signaling pathway,[18, 19] which is closely
associated with extracellular matrix degradation by matrix metalloproteinases (MMPs) to
evade the primary tumor and to form the focal adhesion. In addition, CuB mediates rapid and
massive ROS generation in various cancer cell lines, contributing to its anticancer and
anti-metastasis activity.[20]
To test our hypothesis, ROS-sensitive thioether-linked paclitaxel-linoleic acid prodrug
(PTX-S-LA)
and
CuB
were
co-encapsulated
into
polyethylene
glycol-block-poly(ε-caprolactone) (PEG-PCL) polymeric micelles, which were further
decorated with EM to fabricate an exosome-like sequential-bioactivating prodrug
nanoplatform (EMPCs). The EMPCs not only effectively target primary tumor but also
specifically capture CTCs leading to remarkable metastasis suppression. Following cellular
uptake, the nanoparticles would release CuB firstly, significantly blocking the adhesion,
migration, and invasion of cancer cells and inhibiting tumor metastasis via FAK/MMP
signaling pathway regulation. Furthermore, the released CuB increases ROS levels in tumor
cells, contributing to the effective PTX release from PTX-S-LA (Figure 1). Integrating
homotypic EM-mediated CTCs elimination and sequential prodrug bioactivation into a
unique biomimetic nanoplatform opens up the opportunity to breast cancer anti-metastasis
treatment.

Figure 1. Schematic representation of the exosome-like sequential-bioactivating paclitaxel
prodrug nanoplatform with CTCs clearance, CuB-mediated metastasis suppression, ROS
enhancement, and cascade amplified PTX chemotherapy.
2. Materials and methods
2.1. Materials.
PEG5k-PCL15k was obtained from Jinan Daigang Biomaterial Co., Ltd, China.
Cucurbitacin B was provided by Chengdu Biopurify Phytochemicals Ltd, China. Paclitaxel

was purchased from Nanjing Jingzhu Reagent Co., Ltd, China. Anti-TSG101 antibody,
Anti-CD9 antibody and Anti-CD81 antibody were purchased from Beijing Solarbio Science
& Technology Co., Ltd, China. Linoleic acid, glycol, 1-hydroxybenzotriazole monohydrate
(HOBt), N - (3 - dimethylaminopropyl) - N’ – ethylcarbodii - mide hydrochloride (EDCI)
were bought from Aladdin Industrial Co., Ltd, China . 3 - ( 4 , 5 - Dimethythiazol - 2 - yl ) -
2 , 5 - diphenyltetrazolium bromide ( MTT ) , 2,7-dichlorodihydrofluorescein diacetate
(DCFH-DA), DiI and DiR were purchased from Dalian Meilun Biotech Co., Ltd, China.
Anti-CD44 antibody and Anti-CD47 antibody were bought from Biosynthesis Biotechnology
Inc. (Beijing, China). Cell culture dishes and plates were obtained from Wuxi NEST
Biotechnology Co., Ltd, China. All solvents and reagents applied in the study were analytical
grade. Antibodies of FAK/FAK (pTyr-397) were purchased from ABclonal Biotechnology
Co.,Ltd, China. Antibodies of MMP-2 and MMP-9 were obtained from Abcam Inc., USA.
Exosome concentration kits (for cell culture media) was purchased from Liaoning Rengen
Biosciences Co., Ltd, China. Scientz-IID Ultrasonic Homogenizer was purchased from
Ningbo Scientz Biotechnology Co., Ltd, China.
2.2. Synthesis of PTX-S-LA.
Linoleic acid at a dosage of 2.1 g, 0.43 g of p-toluenesulfonic acid, and 38 ml of
ethylene glycol were heated to 115 under nitrogen for 2 h. The product was extracted using
methylbenzene, washed with NaHCO3 solution, and purified by column chromatography. The
yield of linoleic acid ethyl ester was 64.9 %. Then, 0.22 g of thiodiglycolic anhydride, 0.32 g
of EDCI and 0.25 g of HOBt were dissolved with 10 mL of dichloromethane in an ice bath.
The linoleic acid ethyl ester was added into the mixture for 35 min of reaction. Next, the
temperature was increased to 25
and reacted for 12 h. The solution was dried under
vacuum. Then, hydrochloric acid was added, and dichloromethane was used to extract
product. The dichloromethane layer was washed with NaCl solution. Sodium sulfate was

added to dry the product, and the product was purified with column chromatography. The
yield of 2′-O-(2-oxo-2-(2-((Z)-linoleoyloxy) ethoxy) ethylsulfenyl) acetic acid was 54.7 %.
Then, 1.4 g of 2′-O-(2-oxo-2-(2-((Z)-linoleoyloxy) ethoxy) ethylsulfenyl) acetic acid, 0.5 g of
EDCI, 0.35 g of HOBt, and 2.2 g of PTX were dissolved into dichloromethane in an ice bath
for 3 h and transferred to 25
for another 48-h reaction. Eventually, the PTX-S-LA was
gained through preparative liquid chromatography, and the yield was 46.1 %.
2.3. Preparation of PCNPs and EMPCs.
Encapsulation of PTX-S-LA and CuB in PEG-PCL NPs (PCNPs) was conducted by
emulsion solvent evaporation approach, where P denotes PTX-S-LA and C denotes CuB.
Briefly, 80 mg of PEG-PCL, 4 mg of PTX-S-LA and 0.6 mg of CuB were solubilized in 0.5
ml of acetone. The solution was dropped into 2 ml deionized water with stirring. Acetone was
then evaporated under vacuum and large particles were removed using a 220 nm aperture
filter. To collect the tumor-derived exosomes, MDA-MB-231 cells were cultured in 37 for
2 or 3 days until they grew to 80 % confluence. Next, the culture medium was replaced with
serum-free medium for 2 days. Then, the medium was collected and centrifuged at 300 g for
10 min to discard cells and further centrifuged at 15,000 g for 30 min to remove cell debris.
Following ultracentrifugation at 100,000 g for 120 min, the exosome pellets were acquired.
Subsequently, exosomes were resuspended in hypotonic buffer containing protease inhibitor
cocktail and stored at 4 for 12 h. The mixture was ultracentrifuged at 100,000 g for 5 h to
obtain EM. To prepare exosome-biomimetic EMPCs, the mixture of PCNPs and EM was
repeatedly co-extruded by a 220 nm polycarbonate porous membrane on an extruder (Antos
Nanotechnology Co., Ltd.). The same treatments were applied to fabricate EMPs or EMCs,
loaded with PTX-S-LA or CuB, respectively.
2.4. Characterization of PCNPs and EMPCs.
The size distribution, polydispersity index, and zeta potential of formulations were

detected by dynamic light scattering (DLS, Malvern, U.K.). The morphology of exosomes
and different nanoparticles were measured by transmission electron microscope (JEM-2100,
JEOL Ltd., Japan). The concentration of PTX-S-LA and CuB loaded in nanoparticles was
monitored with high performance liquid chromatography (HPLC). The colloidal stability of
EMPCs was investigated in PBS and cell culture media with or without 10 % FBS for 24 h at
37 . The particle size was detected by DLS at certain intervals. Western blotting for protein
moiety identification of EM, EMPs, EMCs, and EMPCs was performed as previously
reported. In brief, various molecule weights of proteins were separated with SDS-PAGE
electrophoresis. Next, the Coomassie brilliant blue method was applied for gel imaging
analysis. In western blotting assay, the proteins were transferred to PVDF membranes; then,
the membrane was blocked, and primary antibodies of anti-CD44 and anti-CD47 were added.
Following secondary antibodies incubation, ECL Western Blotting Substrate was added for
CCD imaging. TSG101, CD9, and CD81 were regarded as markers of exosomes and were
also detected with western blotting assay to confirm the success of exosome isolation. The
co-localization of the hydrophobic core and EM was applied to investigate the integrity of
EM-coated nanoparticles. Coumarin-6 (C-6) was encapsulated in the core of nanoparticles,
and DiI was used to stain the EM. The dual-labeled nanoparticles were incubated with
MDA-MB-231 cells for 2 h, and the nucleus was stained with Hoechst 33342. Cells were
washed using PBS and fixed using 4 % paraformaldehyde before confocal laser scanning
microscopy (CLSM, C2SI, Nikon, Japan) imaging.
2.5. ROS-Responsiveness of PTX-S-LA and In Vitro Drug Release.
To test the ROS sensitivity, hydrogen peroxide (H2O2) was applied as an ROS simulant.
PTX-S-LA dissolved in PBS (pH 7.4) with 30 % ethanol, and 0, 2, 5, and 10 mM of H2O2
was shaken (120 rpm) at 37 . At certain intervals, a 100 μL sample was collected for
analysis with HPLC. In vitro release study of PTX/CuB from EMPCs was conducted with a

dialysis bag (MWCO 14,000) under shaking (120 rpm) at 37
and subdivided into four
groups: (i) 0 mM of H2O2, (ii) 2 mM of H2O2, (iii) 5 mM of H2O2, and (iv) 10 mM of H2O2.
Typically, EMPCs were dialyzed against 30 ml of release media. At certain intervals, 100 μL
of solution outside the dialysis bag was collected for HPLC analysis. The release study was
carried out in triplicate, and results were presented as mean ± SD.
2.6. Cell Culture.
MDA-MB-231, MCF-7 and RAW264.7 cells were cultured in high-glucose Dulbecco’s
modified Eagle’s medium (DMEM) containing 12 % FBS and 1 % penicillin/streptomycin
(PS). NIH 3T3 cells were cultured in DME/F12 containing 10 % FBS and 1 % PS. Cells were
maintained under conditions of 37 and 5 % CO2.
2.7. Cytotoxicity Assay.
MTT assay was applied to evaluate the antiproliferative activity of Taxol, EMPs, EMCs,
and EMPCs against MDA-MB-231 cells. Briefly, cells (1,800 cells/well) were seeded in
96-well plates and incubated overnight. Cells were exposed to increasing concentrations of
drugs in 200 μL of fresh culture media and incubated for 48 h and 72 h. Then, MTT solution
was added to the plate and incubated at 37 for 4 h. Finally, the MTT solution was removed,
100 μL of dimethyl sulfoxide (DMSO) was added, and the absorbance was detected with a
microplate spectrophotometer at 570 nm.
2.8. Cellular ROS Assessment and Intracellular Drug Release.
Intracellular ROS production was detected with DCFH-DA staining by inverted
fluorescence microscope and flow cytometer. For inverted fluorescence microscope imaging,
MDA-MB-231 cells (60,000 cells/well) were cultivated in 24-well plates for 24 h. EMPs,
EMCs, or EMPCs were added to cells for 2 h. Subsequently, cells were stained with
DCFH-DA for 0.5 h and washed with PBS 3 times. The DCF fluorescence was detected with
inverted fluorescence microscope. To further quantify CuB-induced ROS generation,

MDA-MB-231 cells (150,000 cells/well) plated onto 12-well plates were treated with EMPs,
EMCs, or EMPCs for 2, 6, or 12 h. Subsequently, cells were stained with DCFH-DA for 0.5 h
and washed 3 times with PBS. Finally, cells were gathered and resuspended in PBS to
monitor the fluorescence signal with a FACS Calibur flow cytometer. To compare the free
PTX released from EMPs and EMPCs (PTX equivalent concentration: 500 ng/mL) after
incubation with MDA-MB-231 cells for 6 and 24 h, cells together with drug-containing
culture media were gathered, and the cells were broken with sonication. After centrifugation,
the concentrations of PTX in samples were detected with an ultra performance liquid
chromatography-tandem mass spectrometer (UPLC-MS-MS, Waters Co., Ltd, USA). Free
PTX operated in the same condition was set as control. The ratio of PTX released from
prodrug nanoparticles and PTX detected in control group was PTX release rate.
2.9. In Vitro Targeting of MDA-MB-231 Cells.
Encapsulation of C-6 in PEG-PCL NPs (C-6 NPs) was conducted by emulsion solvent
evaporation approach. Briefly, 40 mg of PEG-PCL and 2 mg of C-6 were solubilized in 0.5
ml of acetone. The solution was dropped into 2 ml deionized water with stirring. Acetone was
then evaporated under vacuum. To collect the red blood cells, whole blood withdrawn from
the mice was centrifuged at 800 g for 5 min to remove plasma and buffy coat. The resulting
red blood cells were washed 3 times in ice cold 1× PBS and suspended in hypotonic medium
of 0.25× PBS in an ice bath for 30 min. The released hemoglobin was discarded by
centrifuging the solution at 12000 rpm for 5 min. Red blood cell membrane (RM) were
collected after washing 3 times by 1× PBS. To prepare C-6 labeled red blood cell membrane
coated PEG-PCL nanoparticles (RM@C-6 NPs) or C-6 labeled EM-coated nanoparticles
(EM@C-6 NPs), the mixture of C-6 NPs, RM or C-6 NPs, EM was repeatedly co-extruded
by a 220 nm polycarbonate porous membrane on an extruder (Antos Nanotechnology Co.,
Ltd.).

MDA-MB-231, MCF-7, and NIH 3T3 cells were seeded into 12-well plates. After 24 h
of incubation, C-6 NPs, RM@C-6 NPs, or EM@C-6 NPs were added to the wells and
incubated for 2 h. Hoechst 33342 was added to stain the nucleus. Subsequently, cells were
washed using PBS and fixed using 4 % paraformaldehyde before CLSM imaging. To further
quantify the homotypic cancer cell targeting efficiency of EM-coated nanoparticles,
MDA-MB-231, MCF-7, and NIH 3T3 cells seeded into 12-well plates were exposed to C-6
NPs, RM@C-6 NPs or EM@C-6 NPs for 2 h. Finally, cells were gathered and resuspended in
PBS to monitor the fluorescence signal using a FACS Calibur flow cytometer.
2.10. Cellular Uptake of Macrophage RAW264.7 Cells.
The macrophage RAW264.7 cells were utilized to assess the macrophage phagocytosis
evasion ability of nanoparticles. The cells were treated with C-6 NPs, RM@C-6 NPs, and
EM@C-6 NPs for 2 h. Before CLSM imaging, cells were washed using PBS and fixed using
4 % paraformaldehyde. The quantification of intracellular fluorescence was conducted using
a FACS Calibur flow cytometer.
2.11. Tumor Spheroid Penetration and Wound-Healing Assay.
The comparison of penetration ability for PEG-PCL nanoparticles and EM-coated
nanoparticles was evaluated in three-dimensional MDA-MB-231 spheroids by CLSM. Briefly,
0.5×104/well cells suspended in DMEM with 2.5 % Matrigel were seeded into
ultra-low-attachment 96-well round-bottom plates and centrifuged at 1,000 rcf for 10 min.
The cells were cultivated under 37
and 5 % CO2 conditions for 4 days to form compact
spheroids. Next, the C-6 solution, C-6 NPs, and EM@C-6 NPs (33 ng ml-1 for C-6) were
added to tumor spheroids transferred in the confocal dishes and incubated at 37
for 8 h.
The tumor spheroids were rinsed with ice-cold PBS twice before observation with CLSM.
The inhibitory effect of the formulations on migration MDA-MB-231 cells was measured by
the wound-healing assay. The 105/well cells were seeded into 12-well plates and allowed to

grow to 90–95 % confluency. The confluent monolayer was scraped with a sterile 200 μL
pipette tip to form a cell-free zone. Then, cells were incubated with EMPs, EMCs, and
EMPCs (30 nM CuB and 120 nM PTX-S-LA) for the wound-healing assay.
2.12. Western Blotting.
The MDA-MB-231 cells cultured in 6-well plates were incubated with EMPs, EMCs,
EMPCs, and blank culture medium as control for 24 h. Radio Immunoprecipitation Assay
(RIPA)-lysis buffer was used to acquire cell lysates. The proteins were fractionated with
SDS-PAGE electrophoresis, transferred to PVDF and incubated with primary antibodies:
p-FAK, FAK, MMP-9, MMP-2, and GAPDH at 4
overnight. Following secondary
antibody incubation, ECL Western Blotting Substrate was added to visualize the protein
bands.
2.13. In Vitro Capture of CTCs in PBS.
The magnetic beads (MB, MACS, 50 nm) were mixed with EM in PBS. The mixture
was repeatedly extruded by a 220 nm polycarbonate porous membrane on an extruder (Antos
Nanotechnology Co., Ltd.) to prepare EM-coated magnetic beads (EM@MBs).
MDA-MB-231 cells (ie., about 50, 100, 200, 300, and 400 cells) were suspended in 1 ml of
PBS and incubated with EM@MBs or MBs for 1 h at 37 under 30 rpm. Then, cells were
separated by an external magnet and cell numbers were determined by cell count plate
(Watson, 177-112C). The cell capture efficiency was defined as the ratio of the captured cells
against the total cells.
2.14. Animal Studies.
All experiments were performed in accordance with the guide approved by the
Institutional Animal Ethical Care Committee (IAEC) of Shenyang Pharmaceutical University.
2.15. Pharmacokinetics.
The pharmacokinetic study was conducted with Sprague-Dawley rats (220–270 g, n = 3),

which were intravenously administrated with Taxol, PCNPs, EMPCs (4 mg kg-1 for PTX) or
CuB solution, PCNPs, EMPCs (2 mg kg-1 for CuB). Blood samples were collected at timed
intervals and centrifuged to get plasma. The concentration of PTX-S-LA, PTX and CuB was
monitored with UPLC-MS-MS (Waters Co., Ltd, USA) after protein precipitation.
2.16. In Vivo Clearance of CTCs.
Female nude mice were intravenously administrated with saline, PCNPs, and EMPCs
(equivalent to 6 mg kg-1 PTX and 0.98 mg kg-1 CuB). A half hour later, 4×105 MDA-MB-231
cells were intravenously injected into the tail vein. At 15 days post-administration, the CTCs
in blood were separated through ISET technology with a polycarbonate membrane
(Cyclotron, 8 μm) filtration. The number of CTCs was determined by cell count plate
(Watson, 177-112C). The elimination efficiency was calculated as the ratio of reducing CTCs
number in PCNPs or EMPCs group against remaining CTCs number in saline group. In
addition, lungs were excised and stained in Bouin’s solution, the visible metastatic nodules
was quantified. Hematoxylin and eosin (H&E) staining was also applied to detect the lung
metastasis.
2.17. Study of MDA-MB-231 Xenograft Tumor Model.
To establish the MDA-MB-231 back tumor model, MDA-MB-231 cells (1×107 cells)
were subcutaneously implanted into the right back side of female nude mice. For the
biodistribution experiment, the mice bearing tumor cells were divided into two groups. When
the tumor volume reached approximately 300 mm3, the DiR-labeled PEG-PCL nanoparticles
(DiR NPs) and DiR-labeled EM-coated nanoparticles (EM@DiR NPs) at 2 mg DiR/kg were
injected intravenously. Fluorescence imaging was performed at different time intervals (2 h, 8
h, 12 h, 24 h) post intravenous injection using in vivo imaging system (IVIS). At 24 h post
injection, the mice were sacrificed, and the fluorescence imaging in the major organs (heart,
liver, spleens, lungs, kidneys) and tumors were analyzed by IVIS. For in vivo antitumor study,

the animals (n = 5) received tail vein injections of the various drug formulations (saline,
Taxol, EMCs, EMPs, PCNPs, and EMPCs, equivalent to 6 mg kg-1 PTX and 0.98 mg kg-1
CuB) every 3 days for 30 days once the tumor size reached an average volume of 100 mm3.
During the treatment, the tumor size was calculated using the formula V = 1/2ab2 (a: long
axis, b: short axis), and body weight was monitored. Mice were sacrificed after 30 treatment
days, and their lungs were collected and stained in Bouin’s solution. The heart, liver, spleen,
lungs, kidneys and tumor of one mouse in every group was excised and sliced for H&E
staining. The weight of the tumor was recorded.
2.18. Study of MDA-MB-231 Orthotopic Tumor Model.
To establish the MDA-MB-231 orthotopic tumor model, 1×107 MDA-MB-231-luc cells
were injected into the mammary fat pads of female nude mice. The orthotopic biodistribution
study and orthotopic in vivo antitumor experimental profiles were similar to that of the
subcutaneous breast cancer xenografts. After 30 treatment days in the orthotopic antitumor
experiment, mice were sacrificed, and fresh lungs were collected and incubated in 15 mg ml-1
D-luciferin for 10 min to assess the anti-metastasis effect. The ex vivo bioluminescent
imaging of lungs was performed with IVIS. To further detect the prodrug bioactivation in
tumor tissue, MDA-MB-231 orthotopic tumor-bearing mice were divided into two groups.
EMPs and EMPCs (equivalent to 6 mg kg-1 PTX and 0.98 mg kg-1 CuB) were administered
intravenously once the tumor volume reached approximately 300 mm3. Mice were sacrificed
at 12 h post-administration, and tumors were collected. The concentration of released PTX
was monitored with UPLC-MS-MS (Waters Co., Ltd., USA) after homogenization and
protein precipitation.
3. Results and Discussion
3.1. Preparation and Characterization of the Nanosystem.

Firstly, the ROS-sensitive PTX prodrug was prepared by conjugating linoleic acid to
PTX by a single thioether linkage (Scheme S1). The chemical structure was verified with MS
1
and H NMR (Figure S1). The PTX-S-LA and CuB were co-loaded into PEG-PCL
nanoparticles using the emulsion solvent evaporation approach to fabricate PCNPs,[21-23]
where P denotes PTX-S-LA and C denotes CuB. The PCNPs were then incubated with EM to
acquire the EMPCs. The same treatments were applied to fabricate EMPs or EMCs, loaded
with PTX-S-LA or CuB, respectively. Figure 2 displayed the hydrodynamic diameter and
morphology of PCNPs, which had a regular spherical shape with an average size of nearly 80
nm. The drug-loading capacities of PTX-S-LA and CuB in PEG-PCL nanoparticles were 9
wt% and 5 wt%, and the encapsulation efficiency were 87.6 % and 69.2 %, respectively.
The gradient centrifugation method was used to extract exosomes. The saucer-like
structure of exosomes by transmission electron microscopy (TEM) imaging together with
western blotting analysis for characteristic exosome markers TSG101, CD9, and CD81,
indicated the successful isolation of exosomes.[24] EM was prepared via the hypotonic
treatment of extracted exosomes. The PEG-PCL nanoparticles were further incubated with
EM to prepare EMPCs.[25] The decreased zeta potential, increased particle size (~100 nm),
and TEM images demonstrated the decoration of EM on the surface of PEG-PCL
nanoparticles. Figure S2 showed that EMPCs had desirable colloidal stability after incubation
in phosphate-buffered saline (PBS) and cell culture media with or without 10 % fetal bovine
serum (FBS) for 24 h.
To examine the existence of membrane proteins on the shell of exosome-biomimetic
nanoparticles, sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and
western blotting assay were performed as previously reported.[26] SDS-PAGE protein
examination of the membrane protein markers showed good preservation of the distinctive
proteins inherited from EM in the protein profile of exosome-biomimetic nanoparticles.

CD44 and CD47 are of critical importance for tumor metastasis. CD44 is a surface adhesion
molecule on MDA-MB-231 cells, which plays a crucial role in the adherence of CTCs to
metastasis lesions. CD47 could protect tumor cells from macrophage phagocytosis, which is
associated with tumor aggression. As shown in Figure 2D, they co-existed on the cancer cell
membrane and EM and were conserved well in EMPCs to mediate homotypic cancer cell
targeting.
Then, EM was labeled with DiI (red), and the hydrophobic core was labeled with
coumarin-6 (C-6, green) to examine the complete coating of PEG-PCL nanoparticles with
EM.[27] The dual-labeled nanoparticles were added to MDA-MB-231 cells for 2 h and
observed with confocal laser scanning microscopy (CLSM). As shown in Figure 2E, the
fluorescence of DiI was co-localized with that of C-6, indicating that exosome-biomimetic
nanoparticles could retain the structural integrity after cellular uptake for a period of time.

Figure 2. (A) TEM image and detection of TSG101, CD9, and CD81 proteins for
MDA-MB-231 derived exosomes. Scale bar = 100 nm. (B) SDS-PAGE protein analysis of
EM, EMPs, EMCs and EMPCs. (C) TEM images of PCNPs, EMPs, EMCs, and EMPCs.
Scale bar = 100 nm. (D) Western blotting analysis of whole MDA-MB-231 cells,
MDA-MB-231 cancer cell membrane (CCM), EM and EMPCs, for CD44 and CD47 proteins.
(E) Intracellular co-localization of EM shell (red) and PEG-PCL core (green). Scale bar = 10
μm. (F) Hydrodynamic diameter and zeta potential of PCNPs, EMPs, EMCs and EMPCs.

3.2. ROS-Responsiveness of PTX-S-LA and In Vitro Drug Release.
To investigate the ROS-responsiveness of PTX-S-LA, H2O2 was used as an ROS
simulant.[28] As shown in Figure 3A, only 10 % PTX was released from PTX-S-LA after 12 h
of incubation without H2O2. In comparison, PTX-S-LA released nearly 100 % of PTX in the
presence of 10 mM of H2O2 within 6 h, suggesting an excellent ROS-sensitive bioactivation
capability of PTX-S-LA. Furthermore, EMPCs were incubated with different concentrations
of H2O2 to explore the sequential-bioactivation process.[29] Approximately 94 % of CuB was
released faster than PTX-S-LA (4 %) after 12 h of incubation in the absence of H2O2. By
contrast, after the addition of H2O2, the PTX release percentage of PTX-S-LA increased in an
H2O2 concentration-dependent manner and dramatically reached 95 % in 12 h after
incubation with 10 mM of H2O2 (Figure 3B).
3.3. Cytotoxicity Assay.
We subsequently detected the cytotoxicity in MDA-MB-231 cells with MTT assay. The
Chou-Talalay method is now the most widely used method in the literature for analyzing
combinations. The combination index (CI) was calculated according to equation below: CI
=(D)1 /(Dx)1 + (D)2 /(Dx)2, in which the denominator (Dx)1 and (Dx)2 are the concentration
of Drug 1 and Drug 2 alone that inhibit the growth of cells by x %. Similarly, the numerator
(D)1 and (D)2 represent the respective dosage of Drug 1 and Drug 2 at which their drug
combo inhibits the cell growth by the same x %. CI < 1, CI = 1, and CI > 1 represents
synergism, additive effect, and antagonism, respectively. The growth inhibition effect
(fraction affected, Fa) was calculated according to equation below: Fa=1-(%growth/100). A
plot of CI values at different levels of growth inhibition effect (fraction affected, Fa) can be
determined by computer simulation. Such a combination index plot is also called the Fa–CI

plot. In Figure S3, the plots showed that molar ratio 4:1 of PTX-S-LA to CuB has the best
synergism at the most effect level (Fa). Figure 3C-D showed that the cytotoxicity of EMPs
was much lower than that of Taxol due to the delayed prodrug bioactivation. This was
resulted from the blocking of C2’-hydroxyl, essential to the antitumor efficacy of PTX. By
contrast, EMPCs displayed better cancer cell killing capability in comparison with Taxol. As
shown in Table S1, the half maximal inhibitory concentration (IC50) at the equivalent
concentrations of PTX were 4,206 nM and 1,365 nM for EMPs and EMPCs, in comparison
with 1,572 nM for Taxol at 48 h against MDA-MB-231 cells. Moreover, the CI values of
EMPCs against MDA-MB-231 cells were 0.37 and 0.31 for 48 h and 72 h, suggesting a
strong synergic effect and good sequential prodrug bioactivation.
3.4. Cellular ROS Assessment and Intracellular Drug Release.
To investigate the ROS generation, MDA-MB-231 cells incubated with different
formulations were stained by 2,7-dichlorodihydrofluorescein diacetate (DCFH-DA). The
ROS level exhibited no enhancement after incubation with EMPs, while CuB-loaded EMCs
and EMPCs considerably improved intracellular ROS levels (Figure 3E). To further quantify
the CuB-induced ROS production, the fluorescence intensity was detected with flow
cytometry after 2, 6, or 12 h of treatment. The EMCs and EMPCs remarkably improved the
intracellular ROS level in a time-dependent manner within 12 h (Figure 3F). Therefore, the
co-loaded CuB could enhance ROS levels in MDA-MB-231 cells, further increasing prodrug
bioactivation. Encouraged by the results of ROS detection, we further monitored the released
PTX from EMPCs and EMPs after 6 or 24 h incubation with MDA-MB-231 cells. As shown
in Figure 3G, the PTX release amount of EMPCs was much higher than that of EMPs,
agreeing well with the in vitro drug release result.

Figure 3. (A) Transformation of PTX-S-LA into PTX in the presence of H2O2. (B) In vitro
drug release profiles of CuB and PTX from EMPCs with various concentrations of H2O2.
Cytotoxicity of EMCs, EMPs, Taxol, and EMPCs against MDA-MB-231 cells for (C) 48 h

and (D) 72 h. (E) Fluorescence microscopy images and (F) flow cytometry analysis of
MDA-MB-231 cells stained with DCFH-DA after incubation with blank culture medium,
EMPs, EMCs, or EMPCs. Scale bar = 20 μm. (G) PTX release amount of EMPs and EMPCs
＊
in MDA-MB-231 cells after 6 h and 24 h incubation (＊＊P < 0.01, P < 0.05, n = 3).
3.5. In Vitro Targeting of MDA-MB-231 Cells.
We next investigated the cellular uptake of C-6 labeled PEG-PCL nanoparticles (C-6
NPs), C-6 labeled red blood cell membrane coated PEG-PCL nanoparticles (RM@C-6 NPs)
and C-6 labeled EM-coated nanoparticles (EM@C-6 NPs) in MDA-MB-231, MCF-7, and
NIH 3T3 cells to confirm the homotypic cancer cell targeting capability of
exosome-biomimetic nanoparticles. The decoration of PEG-PCL nanoparticles with red blood
cell membrane was conducted to verify the critical role of proteins on EM in promoting
internalization of MDA-MB-231 cells. As shown in Figure 4A, the C-6 fluorescence of C-6
NPs and RM@C-6 NPs was rather lower than that of EM@C-6 NPs, demonstrating that the
improved cellular uptake was attributed to proteins on EM. Meanwhile, MCF-7 and NIH 3T3
cells were set as the negative controls. No formulation exhibited improvement of green
fluorescence in MCF-7 and NIH 3T3 cells, further illustrating that the decoration of
nanoparticles with EM specifically enhanced internalization of the same-source cancer cells.
The quantification of targeting effect was then detected with flow cytometry. As shown in
Figure 4B and Figure S4, the fluorescence signal of EM@C-6 NPs was 3.5 times higher than
those of C-6 NPs and RM@C-6 NPs in MDA-MB-231 cells. By contrast, EM@C-6 NPs
showed no increase of fluorescence in MCF-7 and 3T3 cells in comparison with C-6 NPs and
RM@C-6 NPs. All results mentioned above indicated that exosome-biomimetic nanoparticles
had a selective affinity to the homotypic cancer cells.

3.6. Cellular Uptake of Macrophage RAW264.7 Cells.
Then, we verified the macrophage phagocytosis evasion ability of CD47 on EM. Mouse
macrophage RAW264.7 cells seeded into 12-well plates were incubated with C-6 NPs,
RM@C-6 NPs and EM@C-6 NPs. As shown in Figure 4C, the RM@C-6 NPs and EM@C-6
NPs exhibited a much lower internalization than C-6 NPs. For flow cytometry quantification,
the fluorescence intensity of C-6 NPs displayed a 3.1-fold increase in comparison with that of
RM@C-6 NPs and EM@C-6 NPs. The results confirmed that the CD47 on both RM and EM
could inhibit the macrophage phagocytosis.

Figure 4. (A) CLSM images and (B) flow cytometry analysis of MDA-MB-231, MCF-7,
NIH 3T3 cells incubated with C-6 NPs, RM@C-6 NPs, and EM@C-6 NPs. The nucleus was
stained with Hoechst (blue), and nanoparticles were labeled with C-6 (green). (C)
Intracellular uptake imaging and (D) intracellular fluorescence intensity detected with flow
cytometry of C-6 NPs, RM@C-6 NPs and EM@C-6 NPs incubated in macrophage
RAW264.7 cells for 2 h. Scale bar = 15 μm.
3.7. Tumor Spheroid Penetration and Wound-Healing Assay.
MDA-MB-231 cells spheroids were performed to assess the tumor spatial penetration
ability of exosome-biomimetic nanoparticles.[30] The spheroids were incubated with C-6
solution, C-6 NPs, and EM@C-6 NPs for 8 h. CLSM images exhibited that the fluorescence
signal of C-6 from EM@C-6 NPs was higher in the deep site of cancer cells spheroids than
that of C-6 solution and C-6 NPs (Figure 5A), demonstrating an in-depth tumor penetration
capability of exosome-biomimetic nanoparticles. Since the cancer cell migration was
associated with tumor metastasis, the migration inhibition effects of different formulations
(control, EMPs, EMCs, and EMPCs) were examined on MDA-MB-231 cells through wound
healing trials. After 24 h treatment, the wound gap had nearly disappeared in the control
group. Moreover, the cell migration capability was remarkably suppressed in EMCs- and
EMPCs-treated groups compared with the EMPs group (Figure 5B), demonstrating the good
anti-metastasis effect of CuB.
3.8. Western Blotting.
Western blotting assay was conducted to monitor the expression levels of key molecules
related to tumor metastasis. Previous studies have clarified the role of the FAK/MMP

signaling pathway in cancer migration and invasion, including extracellular matrix
degradation by matrix metalloproteinases (MMPs) to evade primary tumor and to form the
focal adhesion. To evaluate the FAK/MMP signaling pathway suppression caused by different
formulations, we detected the expression of activated FAK (phospho-FAK, p-FAK), FAK,
MMP-9, and MMP-2 in cell lysates. In agreement with the wound healing assay, EMCs and
EMPCs significantly suppressed the FAK/MMP signaling pathway compared with EMPs
(Figure 5C), further confirming the CuB-mediated anti-metastasis effect.
3.9. Pharmacokinetics.
We next investigated the pharmacokinetic profiles of Taxol, CuB solution, PCNPs, and
EMPCs in Sprague-Dawley rats. As presented in Figure 5D, Figure S5 and Table S2, PTX
and CuB were eliminated rapidly from blood circulation after intravenous administration of
Taxol or CuB solution. EMPCs and PCNPs showed extended blood circulation time in
comparison with Taxol or CuB solution. Surprisingly, EMPCs exhibited an approximately
2-fold increase in the area under the curve (AUC) of PTX-S-LA and CuB as well as longer
half-life compared with PCNPs, suggesting that the self-protection ability of CD47 on EM
significantly suppressed macrophage phagocytosis. Moreover, EMPCs decreased the
undesired leakage of PTX during blood circulation compared to PCNPs, demonstrating an
enhanced systemic safety of EM-coating nanoparticles.

Figure 5. (A) In vitro penetration of C-6 solution, C-6 NPs and EM@C-6 NPs incubated with
MDA-MB-231 tumor spheroids. Scale bar = 100 μm. (B) Typical images of wound healing
assessments of control, EMPs, EMCs, and EMPCs on MDA-MB-231 cells at 0, 12, and 24 h.
(C) The effect of EMPs, EMCs, and EMPCs on the expression levels of FAK/MMP signaling
pathway proteins. (D) Drug concentration in plasma as a function of time after a single
intravenous injection of Taxol, PCNPs, or EMPCs.
3.10. In Vitro Capture of CTCs in PBS.

Then, magnetic beads (MBs) and EM-coated magnetic beads (EM@MBs) were
prepared to investigate the in vitro CTCs capture ability of EM-biomimetic nanoparticles.
After 1 h co-incubation with MDA-MB-231 cells under 30 rpm, an external magnet was used
to separate the captured cells. It is worth noting that regression analysis showed a much
higher CTCs capture efficiency of EM@MBs (Figure 6A), which could be attributed to the
adhesion molecule CD44 on EM.
3.11. In Vivo Clearance of CTCs.
Encouraged by in vitro study, we examined the potential of EMPCs to eradicate CTCs in
the circulatory system. The experiments were carried out using nude mice with an inhibited
immune system. After intravenous administration with saline, PCNPs, and EMPCs, mice
were then intravenously injected with MDA-MB-231 cells as CTCs simulant.[31] At 15 days
post-administration, the CTCs in blood were separated through ISET technology[32] with a
polycarbonate membrane (Cyclotron, 8 μm) filtration. As shown in Figure 6B, the CTCs
elimination efficiency of EMPCs was 3.2 times higher than that of PCNPs, confirming the
excellent in vivo CTCs clearance ability of EMPCs. CTCs in blood could spread to various
major organs, particularly the lungs. As shown in Figure 6C-D, mice treated with saline had
remarkable micro-metastases in the lungs. The lung micro-metastases number of mice treated
with PCNPs showed a slight reduction. By contrast, there was negligible metastatic nodule
observed in the lung of EMPCs-treated mice, indicating that EMPCs could target the CTCs
during systemic circulation through adhesion molecule CD44 and further neutralize the CTCs
via sequential prodrug bioactivation.

Figure 6. (A) In vitro CTCs capture efficiency of EM@MBs and MBs in PBS. (B) In vivo
CTCs elimination efficiency of PCNPs and EMPCs (＊＊＊P < 0.001, n = 5). (C) Photos of
Bouin’s solution stained whole lungs and H&E staining of the lung slices collected from
saline, PCNPs and EMPCs groups. Red circles demonstrate the visible metastatic site. Yellow
arrows demonstrate the lung metastases. Scale bar = 100 μm. (D) Quantification of visible
metastatic nodules (＊＊＊P < 0.001, ＊＊P < 0.01, n = 5).
3.12. Study of MDA-MB-231 Xenograft Tumor Model.
To investigate the tumor-targeting capability of exosome-biomimetic nanoparticles, an
MDA-MB-231 xenograft tumor model was employed.[33] As shown in Figure S6, the

fluorescence intensity of DiR-labeled EM-coated nanoparticles (EM@DiR NPs) increased
over time from 2 h to 12 h, notably stronger than that of DiR-labeled PEG-PCL nanoparticles
(DiR NPs). Moreover, the ex vivo fluorescence of EM@DiR NPs in tumor tissue remained
high at 24 h post administration, a 3.6-fold increase in comparison with that of DiR NPs. The
enhanced tumor accumulation of exosome-biomimetic nanoparticles was due to homotypic
targeting capability and prolonged blood circulation time.
We further investigated the in vivo antitumor efficacy of EMPCs with MDA-MB-231
back tumor-bearing mice. As shown in Figure S7, the control group exhibited a quick
increase in tumor volume (~1,400 mm3 at day 30). Within 30 d, the volume of the tumor
increased to ~1,000 mm3 and ~800 mm3 in EMC and Taxol groups, indicating a moderate
tumor-inhibitory ability. The tumor volume reached ~520 mm3 and ~300 mm3 for the EMPs
and PCNPs groups after 30 d of treatment. In contrast, EMPCs indicated a complete tumor
growth inhibition (~130 mm3 at day 30), exhibited the strongest induction of tumor apoptosis
in H&E stained tumor slices (Figure S8), and highlighted the advantages of CuB-mediated
strengthened PTX chemotherapy and EM-related excellent tumor-targeting capability.
Bouin’s solution staining and H&E staining were applied to compare the potential of different
formulations to suppress tumor metastasis in vivo. As presented in Figure S7D, no evident
cancer metastasis was observed in the EMPCs group. The Taxol, EMPs, and PCNPs groups
showed serious tumor metastases in the lungs and liver, but there was a significant reduction
of metastatic nodules in the EMCs group, clarifying the EM-mediated CTCs targeting and
CuB-related metastasis inhibition.
3.13. Study of MDA-MB-231 Orthotopic Tumor Model.
We then compared the tumor-targeting ability of DiR NPs and EM@DiR NPs in
orthotopic MDA-MB-231 tumor-bearing mice. The EM@DiR NPs exhibited the highest

fluorescence signal at tumor site after 12 h administration, much stronger than that of DiR
NPs. Moreover, the ex vivo fluorescence intensity of EM@DiR NPs in tumor tissue showed a
3.1-fold increase compared with that of DiR NPs, implying an excellent targeting ability of
exosome-biomimetic nanoparticles. Encouraged by biodistribution results, the bioactivated
PTX release performance in tumor tissue was further monitored. As shown in Figure 7D, the
concentration of released PTX in the EMPCs group was 2.9 times higher than that of the
EMPs group, further confirming the in vivo CuB-mediated amplified prodrug bioactivation.
Figure 7. In vivo fluorescence imaging in orthotopic MDA-MB-231 tumor-bearing mice at 2,
8, 12, and 24 h following injection with DiR NPs and EM@DiR NPs. (B) Ex vivo
fluorescence imaging and (C) ROI analysis of major organs and tumors harvested at 24 h
post-administration. (D) PTX release amount of EMPs and EMPCs in tumors at 12 h after
injection (＊＊＊P < 0.001, n = 3).