Protein-functionalized nanoparticles derived from end-functional polymers and polymer prodrugs for crossing the blood-brain barrier

Article abstract of DOI:10.1016/j.ejpb.2019.06.004

Nanoparticles may provide a viable way for neuroprotective drugs to cross the blood–brain barrier (BBB), which limits the passage of most drugs from the peripheral circulation to the brain. Heterotelechelic polymer prodrugs comprising a neuroprotective model drug (adenosine) and a maleimide functionality were synthesized by the “drug-initiated” approach and subsequent nitroxide exchange reaction. Nanoparticles were obtained by nanoprecipitation and exhibited high colloidal stability with diameters in the 162–185 nm range and narrow size distributions. Nanoparticles were then covalently surface-conjugated to different proteins (albumin, α2-macroglobulin and fetuin A) to test their capability of enhancing BBB translocation. Their performances in terms of endothelial permeability and cellular uptake in an in vitro BBB model were compared to that of similar nanoparticles with surface-adsorbed proteins, functionalized or not with the drug. It was shown that bare NPs (i.e., NPs not surface-functionalized with proteins) without the drug exhibited significant permeability and cellular uptake, which were further enhanced by NP surface functionalization with α2-macroglobulin. However, the presence of the drug at the polymer chain-end prevented efficient passage of all types of NPs through the BBB model, likely due to a decrease in the hydrophobicity of the nanoparticle surface and alteration of the protein binding/coupling, respectively. These results established a new and facile synthetic approach for the surface-functionalization of polymer nanoparticles for brain delivery purposes.

Full text of DOI:10.1016/j.ejpb.2019.06.004

Accepted Manuscript
Protein-Functionalized Nanoparticles Derived from End-Functional Polymers
and Polymer Prodrugs for Crossing the Blood-Brain Barrier
Alysia Cox, Daniele Vinciguerra, Francesca Re, Roberta Dal Magro, Simona
Mura, Massimo Masserini, Patrick Couvreur, Julien Nicolas
PII:
S0939-6411(19)30319-4
https://doi.org/10.1016/j.ejpb.2019.06.004
EJPB 13072
DOI:
Reference:
European Journal of Pharmaceutics and Biophar-
maceutics
To appear in:
Received Date:
Revised Date:
Accepted Date:
18 March 2019
30 May 2019
6 June 2019
Please cite this article as: A. Cox, D. Vinciguerra, F. Re, R. Dal Magro, S. Mura, M. Masserini, P. Couvreur, J.
Nicolas, Protein-Functionalized Nanoparticles Derived from End-Functional Polymers and Polymer Prodrugs for
Crossing the Blood-Brain Barrier, European Journal of Pharmaceutics and Biopharmaceutics (2019), doi: https://
doi.org/10.1016/j.ejpb.2019.06.004
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. 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.

Protein-Functionalized Nanoparticles Derived from
End-Functional Polymers and Polymer Prodrugs for
Crossing the Blood-Brain Barrier
Alysia Cox,1,§ Daniele Vinciguerra,2,§ Francesca Re,1,* Roberta Dal Magro,1 Simona Mura,2
Massimo Masserini,1 Patrick Couvreur,2 Julien Nicolas2,*
1 School of Medicine and Surgery, Nanomedicine Center NANOMIB, University of Milano-
Bicocca, Via Raoul Follereau 3, 20854 Vedano al Lambro (MB), Italy.
2 Institut Galien Paris-Sud, UMR CNRS 8612, Univ Paris-Sud, Faculté de Pharmacie, 5 rue Jean-
Baptiste Clément, F-92296 Châtenay-Malabry cedex, France.
§ Equally contributing authors
*To whom correspondence should be addressed.
Francesca Re. Tel.: +39 2 64 48 83 11; Email: francesca.re1@unimib.it
Julien Nicolas. Tel.: +33 1 46 83 58 53; Email: julien.nicolas@u-psud.fr
1

Abstract
Nanoparticles may provide a viable way for neuroprotective drugs to cross the blood-brain barrier
(BBB), which limits the passage of most drugs from the peripheral circulation to the brain.
Heterotelechelic polymer prodrugs comprising a neuroprotective model drug (adenosine) and a
maleimide functionality were synthesized by the “drug-initiated” approach and subsequent
nitroxide exchange reaction. Nanoparticles were obtained by nanoprecipitation and exhibited high
colloidal stability with diameters in the 162–185 nm range and narrow size distributions.
Nanoparticles were then covalently surface-conjugated to different proteins (albumin, 2-
macroglobulin and fetuin A) to test their capability of enhancing BBB translocation. Their
performances in terms of endothelial permeability and cellular uptake in an in vitro BBB model
were compared to that of similar nanoparticles with surface-adsorbed proteins, functionalized or not
with the drug. It was shown that bare NPs (i.e., NPs not surface-functionalized with proteins)
without the drug exhibited significant permeability and cellular uptake, which were further
enhanced by NP surface functionalization with 2-macroglobulin. However, the presence of the
drug at the polymer chain-end prevented efficient passage of all types of NPs through the BBB
model, likely due to a decrease in the hydrophobicity of the nanoparticle surface and alteration of
the protein binding/coupling, respectively. These results established a new and facile synthetic
approach for the surface-functionalization of polymer nanoparticles for brain delivery purposes.
Keywords
Polymer, prodrug, nanoparticle, blood-brain barrier, proteins, drug delivery
2

1. Introduction
The blood-brain barrier (BBB) remains one of the major limiting obstacles preventing
pharmacological advancement in the treatment of neurological, neurodegenerative, and psychiatric
disorders [1]. Although nanoparticles (NP) offer a promising approach to cross this barrier, surface-
modification is necessary to increase drug delivery to the central nervous system (CNS). With this
in mind, synthetic and physiological peptides and proteins have extensively been used to coat or
functionalize NPs for BBB targeting [2-4]. Many techniques have been utilized to identify novel
synthetic peptides or previously untested physiological proteins, including peptide phage display
and computer-assisted design [5]. A wide array of these ligands has been successfully used to target
receptor-mediated endocytosis and encourage passage of NPs through the BBB [6], commonly
through transferrin receptor [7-9], lactoferrin receptor [10, 11], and low-density lipoprotein receptor
(LDLr) [12, 13]. In particular, a modified fragment of apolipoprotein E, named mApoE,
successfully enhanced brain delivery of liposomes designed for the treatment of Alzheimer’s
disease via transcytosis targeting LDLr [12, 13].
We recently analyzed the protein composition of NP corona after passage through an in vitro
BBB model and found that specific proteins were able to cross endothelial BBB cells in relatively
high abundance [14]. Among these proteins, we have selected three to test their capability to
improve brain targeting of polymeric NP: albumin, α2-macroglobulin (α2M) and Fetuin A (Fet A).
Albumin, the most abundant serum protein, crosses the BBB in vivo in healthy conditions [15], and
in higher quantities after traumatic brain injury [16], in Huntington disease [17], diabetes and
dementia [18]. α2M is a broad spectrum anti-protease that also crosses the healthy BBB [19] and is
a biomarker for multiple sclerosis [20]. It has anti-inflammatory effects which seem particularly
promising in treating neurological conditions that are coupled with increased local or generalized
inflammation [21]. Fet A, also known as α2-HS-glycoprotein, is predominantly found in fetal serum
where it prevents tissue calcification [22]. It can cross the BBB during ischemic injury [23].
3

Although albumin has previously been used to synthesize NPs that can cross the BBB [24], the use
of α2M or Fet A for functionalizing NPs to specifically target the brain has never been investigated.
Among the different families of materials suitable for the design of targeted drug delivery
systems, polymer NPs have attracted much attention because of their numerous benefits. Polymer
synthesis methods indeed offer great flexibility in terms of control, macromolecular architecture,
physico-chemical properties, and enable easy access to functional materials. For instance, different
polymer NPs have been engineered to allow surface modification for CNS drug delivery with
various ligands to enhance their permeability through the BBB [25-27]. NP coating by simple
adsorption of targeting peptides has been extensively investigated [28-30], but surface
functionalization via covalent linkage remains the method of choice as it prevents protein
desorption and loss of targeting, though it requires more elaborate synthetic strategies. Developing
polymer prodrug nanocarriers, where the drug is linked to the polymer, usually comes with
important benefits compared to traditional drug-loaded systems [31], including: (i) absence of
“burst release” (i.e., uncontrolled release of drugs post-administration); (ii) the possibility to obtain
higher drug loading and (iii) enhancement of the drug compatibility with the polymer matrix/core,
allowing a broader range of drugs to be transported.
Heterotelechelic polymers are gaining great interest in the biomedical field [32], as they
offer an opportunity to embed two different biologically active moieties positioned at both
extremities (e.g., drug, targeting ligands, imaging probes). In this context, we proposed the
synthesis of heterotelechelic polymers bearing a drug of interest in the -position and a reactive
group for further conjugation with BBB-crossing proteins in the -position.
As a proof of concept, adenosine (Ade) was chosen as a model drug for this drug delivery
system. Its neuroprotective characteristics have already been demonstrated in vivo after stroke when
delivered to BBB endothelial cells as a squalene-Ade molecular prodrug [33].
Here, we either adsorb or conjugate bovine serum albumin (BSA), α2M or Fet A at the
surface of heterobifunctional polymer prodrug NPs (Figure 1) and report on the proteins effect on
4

NP ability to cross the in vitro BBB, possibly enhancing Ade transport and therefore its
neuroprotective activity. Moreover, targeting is vital to avoid side reactions, given the widespread
expression of Ade receptors throughout the body [34, 35], and peripheral degradation [36]. Though
investigation of the pharmacological activity of the prodrug NPs is outside the scope of this article,
we investigate here the possibility to embed a relevant drug for the treatment of CNS-related
diseases and study its impact on the nanoparticle interaction/functionalization with proteins, and
their subsequent cellular uptake and permeability through a well-established and widely used in
vitro transwell BBB model [3].
Figure 1. General strategies for the preparation of (a) polymer nanoparticles and (b) polymer
prodrug nanoparticles surface-adsorbed or surface-functionalized with proteins for the crossing of
hCMEC/D3 human brain endothelial cells as a blood-brain barrier (BBB) model. BSA = albumin,
2M = 2-macroglobulin and Fet A = fetuin A.
2. Materials and Methods
2.1 Materials
5

Adenosine, tert-Butyldimethylsilyl chloride (TBDMSCl), imidazole, N,N-diisopropylethylamine
(DIPEA), isoprene, 4-(dimethylamino)pyridine (DMAP), succinic anhydride, 4-hydroxy-TEMPO,
1-[bis(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5-b]pyridinium 3-oxid hexafluorophosphate
(HATU), rhodamine
B
base (Rho), N-(3-Dimethylaminopropyl)-N'-ethylcarbodiimide
exo-3,6-epoxy-1,2,3,6-tetrahydrophthalic anhydride, 2-(2-
hydrochloride (EDC•HCl),
aminoethoxy)-ethanol as well as dry solvents, corning transwells (polystyrene membrane, 12 well
plates, 0.4 µm pores), bovine α2M, bovine fetuin (Fet A), bovine serum albumin (BSA) and cell
medium components were purchased from Sigma-Aldrich (Lyon, France) and used as received.
Tetrabutylammonium fluoride (TBAF) was purchased from Alfa Aesar (Kandel, Germany). N-tert-
butyl-N-(1-diethylphosphono-2,2-dimethylpropyl) nitroxide (SG1, 85%) was kindly supplied by
Arkema. AMA-SG1 alkoxyamine [37], Rho-TEMPO [38] and tris-O-silylated adenosine [39] were
prepared as reported previously. All other reactants were purchased from Sigma-Aldrich (Lyon,
France) at the highest available purity and used as received. Deuterated chloroform (CDCl3) was
obtained from Eurisotop (France). All other solvents were purchased from Carlo-Erba (France) at
the highest grade. EBM-2 basal medium was from Lonza (Switzerland). Nanosep centrifugal
devices (Omega membrane, 300K) were from VWR International (Fontenay-sous-Bois, France).
BCA assay kit was from Thermo Fisher Scientific (Illkirch, France). hCMEC/D3 cells were
provided by Pierre-Olivier Couraud (Université Paris Descartes (Paris, France). FITC-dextran 40
kDa was purchased from Sigma-Aldrich (Milan, Italy).
2.2 Analytical methods
Nuclear Magnetic Resonance Spectroscopy (NMR). NMR spectroscopy was performed in 5 mm
1
diameter tubes in CDCl3 at 25 °C. H and 13C NMR spectroscopy was performed on a Bruker
Avance 300 spectrometer at 300 MHz (1H) or 75 MHz (13C). The chemical shift scale was
calibrated based on the internal solvent signals. To characterize nitroxide derivatives,
pentafluorophenylhydrazine was added in situ and allowed to react before the analysis [40, 41].
6

Mass spectrometry. Mass spectra were recorded with a Bruker Esquire-LC instrument. High-
resolution mass spectra (ESI) were recorded on an ESI/TOF (LCT, Waters) LC-spectrometer.
Size Exclusion Chromatography (SEC). SEC was performed at 30 °C with two columns from
Polymer Laboratories (PL-gel MIXED-D; 300 × 7.5 mm; bead diameter 5 mm; linear part 400 to 4
× 105 g.mol-1), a differential refractive index detector (Spectra System RI-150 from Thermo
Electron Corp.) and a scanning fluorescence detector (Waters 474). The eluent was chloroform at a
flow rate of 1 mL.min-1 (Waters 515 pump) and toluene was used as a flow-rate marker. The
calibration curve was based on polystyrene (PS) standards (peak molar masses, Mp = 162–523 000
g.mol-1) from Polymer Laboratories. A polyisoprene (PI) calibration curve was constructed by
converting the PS standard peak molecular weights (MPS) to PI molecular weights (MPI) using
Mark-Houwink-Sakurada (MHS) constants determined for both polymers in CCl4 at 25 °C. For PI,
the MHS constants used were KPI = 2.44 × 104 and αPI = 0.712. For PS, KPS = 7.1 × 104 and αPS =
0.54 (Mw < 16 700 g.mol-1) or KPS = 1.44 × 104 and αPS = 0.713 (Mw > 16 700 g.mol-1) [42]. This
technique allowed the number-average molar mass (Mn), the weight-average molar mass (Mw) and
the dispersity (Mw/Mn, Ð) to be determined.
Dynamic Light Scattering (DLS) and zeta potential. Nanoparticle diameters (Dz) and zeta
potentials (ζ) were measured by dynamic light scattering (DLS) with a Nano ZS from Malvern
(173° scattering angle) at a temperature of 25 °C. The surface charge of the NPs was investigated
by ζ-potential (mV) measurement at 25 °C after dilution with 1 mM NaCl, using the Smoluchowski
equation.
2.3 Synthesis
Synthesis of PI. AMA-SG1 (120 mg, 0.327 mmol) was placed in a 15 mL-capacity pressure tube
(Ace Glass 8648-164) fitted with a plunger valve and thermowell. After addition of isoprene (6.5
mL, 6.53 mmol) and dry 1,4-dioxane (6.5 mL), the tube underwent three cycles of freeze-thaw
degassing and was then backfilled with argon. The tube was placed in a preheated oil bath at 115 °C
for 16 h (PI1) and then cooled to r.t. by placing it under cold water. The residue was concentrated
7

under reduced pressure and precipitated in cold methanol to give PI as a colorless viscous oil. Two
other polymerizations were performed for 16 h with [isoprene]0/[AMA-SG1]0 = 500/1 (PI2) and
400/1 (PI3). All polymers were characterized by SEC and 1H NMR. PI1: Mn,SEC = 2860 g.mol-1, Đ
= 1.13. PI2: Mn,SEC = 4940 g.mol-1, Đ = 1.18. PI3: Mn,SEC = 4320 g.mol-1, Đ = 1.16.
Synthesis of PI-Rho. Briefly, PI1 (300 mg, 0.105 mmol, 1 eq) and TEMPO-Rho (72 mg, 0.105
mmol, 1 eq) were placed in a 7 mL-vial and dissolved in dry pyridine (1 mL). After 20 min of
degassing under argon, the vial was placed in a preheated oil bath at 110 °C and stirred for 16 h. PI-
Rho was then precipitated two times in cold methanol and dried under reduced pressure until
constant weight. Functionalization was confirmed by SEC equipped with a fluorescent detector (λex
= 570 nm, λem = 595 nm). Mn,SEC = 2690 g.mol-1, Đ = 1.13.
Synthesis of PI-OH. Briefly, P2 (0.500 g, 0.101 mmol, 1 eq) and 4-hydroxy-TEMPO (0.017 g,
0.101 mmol, 1 eq) were placed in a 7 mL-vial and dissolved in 1 mL of dry pyridine. After 20 min
of degassing under argon, the vial was placed in a preheated oil bath at 110 °C and stirred for 16 h.
PI-OH was then precipitated two times in cold methanol and dried under reduced pressure until
constant weight. The polymer was characterized by SEC and 1H NMR. Mn,SEC = 5030 g.mol-1, Đ =
1.15.
Synthesis of N-[2-(2-hydroxyethoxy)ethyl]-exo-3,6-epoxy-1,2,3,6-tetrahydrophthalimide. N-[2-
(2-Hydroxyethoxy)ethyl]-exo-3,6-epoxy-1,2,3,6-tetrahydrophthalimide was prepared according to a
modified published procedure [43]. Exo-3,6-epoxy-1,2,3,6-tetrahydrophthalic anhydride (0.50 g,
3.01 mmol) was dissolved in 2-(2-aminoethoxy)-ethanol (1.58 g, 15.0 mmol) and stirred at 80 °C
for 7 h. The reaction mixture was cooled to r.t. before addition of DCM. The solution was washed
1
with brine, dried over MgSO4 and concentrated to give a colorless oil. Yield: 39 % (0.318 g.). H
NMR (300 MHz, CDCl3) δ 6.51 (s, 2H), 5.29 (s, 2H), 3.76-3.50 (m, 8H), 2.87 (s, 2H), 2.59 (b, 1H)
ppm. 13C NMR (75 MHz, CDCl3) δ 176.38, 136.30, 81.02, 72.25, 67.06, 61.63, 47.40, 38.58.
Synthesis
of
succinic-mal-furan
(N-[2-(2-hydroxyethoxy)ethyl]-exo-3,6-epoxy-1,2,3,6-
tetrahydrophthalimide-4-oxobutanoic acid). Succinic-mal-furan was prepared according to a
8

modified published procedure [44]. Briefly, a solution of DMAP (0.290 g, 2.37 mmol) and succinic
anhydride (0.237 g, 2.37 mmol) in dry DCM (10 mL) was added dropwise at 0 °C and under an
argon atmosphere, to a solution of N-[2-(2-hydroxyethoxy)ethyl]-exo-3,6-epoxy-1,2,3,6-
tetrahydrophthalimide (0.200 g, 0.833 mmol) in dry DCM (2 mL). After stirring overnight at r.t.,
the solution was quenched with 1 M HCl. The product was extracted three times with DCM. The
combined organic layers were washed with brine and dried over MgSO4. After evaporation of the
1
solvent, succinic-mal-furan was obtained as a white solid. Yield: 96%. H NMR (CDCl3, 300
MHz): δ 6.53 (s, 2H), 5.30 (d, 2H), 4.27 – 4.15 (m, 2H), 3.66 (dq, 6H), 2.90 (s, 2H), 2.68 (s, 4H).
13C NMR (75 MHz, CDCl3) δ 176.39, 171.95, 136.54, 80.77, 68.50, 67.05, 63.81, 47.49, 38.24,
29.05, 28.88. MS (ESI-): m/z = 352.2 (M)-. Calc. for C16H19NO8: 353.1.
Synthesis of PI-mal-furan. A solution of succinic-mal-furan (0.043 g, 0,121 mmol, 2 eq),
EDC·HCl (0.023 g, 0.121 mmol, 2 eq) and DMAP (cat) in dry DCM (1 mL) was stirred for 15 min
and added dropwise to a solution of PI-OH (0. 300 g, 0.061 mmol, 1 eq) in DCM (1 mL). The
mixture was stirred overnight under argon at r.t., before being quenched with water and extracted
with DCM. The organic phases were then washed with brine and dried with MgSO4. After
evaporation of the solvent, two precipitation in cold methanol were performed to obtain PI-mal-
furan as a colorless oil. The polymer was characterized by SEC and 1H NMR. Mn,SEC = 5200 g.mol-
1, Đ = 1.15.
Synthesis of PI-mal. PI-mal-furan (0.200 g, 0.038 mmol) was dissolved in dry toluene (1 mL).
After 20 min of degassing under argon, the vial was placed in a preheated oil bath at 100 °C and
stirred for 16 h. PI-mal was then precipitated two times in cold methanol and dried under reduced
1
pressure until constant weight. The polymer was characterized by SEC and H NMR, which also
confirmed the disappearance of the maleimide protecting group. Mn,SEC = 5650 g.mol-1, Đ = 1.17.
Synthesis of Ade-PI. The TBDMS-protected TBDMS-Ade-PI2 (0.300 g, 0.050 mmol) was
dissolved in 1.5 mL THF and TBAF (1 M in THF, 150 μL) was added. The solution was stirred for
30 min and then purified by two consecutive precipitation in cold methanol to give Ade-PI as a
9

1
colorless viscous oil. The polymer was characterized by SEC and H NMR, which confirmed also
the disappearance of TBDMS protecting groups. Mn,NMR was calculated according to Mn,NMR
=
(DPn,NMR x MWisoprene) + MWTBDMS-Ade-AMA-SG1 – MWTBDMS with DPn,NMR calculated from ratio of
areas under the peak at 8.69, 8.31 and 5.98 ppm (aromatic and anomeric proton of Ade) and 5.0-5.5
ppm (vinylic H in isoprene repeat unit (1,4-addition), corresponding to ~81% of total isoprene units
[37]). Mn,SEC = 6260 g.mol-1, Mn,NMR = 6280 g.mol-1, Đ = 1.15.
Synthesis of tris-O-silylated adenosine-AMA-SG1 (TBDMS-Ade-AMA-SG1). Tris-O-silylated
adenosine (TBDMS-Ade, 2.0 g, 3.28 mmol) was dissolved in 10 mL of dry DMF. HATU (2.49 g,
6.56 mmol) and AMA-SG1 (2.37 g, 6.56 mmol) were dissolved in 10 mL of dry DMF in another
round-bottomed flask, DIPEA (2.12 g, 16.4 mmol) was added dropwise by syringe. The mixture
was stirred for 1 h and added by syringe to the first solution. The solution was stirred under argon
atmosphere for 96 h at room temperature (r.t.), then diluted with 100 mL EtOAc.The organic phase
was washed with 1 M HCl, sat. NaHCO3 aqueous solution, and brine before being dried over
MgSO4. The residue was concentrated under reduced pressure and purified by flash
chromatography (SiO2, gradient from PetrolEther/EtOAc 9/1 v/v to PetrolEther/EtOAc 5/5 as
1
eluent) to give 1.95 g of TBDMS-Ade-AMA-SG1 as a white-yellow solid. Yield: 62%. H NMR
(CDCl3, 300 MHz): δ 11.01 (d, 1H), 8.76 (s, 1H), 8.31 (d, 1H), 6.09 (m, 1H), 4.93-4.82 (d, 1H),
4.69 (dt, 1H), 4.37-4.29 (m, 2H), 4.23-3.98 (m, 5H), 3.81 (dt, 1H), 3.49 (d, 1H), 1.62 (d, 3H), 1.32
(t, 6H), 1.22-1.20 (m, 18H), 0.95 (d, 18H), 0.81 (d, 9H), 0.12 (d, 12H), -0.03 (d, 3H), -0.24 (d, 3H)
13
ppm. C NMR (75 MHz, CDCl3): δ 169.75, 152.42, 152.06, 149.66, 141.76, 88.39, 85.51, 85.19,
78.82, 78.69, 75.58, 72.02, 71.72, 70.04, 68.21, 62.63, 62.44, 61.82, 60.65, 35.41, 30.54, 28.24,
26.08, 25.85, 25.68, 18.50, 18.08, 17.83, 16.50, 16.23, -4.40, -4.71, -5.05, -5.39 ppm. MS (ESI+):
m/z = 981.6 (M+Na)+. Calc. for C29H48N3O7P: 958.6.
Synthesis of TBDMS-Ade-PI. TBDMS-Ade-AMA-SG1 (110 mg, 0.115 mmol) was placed in a 15
mL-capacity pressure tube (Ace Glass 8648-164) fitted with plunger valves and thermowells. After
addition of isoprene (5.7 mL, 57.2 mmol) and dry 1,4-dioxane (5.7 mL), the tube underwent three
10

cycles of freeze-thaw degassing and was then backfilled with argon. The tube was then placed in a
preheated oil bath at 115 °C for 16 h (TBDMS-Ade-PI1) and cooled to r.t. by placing under cold
water. The residue was concentrated under reduced pressure and precipitated in cold methanol to
give TBDMS-Ade-PI as a colorless viscous oil. Another polymerization was performed for 16 h
with [isoprene]0/[TBDMS-Ade-AMA-SG1]0 = 780/1 (TBDMS-Ade-PI2). All polymers were
1
characterized by SEC and H NMR. Mn,NMR was calculated according to Mn,NMR = (DPn,NMR
x
MWisoprene) + MWTBDMS-Ade-AMA-SG1 with DPn,NMR calculated from the ratio of areas under the peak
at 8.69, 8.31 and 5.98 ppm (aromatic and anomeric proton of Ade) and 5.0-5.5 ppm (vinylic H in
isoprene repeat unit (1,4-addition), corresponding to ~81% of total isoprene units [37]). TBDMS-
Ade-PI1: Mn,SEC = 6000 g.mol-1, Mn,NMR = 6450 g.mol-1 and Đ = 1.16. TBDMS-Ade-PI2: Mn,SEC
5100 g.mol-1, Mn,NMR = 5050 g.mol-1, and Đ = 1.16.
=
Synthesis of TBDMS-Ade-PI-OH. Briefly, TBDMS-Ade-PI1 (0.500 g, 0.098 mmol, 1 eq) and 4-
hydroxy-TEMPO (0.017 g, 0.098 mmol, 1 eq) were placed in a 7 mL-vial and dissolved in 1 mL of
dry pyridine. After 20 min of degassing under argon, the vial was placed in a preheated oil bath at
110 °C and stirred for 16 h. TBDMS-Ade-PI-OH was then precipitated two times in cold methanol
and dried under reduced pressure until constant weight. The polymer was characterized by SEC and
1H NMR. Mn,NMR was calculated according to Mn,NMR = (DPn,NMR x MWisoprene) + MWTBDMS-Ade-AMA-
– MWSG1 + MWTEMPO-OH with DPn,NMR calculated from ratio of areas under the peak at 8.69,
SG1
8.31 and 5.98 ppm (aromatic and anomeric proton of Ade) and 5.0-5.5 ppm (vinylic H in isoprene
repeat unit (1,4-addition), corresponding to ~81% of total isoprene units [37]). Mn,SEC = 5330 g.mol-
1, Mn,NMR = 7590 g.mol-1, Đ = 1.15.
Synthesis of TBDMS-Ade-PI-mal-furan. A solution of succinic-mal-furan (0.041 g, 0.117 mmol,
2 eq), EDC·HCl (0.022 g, 0.117 mol, 2 eq) and DMAP (cat) in dry DCM (1 mL) was stirred for 15
min and added dropwise to a solution of TBDMS-Ade-PI-OH (0.300 g, 0.059 mmol, 1 eq) in dry
DCM (1 ml). The mixture was stirred overnight under argon at r.t., before being quenched with
water and extracted with DCM. The organic phases were then washed with brine and dried with
11

MgSO4. After evaporation of the solvent, two precipitations in cold methanol were performed to
obtain TBDMS-Ade-PI-mal-furan as a colorless oil. The polymer was characterized by SEC and 1H
1
NMR. The post-functionalization yield was calculated by H NMR using the chemical shifts of
protons of mal ( = 6.53, 4.22, 2.90 and 2.68 ppm) and the chemical shifts of aromatic protons and
anomeric proton of Ade ( = 8.69, 8.31 and 5.98 ppm). Mn,NMR was calculated according to Mn,NMR
= (DPn,NMR x MWisoprene) + MWTBDMS-Ade-AMA-SG1 – MWSG1 + MWTEMPO-mal-furan with DPn,NMR
calculated from ratio of areas under the peak at 8.69, 8.31 and 5.98 ppm (aromatic and anomeric
proton of Ade) and 5.0-5.5 ppm (vinylic H in isoprene repeat unit (1,4-addition), corresponding to
~81% of total isoprene units [37]). Mn,SEC = 5370 g.mol-1, Mn,NMR = 9120 g.mol-1, Đ = 1.15.
Synthesis of Ade-PI-mal-furan. The TBDMS-protected TBDMS-Ade-PI-mal-furan (0.300 g,
0.056 mmol) was dissolved in 1.5 mL THF and TBAF (1 M in THF, 150 μL) was added. The
solution was stirred for 30 min and then purified by two consecutive precipitations in cold methanol
to give Ade-PI-mal-furan as a colorless viscous oil. It was then directly used for the next reaction.
Synthesis of Ade-PI-mal. Ade-PI-mal-furan (0.200 g, 0.036 mmol) was dissolved in dry toluene (1
mL). After 20 min of degassing under argon, the vial was placed in a preheated oil bath at 100 °C
and stirred for 16 h. Ade-PI-mal was then precipitated two times in cold methanol and dried under
1
reduced pressure until constant weight. The polymer was characterized by SEC and H NMR,
which also confirmed the disappearance of maleimide protecting group. The post-functionalization
yield was calculated by 1H NMR from ratio of areas under the peak of protons of mal at 6.53, 4.22,
2.90 and 2.68 ppm and aromatic and anomeric protons of Ade at 8.69, 8.31 and 5.98 ppm. Mn,NMR
was calculated according to Mn,NMR = (DPn,NMR x MWisoprene) + MWTBDMS-Ade-AMA-SG1 – MWSG1
–
MWTBDMS + MWTEMPO-mal with DPn,NMR calculated from ratio of areas under the peak at 8.69, 8.31
and 5.98 ppm (aromatic and anomeric proton of Ade) and 5.0-5.5 ppm (vinylic H in isoprene repeat
unit (1,4-addition), corresponding to ~81% of total isoprene units [37]). Mn,SEC = 5390 g.mol-1,
Mn,NMR = 7230 g.mol-1, Đ = 1.37. mal/Ade = 0.63
12

Synthesis of Ade-PI-Rho. Briefly, Ade-PI (0.300 g, 0.048 mmol, 1 eq) and TEMPO-Rho (0.038 g,
0.048 mmol, 1 eq) were placed in a 7 mL-vial and dissolved in 1 mL of dry pyridine. After 20 min
of degassing under argon, the vial was placed in a preheated oil bath at 110 °C and stirred for 16 h.
Ade-PI-Rho was then precipitated two times in cold methanol and dried under reduced pressure
1
until constant weight. The post-functionalization yield was calculated by H NMR using the
chemical shifts of aromatic protons of Rho ( = 7.70, 7.55, 7.08 and 6.93-6.67 ppm) and the
chemical shifts of aromatic protons and anomeric proton of Ade ( = 8.69, 8.31 and 5.98 ppm).
Functionalization was confirmed by SEC equipped with a fluorescent detector (λex = 570 nm, λem
=
595 nm). Mn,NMR was calculated according to Mn,NMR = (DPn,NMR x MWisoprene) + MWTBDMS-Ade-AMA-
– MWSG1 – MWTBDMS + MWTEMPO-Rho with DPn,NMR calculated from ratio of areas under the
SG1
peak at 8.69, 8.31 and 5.98 ppm (aromatic and anomeric proton of Ade) and 5.0-5.5 ppm (vinylic H
in isoprene repeat unit (1,4-addition), corresponding to ~81% of total isoprene units [37]). Mn,SEC
7540 g.mol-1, Mn,NMR = 6120 g.mol-1, Đ = 1.32. Rho/Ade = 0.94
=
2.4 Nanoparticle preparation
PI (PI3), Ade-PI, PI-mal and Ade-PI-mal NPs were prepared by the nanoprecipitation technique
[45]. All NPs were obtained by co-nanoprecipitation of the desired polymer with a Rho-containing
PI to reach 5 wt.% of Rho moieties with respect to the overall weight of NPs, as follows: Ade-PI-
Rho/Ade-PI, Ade-PI-Rho/Ade-PI-mal, PI-Rho/PI and PI-Rho/PI-mal. This percentage gave the
best compromise in terms of fluorescence signal and NP colloidal characteristics (e.g., average size,
stability). Briefly, 2.5 mg of the corresponding polymers were dissolved in 0.5 mL of THF and the
solution was quickly added to 1 mL MilliQ water. THF was evaporated at r.t. using a Rotavapor.
Intensity-averaged diameter (Dz) in MilliQ water and zeta potential measurements in 1 mM NaCl
were carried out in triplicate by DLS. Their colloidal stability was assessed in water for 14 days.
NPs were stored at 4 °C and allowed to reach r.t. before each measurement. To establish the best
conditions for protein functionalization, PI and Ade-PI NP colloidal stability was tested following
13

incubation at r.t. overnight and at 37 °C for 30 min, in either PBS or hCMEC/D3 cell medium
without serum (which is important to avoid coating with serum proteins).
2.5 hCMEC/D3 cell culture
Immortalized human cerebral microvascular endothelial cells (hCMEC/D3) cells were used as a
well-established model of the BBB that forms tight junctions (TJs), the major mechanism that
prevents passage of most biomolecules and pharmacological agents from the blood to the brain [3,
48-50]. Briefly, hCMEC/D3 cells were seeded (60.000 cells.cm-2) between passages 25 and 35 on
the apical side of transwell filters coated with rat-tail collagen type I (4 µg.cm-2). Cells were grown
for 14 days in EBM-2 medium supplemented with fetal bovine serum (5%), hydrocortisone (1.4
µM), HEPES (10 mM), penicillin-streptomycin (1%), basic fibroblast growth factor (1 ng.mL-1),
ascorbic acid (5 µg.mL-1), and lipid concentrate (1/100). Medium was changed every 2-3 days (500
µl in the apical compartment, 1 ml in the basolateral compartment). Trans-endothelial electrical
resistance (TEER) was measured every day to monitor the formation of tight junctions (EVOM2,
World Precision Instruments), and was also measured in the presence of NPs (0.1 mg.mL-1 added to
the apical compartment for 3 h at 37 °C) and medium with no serum. Integrity of tight junctions
was assessed by measuring the paracellular permeability of lucifer yellow (LY) across hCMEC/D3
monolayer. Briefly, LY (50 µM in PBS) was added to the apical compartment of the transwell
system for 60 min. Every 30 min, samples were collected from the basolateral compartment.
Quantification of fluorescence (λex = 425 nm, λem = 530 nm) was carried out using a
spectrofluorimeter (Perkin Elmer, LS50 B). The apparent permeability coefficient (Papp) for LY
was calculated as previously described [51]. Permeability of FITC-dextran (40 kDa) was assessed
after incubating cells in transwells with medium supplemented with either 0 % or 5 % serum for 24
h. All medium was removed, and FITC-dextran was added to the apical compartment (1 mg.mL-1 in
14

PBS), with 1 mL serum-depleted medium in the basolateral compartment. After 3 h incubation,
fluorescence spectroscopy was used to quantify FITC-dextran in both compartments, and
endothelial permeability was calculated [51].
2.6 Cell viability in the presence of polymer NPs
The viability of hCMEC/D3 cells after exposure to medium with no serum or NPs for 3 h at 37 °C
was assessed using the MTT assay, as previously described [52]. Four types of bare NPs were used,
each labelled with 5% wt. Rho: PI (PI), Ade-PI (Ade-PI), PI-mal (PI-mal) and Ade-PI-mal (Ade-
PI-mal).
2.7 NP functionalization with physiological proteins and protein corona formation
Protein functionalization. 100 L of PI and Ade-PI NPs (for adsorption) or PI-mal and Ade-PI-
mal NPs (for covalent conjugation) at 2.5 mg.mL-1 were incubated with 400 L of one of the
proteins (MWFet A = 49 kDa, MWα2M = 725 kDa and MWBSA = 66 kDa; 1.25 mg.mL-1) in
hCMEC/D3 serum-free cell medium overnight at r.t., under which conditions the NPs were stable,
as determined by previous stability tests. Free proteins were separated from NP-protein complexes
by ultrafiltration (MWCO 300 kDA) at r.t. for 3 min at 14,000 g. NPs were resuspended in serum-
free medium, and BCA assay (Thermo Fischer Scientific) was used to determine protein binding.
Protein corona formation. To test potential size changes upon protein corona formation, 25 L
bare, protein-adsorbed or protein-conjugated NPs (62.5 g) were incubated in 200 L cell medium
containing 5% FBS for 30 min at 37 C. Unbound proteins were removed by ultrafiltration (MWCO
300 kDA) for 6 min at 14,000 g. NPs were resuspended in MilliQ water and size was measured by
DLS.
15

Assessment of protein covalent binding. To establish whether proteins can covalently bind to the
maleimide moieties present on the polymeric NPs, Ellman’s protocol was used to determine free
thiol groups [53, 54]. PI, Ade-PI, PI-mal and Ade-PI-mal NPs (50 g) were incubated with each
protein (150 g) in 500 l serum-free medium overnight at r.t., and free protein was removed by
ultrafiltration as before. 5 L of Ellman’s reagent (2 mM DTNB in 50 mM sodium acetate) was
added to NP-protein samples diluted 1:15 in PBS (final volume 150 L) in 96 well plates. Plate was
shaken and incubated at r.t. for 5 min. Optical absorbance was measured at 412 nm. A standard
curve was created using known amounts of N-acetyl-L-cysteine, and nmol of free cysteine per mg
protein was calculated. All samples were measured in triplicate and experiments were repeated
three times.
2.8 Endothelial permeability and cellular uptake of NPs in an in vitro model of the healthy
BBB
On the 14th day after cell seeding in the transwell system, 0.1 mg.mL-1 of bare, protein-adsorbed or
protein-conjugated NPs (determined to be non-toxic by MTT assay) suspended in 500 µL serum-
free medium were added to the apical “blood” compartment of the transwell for 3 h at 37 °C. Note
that the use of serum-free cell medium has already been utilized for similar NP permeability studies
using this cell model [46, 47]. Also, cell viability (MTT), permeability (TEER and EP after
incubation with FITC-dextran, MW = 40 kDa) were tested using serum-free medium and there was
no effect on hCMEC/D3 permeability and viability after 3 h incubation. The basolateral “brain”
compartment contained 1 mL of serum-free medium. At different time points for up to 3 h, an
aliquot from the basolateral compartment was collected and fluorescence of Rho was measured
using a spectrofluorimeter (Perkin Elmer, LS50 B). Excitation was set at 555 nm, and emission at
578 nm for PI and PI-mal NPs or 587 nm for Ade-PI and Ade-PI-mal NPs. For each experiment, a
standard curve was established using bare NPs of a known concentration that were not subjected to
ultrafiltration. Each experiment used triplicate wells and was repeated at least three times.
16

Transwell filters without cells were used as a control to calculate the endothelial permeability (EP)
of NPs across the filter. EP takes into account passage of NPs through the transwell filter alone to
give a value representative of the rate of passage through the cell monolayer, and was calculated as
previously described [51].
To measure cellular uptake of NPs after 3 h incubation, cells were washed twice with
Dulbeccos’s PBS (DPBS) and incubated with 0.25% trypsin-EDTA for 5 min at 37 °C. Control
wells without cells were used to rule out NPs adherence to the filter. Trypsinized cells were
collected and centrifuged at 2,000 rpm (1500 g) for 5 min. Supernatant was discarded and the pellet
of cells was resuspended in 750 µL cold lysis buffer (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 2
mM EDTA, 1% v/v Triton X-100, 0.1% SDS, 1 mM DTT). Cells were incubated for 30 min at r.t.
while rotating in the dark and then centrifuged at 10,000 rpm (7500 g) for 5 min. The supernatant
was collected and fluorescence was measured as above described. Cellular uptake was expressed as
a percentage of the initial amount of NPs added to the apical compartment, subtracting the
fluorescence measurements obtained from wells without cells.
2.9 Visualization of cellular uptake of NPs by hCMEC/D3 cells using fluorescence microscopy
hCMEC/D3 cells were seeded on rat collagen type I-coated Cell Carrier Ultra 96 well plates
(10,000 cells/well) and grown for two days. When cells were confluent, they were incubated with
100 µL bare PI, Ade-PI, PI-mal and Ade-PI-mal NPs in serum-free cell medium (0.1 mg.mL-1) for
3 h at 37 °C. NPs were then removed from the wells and cells were washed twice with DPBS. Cells
were fixed with 10% formalin solution (100 µL) at r.t. for 20 min, followed by three 10 min washes
with DPBS. Cells were permeabilized with 100 µL Triton (0.2 % v/v) for 15 min at r.t. and washed
once with DPBS. The actin cytoskeleton was stained with Phalloidin Alexafluor 633 (1:100 in
DPBS) for 1 h at r.t., and then washed 3 times with DPBS. Hoescht (1:5,000 in DPBS) was
incubated for 15 min at r.t. to stain nuclei. Cells were washed three times in DPBS and all wells
were filled with MilliQ water (100 µL). Plates were stored in the dark at 4 °C until visualization.
17

Fluorescent microscopy images were acquired using the Operetta CLS High-Content Analysis
System (PerkinElmer) equipped with 40x water objective. Excitation of NPs was set to 530-560
nm, emission at 570-650 nm.
3. Results and Discussion
3.1 Overall synthetic strategy
Heterotelechelic polymers were synthesized by the “drug-initiated” method [55]. It relies on the
controlled growth of a polymer chain from a drug covalently linked to an “initiator” using
reversible-deactivation radical polymerization (RDRP) techniques, especially nitroxide-mediated
polymerization (NMP) [42, 56-59] and reversible addition-fragmentation chain transfer
polymerization (RAFT) [59-64]. This synthetic approach has key advantages: (i) purification of the
drug-polymer prodrug is simplified as only unreacted monomer has to be removed; (ii) the
conjugation of the drug is quantitative, leading to well-defined conjugates with one drug molecule
attached to each polymer chain and (iii) the drug loading (DL) can be fine-tuned by varying the
polymer chain length; the lower the chain-length, the higher the DL. Polyisoprene (PI) was selected
as the polymer for its biocompatibility [65] and structural similarity to natural terpenoids, hence
representing a potential material for drug delivery applications [38, 42, 57, 59].
To functionalize the other polymer chain-end and obtain heterotelechelic polymer prodrugs,
we recently developed a general and robust methodology consisting of applying a nitroxide-
exchange reaction to “drug-initiated” synthesized polymers obtained by NMP [38, 66]. The
terminal SG1 nitroxide was quantitatively replaced by a functional TEMPO nitroxide bearing the
molecule of interest, which was either a fluorescent probe for imaging applications or a second drug
for combination therapy. This methodology has important benefits such as nearly quantitative post-
functionalization yields and easy purification given the absence of any catalyst or reactant other
than the free nitroxide and the polymer. Also, TEMPO can be functionalized by a variety of
different moieties at the 4-position.
18

To allow NP surface functionalization with cysteine-containing proteins, the TEMPO
nitroxide was derivatized by a protected maleimide group [67-69]. TEMPO was also functionalized
with a Rho moiety to endow NPs with fluorescence properties by a simple co-nanoprecipitation
approach. Rho was chosen as a fluorescent probe because of its favorable characteristics such as
good photostability, high quantum yield, high extinction coefficient and higher emission
wavelengths than those coming from cell autofluorescence [70]. Rho was also previously modified
with a piperazine ring to avoid intramolecular cyclization and allow good stability with pH
modifications [71]. The covalent linkage of Rho to the polymer, as opposed to simple
encapsulation, is intended to avoid potential leakage of the dye, which could result in the detection
of artifact coming from the free dye by fluorescence spectroscopy [38, 58].
3.2 Synthesis of heterotelechelic polymer prodrug
PI was synthesized from AMA-SG1 alkoxyamine initiator (Scheme 1a) with Mn ranging from 2860
to 4940 g.mol-1 and low dispersity (1.13–1.18, Table 1). PI-Rho was obtained by nitroxide
exchange reaction from PI and TEMPO-Rho (Scheme 1b and Table 1). Unfortunately, it was not
possible to calculate the post-functionalization yield by 1H NMR because of the lack of a suitable
proton in α position to be compared with those of Rho, but previous demonstration by electron spin
resonance [38] and characterization of heterotelechelic polymers showed below, led us to assume it
was quasi-quantitative.
19

Scheme 1. (a) Synthesis of polyisoprene (PI) by nitroxide-mediated polymerization (NMP) initiated
by the AMA-SG1 alkoxyamine. (b) Synthesis of PI-rhodamine (PI-Rho) by the nitroxide exchange
reaction with TEMPO-rhodamine (TEMPO-Rho) on PI. (c) Synthesis of adenosine-polyisoprene
(Ade-PI) by NMP of isoprene from tris-O-silylated-adenosine-AMA-SG1 alkoxyamine (TBDMS-
Ade-AMA-SG1), followed by deprotection of Ade. (d) Synthesis of adenosine-polyisoprene-
rhodamine (Ade-PI-Rho) by the nitroxide exchange reaction with TEMPO-rhodamine (TEMPO-
Rho) on Ade-PI.
20

Table 1. Characterization of Polyisoprene (PI), Polyisoprene-Rhodamine (PI-Rho), Adenosine-
Polyisoprene (Ade-PI), Adenosine-Polyisoprene-Rhodamine (Ade-PI-Rho), Adenosine-
Polyisoprene-Maleimide (Ade-PI-mal) and Polyisoprene-Maleimide (PI-mal).
a
e
Prodrug
Mn,SEC
Ða
Mn,NMR
DPn,NMR
Rho or mal
Drug loading
(g.mol-1)
(g.mol-1)
/Adef
(wt.%) g
PI
2860
2690
6260
7540
5390
5650
1.13
1.13
1.15
1.32
1.37
1.17
-
-
-
-
-
-
PI-Rho
-
-
-
Ade-PI
6280b
6120c
7230d
-
83
74
95
-
4.3
3.5
5.0
-
Ade-PI-Rho
Ade-PI-mal
PI-mal
0.94
0.63
-
a
Determined by SEC, calibrated with PS standards and converted into PI by using Mark-Houwink-Sakurada
parameters. b Calculated according to Mn,NMR = (DPn,NMR x MWisoprene) + MWTBDMS-Ade-AMA-SG1 – MWTBDMS . c Calculated
according to Mn,NMR = (DPn,NMR x MWisoprene) + MWTBDMS-Ade-AMA-SG1 – MWSG1 – MWTBDMS + MWTEMPO-Rho. d Calculated
according to Mn,NMR = (DPn,NMR x MWisoprene) + MWTBDMS-Ade-AMA-SG1 - MWSG1 – MWTBDMS + MWTEMPO-mal. e Calculated
from ratio of areas under the peak at 8.69, 8.31 and 5.98 ppm (aromatic and anomeric proton of Ade) and 5.0-5.5 ppm
(vinylic H in isoprene repeat unit (1,4-addition), corresponding to ~81% of total isoprene units [37]). f Calculated from
ratio of areas under the peak of aromatic protons of Rho at 7.70, 7.55, 7.08 (or of mal at 6.53, 4.22, 2.90 and 2.68 ppm)
and 6.93-6.67 ppm and aromatic and anomeric protons of Ade at 8.69, 8.31 and 5.98 ppm. g Calculated according to
MWAde/Mn,SEC
.
As for Ade-PI (Scheme 1c), Ade hydroxyl groups were first protected by TBDMS and the resulting
TBDMS-Ade was covalently linked to AMA-SG1 alkoxyamine initiator by HATU coupling to give
TBDMS-Ade-AMA-SG1 (62% overall yield). By varying the experimental conditions, it then
served as an initiator for the NMP of isoprene to yield well-defined TBDMS-Ade-PI (see
experimental part) with Mn,SEC ~5500 g.mol-1 and Đ = 1.16. After deprotection in the presence of
1
TBAF, Ade-PI was obtained. The effectiveness of the reaction was confirmed by H NMR, which
showed the disappearance of the silylated protecting groups (Figure S1) and a slight decrease of the
Mn,NMR because of TBDMS group removal (see experimental part and Table 1).
Post-functionalization by nitroxide exchange reaction was performed on Ade-PI in presence
of 1 eq. of TEMPO-Rho to yield Ade-PI-Rho as a fluorescent heterotelechelic polymer prodrug
1
(Scheme 1d and Table 1). Ade-PI-Rho was characterized by H NMR spectroscopy and gave the
expected structure. Rho aromatic protons were detected in the 6.5–8 ppm region, together with
those from the piperazine ring in the 4-4.5 ppm region (Figure S2). Complete disappearance of SG1
21

protons in the 3.2-3.4 ppm region was also observed. Integration of aromatic proton signals of Ade
and Rho gave a nearly quantitative post-functionalization yield as the Rho/Ade molar ratio was as
high as 94 mol.% (Table 1). SEC equipped with a fluorescent detector revealed comparable molar
mass distribution profiles for both RI and fluorescence traces, supporting a uniform end-capping
process by TEMPO-Rho (Figure S3a). Moreover, RI traces between Ade-PI and Ade-PI-Rho were
almost perfectly overlaid, indicating negligible irreversible termination by recombination (Figure
S3b).
To allow NP surface functionalization with BBB-targeting proteins by covalent linkage, a
maleimide moiety was installed at the -chain end of PI and Ade-PI through the nitroxide
exchange reaction. However, direct nitroxide exchange from a TEMPO bearing a furan-protected
maleimide group led to extensive occurrence of termination reactions and doubling of the starting
Mn (Figure S4), likely because of in situ deprotection of the maleimide group at elevated
temperature and radical addition onto it. The furan-protected maleimide was therefore introduced in
a two-step process, by first performing the nitroxide exchange reaction with 4-hydroxy-TEMPO,
followed by coupling with a succinate furan-protected maleimide (succinic-mal-furan, Scheme 2a).
PI-mal was also prepared by subjecting PI to nitroxide exchange in the presence of 4-hydroxyl-
TEMPO to give PI-OH. After coupling with succinic-mal-furan under EDC assistance to give PI-
mal-furan, PI-mal was recovered after maleimide deprotection (see experimental part, Scheme 2b
1
and Table 1). SEC analysis did not show noticeable changes in molar mass distributions and H
NMR confirmed the presence of maleimide and disappearance of SG1 protons. Even though
absence of a suitable proton in α-position prevented determination of the coupling yield, it was
expected to be very similar to Ade-PI-mal.
22

Scheme 2. (a) Synthesis of adenosine-polyisoprene-maleimide (Ade-PI-mal) by nitroxide-mediated
polymerization (NMP) of isoprene from tris-O-silylated-adenosine-AMA-SG1 alkoxyamine
(TBDMS-Ade-AMA-SG1), followed by the nitroxide exchange with 4-hydroxy-TEMPO on
TBDMS-Ade-AMA-SG1, EDC-assisted functionalization by succinic-mal-furan and subsequent
deprotections of Ade and maleimide. (b) Structure of PI-mal obtained by the nitroxide exchange
with 4-hydroxy-TEMPO on PI, EDC-assisted functionalization by succinic-mal-furan and
subsequent deprotection of maleimide.
A furan-protected maleimide derivative was reacted with succinic anhydride to give succinic-mal-
furan with a 96% yield (Scheme 2a), meanwhile the nitroxide exchange reaction was performed on
TBDMS-Ade-PI with 4-hydroxy-TEMPO to give TBDMS-Ade-PI-OH (Scheme 2b and Figure
S5). Even though absence of visible protons peaks on TEMPO prevented calculation of the
23

1
nitroxide exchange yield by H NMR, disappearance of SG1 protons (Figure S6) and the previous
successful use of this reaction [38] indicates that it was quantitative. Subsequent coupling between
TBDMS-Ade-PI-OH and a furan-protected maleimide assisted by EDC yielded the expected
TBDMS-Ade-PI-mal-furan. The post-functionalization yield was calculated by comparing Ade
aromatic and anomeric protons ( = 6.53, 4.22, 2.90 ppm) and the maleimide proton ( = 2.68
ppm), and reached 60% (Figure S5, experimental part). Ade was then deprotected with TBAF
followed by maleimide deprotection in toluene at 100 °C to give Ade-PI-mal (Table 1). Complete
disappearance of both protecting groups, appearance of a signal corresponding to maleimide
protons at 7.14 ppm, and shift of Ade anomeric proton from 6.10 to 6.24 ppm (likely due to a
change of the chemical environment), confirmed the effectiveness of the deprotections (Figure S6).
All polymers were also characterized by SEC and showed good conservation of the macromolecular
characteristics throughout the different steps (Figure S7, experimental part).
In summary, six different polymers were synthesized (Table 1): (i) two Ade prodrug
polymers, bearing a maleimide group (Ade-PI-mal) or not (Ade-PI); (ii) two Ade-free polymers,
bearing a maleimide group (PI-mal) or not (PI) and (iii) two fluorescent polymers, bearing Ade
(Ade-PI-Rho) or not (PI-Rho).
3.2 NP formulation and colloidal characteristics
The previously synthesized polymers were formulated into NPs in order to compare bare (i.e.,
protein-free) NPs to NPs functionalized with BBB-targeting proteins, either by adsorption or
conjugation, and to investigate the influence of the drug on NP colloidal properties. Drug loading of
the different polymer prodrugs ranged from 3.5 to 5 wt.%, depending on the Mn of the polymer
(Table 1). It is worth noting that higher drug loading values could be easily obtained by targeting
lower molar mass polymers. Here, a relatively high Mn was targeted to increase NP surface
hydrophobicity, which is known to increase protein affinity towards the NP surface [72, 73].
24

Four different types of NPs were formulated by the nanoprecipitation technique at a
concentration of 2.5 mg.mL-1: drug-free NPs (PI) and polymer prodrug NPs (Ade-PI) for protein
adsorption or drug-free NPs containing maleimide groups (PI-mal) and polymer prodrug NPs
containing maleimide groups (Ade-PI-mal) for protein conjugation (Table 2). All NPs were also
fluorescently-tagged by co-nanoprecipitation with PI-Rho or Ade-PI-Rho (5 wt.% Rho) to confer
fluorescent properties and allow tracing during biological evaluations.
Table 2. Colloidal Characteristics of NPs Prepared by Co-nanoprecipitation of PI, Ade-PI, PI-mal
or Ade-PI-mal with PI-Rho or Ade-PI-Rho (5 wt.% Rho).
a
Dz
 a
Sample
PSD a
(nm)
(mV)
PIb
185 ± 1
179 ± 3
164 ± 14
162 ± 3
0.13 ± 0.001
0.11 ± 0.03
0.21 ± 0.01
0.10 ± 0.02
49 ± 2
45 ± 1
41 ± 15
50 ± 4
Ade-PIc
PI-malb
Ade-PI-malc
a Measured by dynamic light scattering (DLS) as an average of three different measures. b Co-nanoprecipitated with PI-
Rho (5 wt.% Rho). c Co-nanoprecipitated with Ade-PI-Rho (5 wt.% Rho).
NPs were very similar in terms of average size (Dz = 162–185 nm) and had low particle size
distribution (PSD) values varying from 0.10 to 0.21. Whereas zeta potential of PI-based prodrug
NPs is usually strongly negative [42, 57], all NPs here had a strong positive zeta potential value,
between +40 and +50 mV. This is due to the presence of positive charges on the Rho molecule, as
already observed for other PI NPs labeled with this fluorophore [38]. In addition to the nanoparticle
surface hydrophobicity provided by the polymer, the positive charge from Rho could also increase
the NP affinity towards the proteins used for functionalization [74-78]. Note that Rho has been
extensively used for NP and peptide labelling in cell uptake/permeability studies, but the positive
charge of the Rho alone has not been enough to dictate the cell-NP interactions [79].
NP colloidal stability was assessed in water and they were all stable for at least 14 days
(Figure 2a). To optimize protein functionalization conditions, PI and Ade-PI colloidal stability was
25

also tested in PBS and hCMEC/D3 cell culture medium without serum at r.t., at 4 °C overnight, or
at 37 °C for 30 min (Figure 2b and 2c). The hCMEC/D3 immortalized human brain endothelial cell
line is a widely used model of the BBB, forming the characteristic tight junctions, architectural
structure, and protein expression found in the human barrier [80]. NPs were stable in cell medium
but aggregated in PBS, therefore the former was used in further experiments.
Figure 2. (a) Evolution of the intensity-averaged diameters (Dz) and particle size distributions
(PSD) in water of Ade-PI, PI, Ade-PI-mal and PI-mal NPs blended with PI-Rho or Ade-PI-Rho (5
wt.%. Rho) measured by DLS over 14 days. Stability in milliQ water, PBS and cell culture medium
without serum incubated overnight (O.N.) at room temperature (r.t.), O.N. at 4 °C, or for 30 min at
37 °C of (b) PI or (c) Ade-PI. NPs at 0.1 mg.mL-1 in order to determine the best conditions for
protein functionalization (n = 3, error bars represent SEM).
3.3 Coating and functionalization of NPs with BBB crossing proteins
We recently found that specific proteins are enriched in the protein corona of gold NPs following
passage through a transwell model of the BBB [48], emphasizing their potential ability to reach the
brain after in vivo administration. It was postulated that the formation of an artificial corona
consisting of one of these specific proteins could enhance the passage of NPs through the BBB.
The presence of maleimide on the polymer should allow covalent binding of NPs with
proteins that have a free cysteine residue. BSA [81], α2M [82] and Fet A [83] have cysteine
26

residues at different oxidation states that could be theoretically available for covalent binding to PI-
mal and Ade-PI-mal NPs. Moreover, given the strongly positive NP surface charge (Table 2),
simple adsorption of the proteins was also investigated, as it has been extensively shown that a wide
array of proteins are easily adsorbed via strong electrostatic interactions with the surface of NPs
[84]. However, there are several advantages of choosing covalent binding over the simple
adsorption of proteins on the surface of NPs, including improved selectivity of protein binding.
Furthermore, as covalent bonds are stronger than electrostatic forces, this may possibly prevent the
displacement of functionalized proteins by serum proteins upon in vivo administration, or their
removal in the purification process [85].
NPs were incubated overnight at r.t. with each protein suspended in hCMEC/D3 cell
medium without serum, to avoid serum protein interferences, then ultrafiltrated to remove unbound
protein. BCA quantification indicated that the highest amount of protein binding occurred with α2M
on all NPs (1.8-2.4 µg protein/µg NP) and the lowest with Fet A (0.5-1.1 µg protein/µg NP),
possibly due to differences in MW as MWα2M = 725 kDa and MWFet A = 49 kDa (Figure 3). The
presence of Ade significantly increased BSA binding in the presence of mal (1.4 µg protein/µg NPs
on Ade-PI-mal vs 0.9 µg protein/µg NPs on PI-mal) but did not significantly affect α2M or Fet A
binding.
27

Figure 3. Amount of protein bound to NPs (g protein/g NP), as measured by BCA assay after
incubation of NPs with protein in serum-free cell medium overnight at r.t., followed by
ultrafiltration to remove unbound protein (n = 3, error bars represent SEM. Student’s t test * p <
0.05, ** p < 0.01).
To confirm that proteins could covalently bind to maleimide moieties present on the NPs and to
discriminate between adsorbed and covalently linked proteins, Ellman’s protocol was used to
determine free thiol groups [53, 54]. If a NP is covalently bound to a protein, the number of free
thiols present on the protein should decrease, as the thiol groups are attached to the maleimide
moiety. Ellman’s reagent was used to quantify free cysteine on BSA, α2M or Fet A following
incubation with either PI, PI-mal, Ade-PI or Ade-PI-mal NPs (Figure S8). For all proteins, there
was significantly less free cysteine following incubation with PI-mal and Ade-PI-mal NPs
compared to incubation with PI and Ade-PI NPs respectively, indicating that covalent binding
occurred in the presence of maleimide. Even though there is likely a mix of adsorbed/linked
proteins at the surface of PI-mal and Ade-PI-mal NPs, their total amount was not significantly
different to that of only adsorbed proteins at the surface of PI and Ade-PI NPs.
Further confirmation of successful protein functionalization was obtained by measuring the
hydrodynamic diameter of NPs. It has been previously shown that NPs increase in size upon protein
functionalization and protein corona formation [86]. Therefore, the size of NPs with or without
28

protein functionalization was also measured after incubation in serum to form a corona, and
subsequent ultrafiltration to remove excess unbound serum proteins. For all NPs, both protein
functionalization and protein corona formation resulted in a significant increase in size, with the
latter causing a more dramatic increase (Figure 4). The size increase upon corona formation is likely
due to the adherence of multiple protein layers, containing proteins of diverse size bound directly to
the NP surface and to each other. In the case of almost every NP-protein combination, prior protein
functionalization prevented corona-associated size increases, and the size was comparable or
identical to protein-functionalized NPs not exposed to serum. This is indicative of functionalization
stability. Notably the ultrafiltration process itself caused a slight increase in PI (+36 nm), PI-mal
(+23 nm), and Ade-PI-mal (+9 nm) NPs without proteins, but PSD values remained comparable
before and after ultrafiltration (Figure S9). Indeed, the PSD values for most nanoparticle-protein or
nanoparticle-serum complexes did not significantly change compared to bare NPs. This indicated
that size changes are not due to aggregation of NPs upon protein exposure, and is in agreement with
studies showing increases in NP size upon protein binding [53, 87-89].
29