Translation of nanomedicines from lab to industrial scale synthesis: The case of squalene-adenosine nanoparticles

Article abstract of DOI:10.1016/j.jconrel.2019.06.040

A large variety of nanoparticle-based delivery systems have become increasingly important for diagnostic and/or therapeutic applications. Yet, the numerous physical and chemical parameters that influence both the biological and colloidal properties of nanoparticles remain poorly understood. This complicates the ability to reliably produce and deliver well-defined nanocarriers which often leads to inconsistencies, conflicts in the published literature and, ultimately, poor translation to the clinics. A critical issue lies in the challenge of scaling-up nanomaterial synthesis and formulation from the lab to industrial scale while maintaining control over their diverse properties. Studying these phenomena early on in the development of a therapeutic agent often requires partnerships between the public and private sectors which are hard to establish. In this study, through the particular case of squalene-adenosine nanoparticles, we reported on the challenges encountered in the process of scaling-up nanomedicines synthesis. Here, squalene (the carrier) was functionalized and conjugated to adenosine (the active drug moiety) at an industrial scale in order to obtain large quantities of biocompatible and biodegradable nanoparticles. After assessing nanoparticle batch-to-batch consistency, we demonstrated that the presence of squalene analogs resulting from industrial scale-up may influence several features such as size, surface charge, protein adsorption, cytotoxicity and crystal structure. These analogs were isolated, characterized by multiple stage mass spectrometry, and their influence on nanoparticle properties further evaluated. We showed that slight variations in the chemical profile of the nanocarrier's constitutive material can have a tremendous impact on the reproducibility of nanoparticle properties. In a context where several generics of approved nanoformulated drugs are set to enter the market in the coming years, characterizing and solving these issues is an important step in the pharmaceutical development of nanomedicines.

Full text of DOI:10.1016/j.jconrel.2019.06.040

Accepted Manuscript
Translation of nanomedicines from lab to industrial scale
synthesis: The case of squalene-adenosine nanoparticles
Flavio Dormont, Marie Rouquette, Clement Mahatsekake,
Frédéric Gobeaux, Arnaud Peramo, Romain Brusini, Serge Calet,
Fabienne Testard, Sinda Lepetre-Mouelhi, Didier Desmaële,
Mariana Varna, Patrick Couvreur
PII:
S0168-3659(19)30373-6
DOI:
Reference:
https://doi.org/10.1016/j.jconrel.2019.06.040
COREL 9845
To appear in:
Journal of Controlled Release
Received date:
Revised date:
Accepted date:
24 April 2019
27 June 2019
28 June 2019
Please cite this article as: F. Dormont, M. Rouquette, C. Mahatsekake, et al., Translation
of nanomedicines from lab to industrial scale synthesis: The case of squalene-adenosine
nanoparticles,
Journal
of
Controlled
Release,
https://doi.org/10.1016/
j.jconrel.2019.06.040
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.

ACCEPTED MANUSCRIPT
Translation of Nanomedicines from Lab to Industrial Scale Synthesis: The
Case of Squalene-Adenosine Nanoparticles
Flavio Dormonta, Marie Rouquettea, Clement Mahatsekakeb, Frédéric Gobeauxc, Arnaud Peramoa
Romain Brusinia, Serge Caletb, Fabienne Testardc, Sinda Lepetre-Mouelhia, Didier Desmaëlea, Mariana
Varnaa, Patrick Couvreura,* patrick.couvreur@u-psud.fr
aInstitut Galien Paris-Sud, CNRS UMR 8612, Université Paris-Sud, Université Paris-Saclay, 92296 Châtenay-
Malabry, France
bHOLOCHEM, Voie de l'Innovation, 27100 Val-de-Reuil, France
cCEA Saclay, CNRS UMR 3685, Université Paris-Saclay, 91191 Gif sur Yvette, France
*Corresponding author.
ABSTRACT
A large variety of nanoparticle-based delivery systems have become increasingly important for
diagnostic and/or therapeutic applications. Yet, the numerous physical and chemical parameters that
influence both the biological and colloidal properties of nanoparticles remain poorly understood. This
complicates the ability to reliably produce and deliver well-defined nanocarriers which often leads to
inconsistencies, conflicts in the published literature and, ultimately, poor translation to the clinics. A
critical issue lies in the challenge of scaling-up nanomaterial synthesis and formulation from the lab to
industrial scale while maintaining control over their diverse properties. Studying these phenomena
early on in the development of a therapeutic agent often requires partnerships between the public and
private sectors which are hard to establish.
In this study, through the particular case of squalene-adenosine nanoparticles, we reported on the
challenges encountered in the process of scaling-up nanomedicines synthesis. Here, squalene (the
carrier) was functionalized and conjugated to adenosine (the active drug moiety) at an industrial scale
in order to obtain large quantities of biocompatible and biodegradable nanoparticles. After assessing
nanoparticle batch-to-batch consistency, we demonstrated that the presence of squalene analogs
resulting from industrial scale-up may influence several features such as size, surface charge, protein
adsorption, cytotoxicity and crystal structure. These analogs were isolated, characterized by multiple
stage mass spectrometry, and their influence on nanoparticle properties further evaluated. We
showed that slight variations in the chemical profile of the nanocarrier’s constitutive material can have
a tremendous impact on the reproducibility of nanoparticle properties. In a context where several
generics of approved nanoformulated drugs are set to enter the market in the coming years,
characterizing and solving these issues is an important step in the pharmaceutical development of
nanomedicines.

ACCEPTED MANUSCRIPT
Keywords: squalene-adenosine nanoparticles, drug development, impurity profile, scaling-up
nanomedicines
ABREVIATIONS
SQAd Squalene-Adenosine
NBS N-Bromosuccinimide
HOBt 1- hydroxybenzotriazole hydrate
HATU 2-(1H-7-azabenzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate
DIIPEA Diisopropylethylamine
Et3N Triethylamine
MTT 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide
DMEM Dulbecco’s Modifidied Eagle’s Medium
DXT Dextrose
THF Tetrahydrofurane
MeOH Methanol
AcOEt Ethyl Acetate
NMR Nuclear Magnetic Resonance
MTBE Methyl tert-butyl ether
NPs Nanoparticles
DLS
KCl
Dynamic Light Scattering
Potassium Chloride
Fetal Bovine Serum
FBS
PBS Phosphate-buffered saline
ES-MS Electrospray Mass Spectrometry
TEM Transmission Electron Microscopy

ACCEPTED MANUSCRIPT
ATP Adenosine tri-phosphate
SDS Sodium dodecyl phospage
PAGE Polyacrylamide gel electrophoresis
INTRODUCTION
Nanostructured drug delivery systems have emerged over the last several decades as an important
field of academic research and are set to bring momentous advances to human health through clinical
and industrial developments. While a first generation of nanocarriers were successfully brought to
market, notably in the oncology field, new classes of therapeutic agents making use of polymers,
proteins, polysaccharides and other biomolecules have emerged for the treatment of numerous
diseases[1,2]. Unfortunately, these new types of nanocarriers are facing difficulties in their translation
to the clinics due to an absence of realistic industrial production, lack of reproducibility and insufficient
cost-effectiveness[3]. Part of the problem lies in the fact that nanostructured drug delivery systems
have grown into increasingly complex structures with poorly understood biological fates[4]. Size[5],
surface charge[6], morphology[7], drug release profiles[8] and interaction with blood components[9]
have all been shown to be key parameters that influence the biomedical outcome of nanocarriers in
vivo. Controlling these numerous parameters in a reproducible manner over intricate structures can be
problematic, as well as difficult to bring to industrial standards. Accordingly, recent studies have
highlighted the importance of developing simple and robust manufacturing methods that ensure the
proper formulation of nanomaterials in large scales[10,11].
Another critical issue lies in the challenge of scaling-up the synthesis of a nanocarrier’s constitutive
materials. Guidelines from the pharmaceutical industry in terms of good manufacturing practice and
pharmaceutical quality can be applied, but the nature and complexity of nanomaterials carry their
extra set of challenges. Indeed, while any scale-up of laboratory processes is difficult, nanomaterials
production is made more challenging by the fact that subtle variations in the synthesis and
manufacturing processes can result in significantly different products. The matter is further
complicated when one considers that carriers (e.g. polymers, lipids, dendrimers) often inherently do
not have a single pharmaceutical identity, but are constituted of a raft of different molecules which are

ACCEPTED MANUSCRIPT
difficult to individually characterize. This is contrary to conventional small-drug therapeutics, where the
active pharmaceutical ingredient is the main concern. While regulations are still unclear regarding the
plural identity of nanomedicines, special care should be taken when considering production of
nanomaterials. Namely, there must be robust quality control systems in place, validating batch-to-
batch consistency and biological equivalence. Identifying critical parameters that might affect the
reproducibility of nanomaterial properties early on and monitoring these parameters on multiple
batches is a good practice which can minimize the appearance of reproducibility issues later on[12].
Recently, squalenoylation (i.e squalene loaded with active molecules) has emerged as a simple, safe
and efficient method to produce nanoparticles encapsulating a vast range of therapeutic agents[13-
15]. Conjugating squalene, an endogenous, biocompatible and biodegradable lipid to a drug candidate
yields nanocarriers with high drug loading efficiency, low toxicity, and high blood residence time[16].
These nanoparticles have been used in pre-clinical models to treat a host of afflictions from cancer[17]
to ischemic stroke[18] and pain[19]. As with other nanomedicines, efforts now need to focus on the
translation from lab-scale experiments to semi-industrial batches, allowing eventual regulatory
approval for clinical trials.
In this article, through the example of squalene-adenosine (SQAd) nanoparticles, we wish to report on
the challenges encountered in the process of scaling-up nanomaterial synthesis by comparing two
batches, one synthesized at the gram scale in the laboratory, the other at the kilogram scale by an
industrial partner. We highlight specific parameters which needed to be closely monitored to ensure
batch-to-batch consistency. Notably, we demonstrate that scaling of lab scale synthetic processes can
induce varying impurity profiles which can have tremendous impact on the reproducibility of the
biological and physico-chemical properties of nanoparticle from batch-to-batch.
MATERIALS AND METHODS
Materials
Squalene, adenosine, N-Bromosuccinimide (NBS), 1-hydroxybenzotriazole hydrate HOBt, 2-(1H-7-
azabenzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate (HATU), tetrabutylammonium

ACCEPTED MANUSCRIPT
fluoride (TBAF), aluminium isopropoxide, propionic acid, triethyl orthoacetate, triethylamine (NEt3),
diisopropylethylamine (DIPEA), formic acid, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium
bromide (MTT), penicillin streptomycin solution, glutamine, norepinephrine, Claycomb’s Medium,
Dulbecco’s Modifidied Eagle’s Medium (DMEM) and dextrose (DXT) were purchased from Sigma-
Aldrich (France). Ultra-pure water was prepared using a MilliQ system (Millipore Corporation, Billerica,
MA). Reagents and HPLC grade solvents as ethanol, methanol, ethyl acetate, and acetonitrile were
provided by Carlo Erba (France). Ammonium acetate CH3CO2NH4 was from VWR.
Industrial Scale Synthesis of SQAd (Ind SQAd)
Synthesis of 3-bromo-2,6,10,15,19,23-Hexamethyl-tetracosa-6,10,14,18,22-pentaen-2-ol (squalene
bromohydrine) 2.
To a stirred solution of squalene 1 (1000 g, 2.4 mol) in THF (5 L), H2O (200 mL) was slowly added until
the solution became opalescent. To this mixture, THF (100 mL) was slowly introduced until solution
became clear again. N-Bromosuccinimide (520 g, 2.9 mol) was then added and the reaction mixture
was stirred 90 min at room temperature after which THF was removed and a liquid-liquid extraction
was performed with 500 mL of brine and 1000 mL of diethyl ether. The organic phase was separated
and the aqueous phase extracted with diethyl ether (3 x 800 mL). All the organic layers were combined,
dried over MgSO4, filtered and concentrated under reduced pressure. The resulting oily residue was
purified by flash chromatography over silica gel (cyclohexane/ethyl acetate : 8/1 v/v) to provide 2,3-
bromohydrine 2 (353 g, 29%) as a yellow oil. 1H NMR (300 MHz, CDCl3) δ: 5.25-5.10 (m,5H, HC=C(CH3)),
4.01 (dd,
J = 2.1, 11.3 Hz, 1H, CHBr), 2.32-2.35 (m, 1H), 1.95-2.19 (m, 18H, CH2), 1.77-1.88 (m, 1H), 1.70
(s, 3H, CH3), 1.63 (m, 15H, CH3), 1.37 (s, 3H, CH3), 1.36 (s, 3H, CH3).
Synthesis
of
2,3-epoxysqualene-2,2-dimethyl-3,7,12,16,20-pentamethylhenicosa-3,7,11,15,19-
pentaen-1-yloxirane (epoxysqualene) 3.
In a 10 L round bottom flask fitted with a mechanical stirrer and a calcium chloride guard tube,
squalene bromohydrine 2 (350 g, 0.69 mol) was solubilized in MeOH (5 L) and K2CO3 (2eq) was added
to this mixture. The reaction mixture was stirred 2 hours at room temperature and was then
concentrated in vacuo. The residue was treated with H2O and extracted with AcOEt (2 x 1 L). The

ACCEPTED MANUSCRIPT
combined organic layers were dried with MgSO4, filtered and the solvent removed under reduced
pressure to obtain a yellow oil (280 g, 95%) which was directly used in the next step without any
1
further purification. H NMR (CDCl3) δ: 5.25-5.10 (m,5H, HC=C(CH3)), 2.69 (t, J = 7.0 Hz, 1H, CH) 1.95-
2.19 (m, 20H, CH2); 1.68 (s, 3H, CH3) 1.60 (s, 3H, CH3); 1.58 (m, 12H, CH3) 1.28 (s, 3H, CH3); 1.24 (s, 3H,
CH3).
Synthesis of 2,6,10,15,19,23-Hexamethyl-tetracosa-1,6,10,14,18,22-hexaen3-ol 4.
To a stirred solution of 2,3-epoxysqualene 3 (280 g, 0.66 mol) in toluene (2.5 L) was added aluminium
isopropoxide (430.8 g, 1.98 mol). The resulting mixture was stirred under reflux for 16 h. After cooling
to room temperature, 1 N HCl was added until the whole solid was solubilized. The organic phase was
collected and the aqueous phase extracted with diethyl ether (4 x 500 mL). The combined organic
layers were washed with brine, dried over MgSO4, filtered and concentrated under reduced pressure.
The resulting oily residue was purified by flash chromatographic purification over silica gel (petroleum
1
ether/diethyl ether : 4/1 (v/v)) to provide allylic alcohol 4 (177.5 g, 63%) as a colourless oil. H NMR
(300 MHz, CDCl3) δ: 5.25–5.10 (m, 5H, HC=C(CH3)), 4.96 (m, 1H, =CH2), 4.86 (m, 1H, =CH2), 4.05 (t, J =
6.5 Hz, 1H, CHOH), 2.20–1.95 (m, 20H, CH2), 1.75 (s, 3H, CH3), 1.70 (s, 3H, CH3), 1.64 (s, 3H, CH3), 1.63
(s, 12H, CH3).
Synthesis of 4,8,12,17,21,25-Hexamethyl-hexacosa-4,8,12,16,20,24-hexaenoic acid ethyl ester
(squalene acetic ethyl ester) 5.
A stirred solution of allylic alcohol 4 (175 g, 0.41 mol) and propionic acid (3.5 g, 0.047 mol) in triethyl
orthoacetate (1 L) was heated under reflux for 3 h. After cooling to room temperature, triethylamine (7
g, 0.069 mol) was added and the mixture was concentrated under reduced pressure. The residue was
stirred for 5 min with 0.1 N HCl solution and extracted with diethyl ether (4 x 500 mL). The combined
organic phases were washed with brine, dried over MgSO4, filtered and concentrated under reduced
pressure to provide the crude ester 5 (202 g, quantitative) as a pale yellow oil which was directly used
in the next step without further purification. 1H NMR (300 MHz CDCl3) δ: 5.20-5.02 (m, 6H, HC=C(CH3)),
4.09 (q, J = 7.2 Hz, 2H, OCH2CH3), 2.40-2.20 (m, 4H, CO2CH2CH2), 2.15-1.85 (m, 20H, CH2), 1.66 (s, 3H,
CH3), 1.59 (s, 18H, CH3), 1.23 (t, J = 7.2 Hz, 3H, OCH2CH3).

ACCEPTED MANUSCRIPT
Synthesis of squalenylacetic acid 4,8,12,17,21,25-Hexamethyl-hexacosa-4,8,12,16,20,24-hexaenoic
acid (squalenylacetic acid) 6.
In a 10 L round bottom flask fitted with a mechanical stirrer, a reflux condenser and a calcium chloride
guard tube, was introduced ester 5 (200 g, 0.4 mol) solubilized 3000 mL of EtOH and 1000 mL of THF.
After stirring, LiOH,H2O (92.5 g, 2.2 mol) was added at room temperature. The mixture was heated to
reflux for 24 h. After cooling, the reaction mixture was concentrated under reduced pressure and
MilliQ water was added. The mixture was acidified with HCl 3N to pH = 3-4 and was then extracted by
MTBE (3 x 1 L). The combined organic extracts were dried over MgSO4, filtered and concentrated under
reduced pressure. The residue was purified by flash chromatography on silica gel
(dichloromethane/petroleum ether : 1/1 (v/v)) to provide squalenylacetic acid 6 (121 g, 65%) as a
1
yellow oil. H NMR (300 MHz CDCl3) δ: 5.15 -5.0 (m, 6H, HC=C(CH3)), 2.49-2.30 (m, 4H, C O2CH2CH2),
2.20-1.90 (m, 20H, CH2), 1.70 (s, 3H, CH3), 1.62 (s, 18H, CH3).
Synthesis of N-{9-2,3-bis[(tert-butyldimethylsilyl)oxy]-5-[(tertbutyldimethylsilyl)oxy]methyloxolan-2-
yl]-9H-purin-6-yl}-4,8,12,17,21,25-hexamethylhexacosa-4,8,12,16,20,24-hexaenamide(Tri-silylated
Squalene Adenosine) 7.
To a stirred solution of squalenylacetic acid 6 (120 g, 0.25 mol) in anhydrous dichloromethane (1 L) and
DMF (5 mL), was added under nitrogen HOBt (46 g, 0.3 mol), HATU (114 g, 0.3 mol), tris-O-silylated
adenosine (130.2 g, 0.25 mol) and DIPEA (9.35 g, 0.075 mol). The reaction mixture was stirred 72 hours
at room temperature and was then concentrated in vacuo. Aqueous sodium hydrogen carbonate was
added and the mixture was extracted with ethyl acetate (3 × 500 mL) and dichloromethane (1 × 300
mL). The combined extracts were washed with brine, dried on MgSO4, and concentrated under reduced
pressure. The crude product was purified by flash chromatography on silica gel eluting with 10 % ethyl
1
acetate in cyclohexane to give tri-silylated Squalene Adenosine 7 as a colourless oil (119 g, 45%); H
NMR (300 MHz, CDCl3) δ : 8.70 (s,1H, H8), 8.64 (s, 1H NHCO), 8.35 (s, 1H, H2), 6.10 (d, , J = 5.2 Hz, 1 H,
H1’), 5.20 (t, J = 6.4 Hz, 1 H, HC=C(CH3)), 5.20-5.10 (m, 5H, HC=C(CH3)), 4.67 (t, J = 4.7 Hz, 1 H, H2’), 4.3
(t, J = 4.7 Hz, 1 H, H2’), 4.15-4.11 (m; 1 H, H4’), 4.02 (dd, J = 11.2 Hz, J = 3.9 Hz, 1H, H5’), 3.78 (dd, J =
11.2 Hz, J = 2.7 Hz, 1H, H5’), 2.96 (t, J = 7.8 Hz, 2 H, O2CCH2CH2), 2.45 (t, J = 7.8 Hz, 2 H, O2CCH2CH2),
2.10-1.95 (m, 20 H, =C(CH3)CH2CH2), 1.67 (s, 6 H, C=C(CH3)2), 1.58 (s, 15 H, C=C(CH3)), 0.95 (s, 9 H, t-

ACCEPTED MANUSCRIPT
BuSi), 0.93 (s, 9 H, t-BuSi), 0.78 (s, 9H, t-BuSi), 0.15 (s, 3 H, CH3Si), 0.14 (s, 3 H, CH3Si), 0.10 (s, 6H,
CH3Si), −0.04 (s, 3 H, CH3Si), −0.24 (s, 3 H, CH3Si).
Synthesis
of
N-{9-2,3-dihydroxy-5-(hydroxymethyl)oxolan-2-yl]-9H-purin-6-yl}-4,8,12,17,21,25-
hexamethylhexacosa-4,8,12,16,20,24-hexaenamide (Squalene-Adenosine) 8.
To a solution of tris-O-silylated Squalene Adenosine 7 (100 g, 0.94 mol) in dry THF (1 L) was added TBAF
(1M solution in THF, 340 mL, 0.34 mol) and the mixture was stirred for 3h at room temperature. The
reaction was then quenched with brine and the mixture was extracted with ethyl acetate (3 × 500 mL).
The combined extracts were dried on MgSO4, and concentrated under vacuum. The crude product was
purified by chromatography on silica gel eluting with 5 % methanol in dichloromethane to give
1
Squalene-Adenosine 8 as a colourless oil (37.1 g, 55 %); H NMR (300 MHz, CDCl3) δ: 9,39 (broad s, 1 H,
NHCO), 8.41 (s, 1 H, H8), 8.21 (s, 1 H, H2), 5.98 (d, 1 H, J = 5.4 Hz, H1’), 5.32 (t, 1 H, J = 6.3 Hz,
HC=C(CH3)), 5.05-4.93 (m, 5 H, HC=C(CH3)), 4.61 (t, 1 H, J = 4.9 Hz, H2’), 4.18 (m, 1 H, H3’), 3.98 (m, 1 H,
H4’), 3.69 (dd, 1 H, J = 11.9 Hz, J = 3.7 Hz, H5’), 3.57 (dd, J = 11.9 Hz, J = 3.3 Hz, 1 H, H5’), 2.98 (t, 2 H, J
= 7.8 Hz, O2CCH2CH2), 2.45 (t, 2 H, J = 7.8 Hz, O2CCH2CH2), 2.11-1.93 (m, 20 H, =C(CH3)CH2CH2), 1.62 (s,
3 H, C=C(CH3)), 1.60 (s, 3 H, C=C(CH3)), 1.54 (s, 15 H, C=C(CH3))
Lab Scale Synthesis of SQAd (Lab SQAd)
SQAd was synthesized as previously reported[18] starting from 20 g of squalene and obtaining 700 mg
of SQAd for a total yield of 2.1 %.
HPLC-UV-ESI-MS analysis.
Chromatography Apparatus: A high-performance liquid chromatograph Waters Alliance 1695 (Waters,
Milford, MA), equipped with a Waters DAD 2996 photodiode array detector, Hewlett-Packard
computer with a Waters Empower 3 software and a Waters autosampler with a 50 µL loop was used,
with simultaneous detection at 220 nm and 270 nm. The spectra (detection wavelengths from 200-600

ACCEPTED MANUSCRIPT
nm) were recorded for all peaks. The HPLC system was coupled on line with a Waters TOF mass
spectrometer.
Columns and Mobile Phase: A Waters Xselect LC-18 column (2.1 x 150 mm, 3.5 µm) was used. The
solvents used were (A) 100% HPLC-grade acetonitrile (B) 0.1% formic acid in MilliQ water. Flow rate:
0.25 mL/min. Solvents and samples were filtered through a 0.45 µm Millipore filter type HA (Millipore
Corp., Bedford, MA).
HPLC Conditions: SQAd and SQAd analogs phase (called Ph1) were separated using a gradient over 10
min from 80% to 100% of A. Peak assignments were made based on spectral data.
Electrospray Mass Spectrometry (ESI-MS). Mass spectrometry was performed using electrospray MS.
The instrument was a Waters TOF LCT Premier mass spectrometer equipped with an ion spray
interface with a capillary voltage of 2800 V. The mass spectrometer was operated in the positive-ion
mode.
High Resolution HPLC-ESI-MSn analysis.
Chromatography Apparatus: A ThermoFisher Scientific (ThermoFisher, Waltham, MA) chromatograph
including a Dionex U-3000 RSLC system equipped with a WPS-3000 autosampler and a TCC-3000
column oven was used. The RSLC system was coupled on-line to a hybrid mass spectrometer LTQ-
Orbitrap Velos Pro (ion trap and orbital trap) from ThermoFisher.
Columns and Mobile Phase: A Macherey-Nagel Nucleodur 100-5 C18 EC column (2.1 x 125 mm) was
used. The solvents used were 50% HPLC-grade acetonitrile/ 50% HPLC-grade MeOH (solvent A) and
0.1% formic acid in MilliQ water (solvent B). Flow rate was 0.3 mL/min. Solvents and samples were
filtered through a 0.45 µm Millipore filter type HA (Millipore Corp., Bedford, MA).
HPLC Conditions: SQAd and SQAd analogs were separated using a solvent gradient over 10 min from
80% to 100% of solvent A.
High Resolution Mass Spectrometry: Mass Spectrometry Full Scan analysis was performed in the
Orbital Trap with a resolution of 30 000 in positive mode. The instrument was a ThermoFisher hybrid

ACCEPTED MANUSCRIPT
mass spectrometer LTQ-Orbitrap Velos Pro equipped with an electrospray ion source with a capillary
voltage of 3400 V.
MSn: MSn was performed in the LTQ ion trap with Collision Induced Dissociation (CID) at an energy of
35 (arbitrary unit). Fragment detection was performed in the orbitrap with a resolution of 7500.
Preparative HPLC.
Preparative HPLC separation of Ph1 from SQAd was performed by injecting 1 mL of a filtered solution
of 10 mg/mL Ind SQAd sample in ethanol onto a Waters Xselect LC-18 (30 × 150 mm, 2.1 µm) HR C-18
column. Elution was performed using Waters model 501 pumps to deliver a constant flow rate of 42
mL/min. The solvent system used was identical to prior HPLC experiments. SQAd was detected by
absorbance at 270 nm using a Waters UV DAD 2996 photodiode array detector. The first product
impurity phase (Ph1) was collected between 13.6 min and 15.3 min and SQAd was collected between
19 min and 21 min.
Peak Extraction.
Fractions of eluates containing compounds of interest were combined, filtered through a 0.22 µm
Teflon filter, and concentrated at 50 °C under 0.01 hPa vacuum on a rotary evaporator to obtain
purified compound.
Preparation and Physico-Chemical Characterization of NPs.
SQAd nanoparticles (SQAd NPs) were prepared by nanoprecipitation, as previously described[20,21]
(Supp. Figure S1). Briefly, for both Lab SQAd and Ind SQAd, SQAd bioconjugates were dissolved in 1 mL
of absolute ethanol (6 mg/mL) and added dropwise under moderate mechanical stirring to a 5% (w/v)
dextrose solution (DXT 5%).The ethanol was then completely evaporated using a Rotavapor (90 rpm,
40 °C, 42 mbar) to obtain an aqueous suspension of pure NPs (2 mg/mL). Ph1 nanoparticles were
obtained from the analog chromatographic fraction using the same protocol. NPs size (hydrodynamic
diameter) and surface charge (zeta potential) were measured using a Malvern Zetasizer Nano ZS 6.12
(173° scattering angle, 25 °C, Material RI: 1.49, Dispersant Viscosity RI: 1.330, Viscosity 0.8872 cP). For
the size measurements by dynamic light scattering (DLS), a good attenuator value (7−9) was obtained

ACCEPTED MANUSCRIPT
when suspending 50 μL of NPs in 1 mL of distilled water. The mean diameter for each preparation
resulted from the average of three measurements of 60 s each. For zeta potential measurements, 70
μL of NPs was dissolved in 2 mL of KCl 1 mM before filling the measurement cell. The mean zeta
potential for each preparation resulted from the average of three measurements in automatic mode,
followed by the application of the Smoluchowski equation. NP stability was assessed by measuring NP
size by DLS at different time intervals after initial formulation up to seven days.
Cryo-Transmission Electron Microscopy (cryo-TEM).
The morphology of SQAd nanoparticles was observed with cryo-transmission electron microscopy.
Drops of the solutions were deposited on electron microscope grids covered with a holey carbon film
(Quantifoil R2/2) previously treated with a plasma glow discharge. The excess liquid on the grids was
blotted out with filter paper, and the grids were quickly immersed in liquid ethane to form a thin
vitreous ice film. The whole process was performed using a Vitrobot apparatus (FEI Company).
Observations were conducted at low temperature (−180°C) on a JEOL 2010 FEG microscope operated
at 200 kV. Images were recorded with a Gatan camera.
Small-Angle X-ray Scattering
SAXS experiments were performed on two different set-ups. Lab SQAd was studied at the SWAXSLab in
CEA Saclay with a Xeuss 2.0 apparatus (Xenocs) equipped with a GeniX3D X-ray generator (λ = 1.542Å)
and a PILATUS3 1M 2D detector. The detector to sample distance was chosen to attain the 0.01-0.5 Å-1
q-range.
Averaging and data treatment were performed with PYSAXS software
(https://pypi.org/project/pySAXS/). The sample Ind SQAd was studied at the synchrotron SOLEIL (Saint-
Aubin, France) on the SWING beamline. Data were recorded at 12keV with a 4M bi-dimensional Eiger
detector in the chosen 0.014-1.9 Å-1 q-range. The data treatments were performed using FOXTROT
software. For each set-up, the samples were inserted in 1.5 mm diameter quartz capillaries sealed with
paraffin wax and analyzed with a counting time of 3600s and 250 ms for respectively the SWAXSLab
and SWING beamlines.
Spiking experiment.

ACCEPTED MANUSCRIPT
SQAd and Ph1 were separately dissolved in absolute ethanol (6 mg/mL). The SQAd and Ph1 ethanolic
solutions were mixed in varying ratios as to obtain different 0.1 %, 1 %, 5 %, 10 % (w/w) solutions of
Ph1 in SQAd. This organic solution was then nanoprecipitated as described above in a 5% (w/v)
dextrose solution.
Cell Culture.
MCEC (Mouse Cardiac Endothelial Cells, TebuBio, France) cells were grown in DMEM medium
supplemented with 10% FBS, 1% penicillin-streptomycin and 1% HEPES at 37 °C and 5% CO2. HL1
(Cardiac Muscle cells, Sigma, France) were grown in Claycomb’s medium supplemented with 10% FBS,
1% penicillin-streptomycin, 1% glutamine and 1% norepinephrine at 37 °C and 5% CO2. HepG2
(Hepatocellular Carcinoma, Sigma, France), HUVEC (Human Umbilical Vein Endothelial cell, ATCC,
France) and MiaPaca (Human Pancreatic Carcinoma Epithelial Cells, ATCC, France) cells were grown in
DMEM medium supplemented with 10% FBS and 1% penicillin-streptomycin at 37 °C and 5% CO2
atmosphere. For HL1 and MCEC cell lines, cells were divided to the tenth every 3-4 days. For MiaPaca,
HepG2 and HUVEC cells, cells were divided to the fifth every 3-4 days.
Cytotoxicity of SQAd NPs.
Cytotoxicity was evaluated using an MTT (3-[4,5-dimethylthiazole-2-yl]-2,5-diphenyltetrazolium
bromide) assay. For a typical 24 hours cytotoxicity study, cells were seeded in a 96 well plate at 8 000
cells/well (30 000 cells/cm2) for 24 hours. Subsequently, cells were treated with different
concentrations of nanoparticles ranging from 1 µM to 200 µM in 200 µL of complete cell medium. Cells
were incubated for 24 hours before the MTT assay was performed. For the MTT assay, cells were
washed with PBS, and incubated for 2 hours with medium containing 0.5 mg/mL of MTT. The medium
was removed after 2 hours and 100 µL of DMSO were added to each well before measuring
absorbance reading on a Perkin Elmer absorbance reader at 570 nm.
Rat Plasma Samples.
Rat blood samples were obtained from healthy male Sprague Dawley rats by cardiac puncture and
collected into heparinated collection tubes. Plasma was obtained by centrifuging the blood samples at
2000 rcf for 10 min. Plasma samples were pooled, split into 5 mL aliquots and stored at -80 °C until

ACCEPTED MANUSCRIPT
further use. For protein binding experiments, the aliquots were thawed at room temperature and then
incubated at 37 °C.
Protein Assay.
Plasma protein binding to SQAd nanoparticles was studied by incubating 250 µL of NP suspension (4
mg/mL) in DXT 5% with 250 µL of rat plasma. Incubation was carried on at 37 °C for 1 h to promote
protein adsorption. The amount of adsorbed proteins on SQAd NPs after incubation in plasma was
quantified using the Protein Assay reagent (Pierce, Thermo Scientific, Waltham, MA, USA) according to
manufacturer’s protocol. Briefly, after incubation, NP-protein complexes were centrifuged at 16 000 rcf
for 10 min to pellet NP- protein complexes. The supernatant was removed and the pellet was
resuspended in 500 µL of PBS. This procedure was repeated twice to remove any loosely bound
proteins until final suspension in 500 µL of PBS. A plasma aliquot not incubated with nanoparticles was
subjected to the same procedure as control. For the assay, 25 µL of each sample were placed in a 96-
well plate and 200 µL of Protein Assay working solution were added. The measures were performed in
triplicate. The plate was covered, mixed on an orbital plate shaker, and incubated at 37 °C for 30 min.
The absorbance of each sample, standard and blank was measured with a Perkin Elmer absorbance
reader at 570 nm. The protein concentration was calculated using interpolation with the standard
curve obtained from PBS solutions of Bovine Serum Albumin.
Protein Separation by 1D-PAGE.
Proteins were separated by 1D-PAGE according to Laemmli[22] with some modifications. Proteins were
desorbed from SQAd NPs by adding SDS-PAGE sample buffer to the pellet and heating the solution to
95 °C for 5 min. A Biorad Mini-PROTEAN® TGX Stain-Free™ 4%-15% Bis–Tris polyacrylamide gel was
used as separating gel. Electrophoresis was carried out at a constant voltage of 300 V in
electrophoresis buffer, until the dye front reached the lower end of the gel (20 min). Separated
proteins were imaged using a Biorad ChemiDoc Imaging system and analyzed with Biorad Image Lab
software. For quantification, total lane fluorescence was measured and for each detected band the

ACCEPTED MANUSCRIPT
band fluorescence count was normalized over total lane fluorescence yielding the normalized
percentage of fluorescence count (NpFlC %).
RESULTS AND DISCUSSION
I. Scale-up and Reproducibility of NPs properties.
a) Synthesis of SQAd
At the onset of this study was a need to scale up the synthesis of the SQAd bioconjugates for use in
multi-lab pre-clinical trials evaluating SQAd efficacy in different animal models. Starting from squalene,
SQAd was first synthesized at a lab scale in limited amounts (ie. 100 mg) with a total yield of ~2 % (Lab
SQAd) and then by an industrial partner in large scale using the same synthetic process (Ind SQAd).
SQAd synthesized from industrial batches was obtained in a higher total yield (~5 %) through industrial
optimization of the synthesis process. The synthetic route to SQAd 8 is illustrated in Figure 1A. Because
of its lack of functional groups, conjugation to squalene 1 can only be achieved after functionalization
of the squalene aliphatic chain. This functionalization process starts with the selective
hydroxybromination of the terminal double bond with NBS in wet THF according to the method of Van
Tamelen[23]. The bromohydrine 2 pathway shows a proclivity for the formation of the terminal 2,3-
bromohydrine which has been attributed in part to conformational changes of the squalene chain in a
water containing solution. This hypothesis is strongly supported by the numerous observations that
most epoxidation reactions performed in apolar organic solvents using peroxy acids or
dimethyldioxirane give a mixture of internal epoxidation products[24,25]. The chemoselectivity of
NBS/H2O for the terminal double bond was, however, not perfect and the crude material needs to be
thoroughly purified by silica gel chromatography before engaging the bromohydrine in the next step.
Nevertheless, this synthetic path was considered preferable to other oxidative methods as the process
didn’t require the involvement of toxic reagents or heavy metals which would further complicate
regulatory approval. As depicted in Figure 1A, once the crucial terminal functionalization was obtained,
2,3-oxydosqualene was easily derived from the 2,3-bromohydrine by potassium carbonate treatment
(b). The one isoprene-unit homologation was then achieved by sequential aluminium isopropoxide

ACCEPTED MANUSCRIPT
Figure 1: A. Scheme of SQAd synthesis: a) Squalene and NBS in THF/H2O, 29%. b) 2 and K2CO3 in MeOH, 95%. c)
3, aluminium-isopropoxide in Toluene, reflux, 63%. d) 4, propionic acid in triethyl orthoacetate, reflux. e) 5,
LiOH,H2O in EtOH/THF, reflux, 65%. f) 6, HOBt, HATU and DIPEA in DCM, 45%. g) 7 and TBAF in THF, 55%. B. DLS

ACCEPTED MANUSCRIPT
measurement of NPs sizes for Lab SQAd and Ind SQAd batches. For measurements performed in cell medium,
NPs were dispersed at a concentration of 50 g/mL or 150 g/mL in complete DMEM medium. C. Zeta-potential
measurements of Lab SQAd and Ind SQAd. D. Stability of SQAd NPs in water determined by DLS monitoring of
NP size.
ring opening reaction (c) followed by Claisen-Johnson rearrangement with triethylorthoacetate (d)[26].
After saponification, the obtained squalenyl acetic acid was coupled to 2’,3’,5’-trisilylated adenosine
using HATU/HOBt as coupling agent with moderate yield. After extensive purification, the silyl
protecting groups were removed upon TBAF treatment. Although the lipidic nature of squalene
precluded crystallization of the intermediates, a satisfactory purity ( 98% by HPLC) was achieved after
optimization of the chromatographic conditions for both Lab SQAd and Ind SQAd. NMR analysis of both
batches yielded spectra consistent with SQAd (Supp. Figure S2). Nanoparticles of both Lab SQAd and
Ind SQAd were then formulated to evaluate the influence of industrial scale-up on NPs properties.
b) NP formulation and physico-chemical characterization
SQAd NPs were prepared by nanoprecipitation[18] of an ethanolic solution of SQAd in a 5% (w/v)
aqueous dextrose solution. Once formulations of different SQAd batches were obtained, size and
polydispersity index (PDI) of the NPs were measured by dynamic light scattering (DLS) (Figure 1B). The
resulting measurements showed substantial differences in the colloidal behavior of formulations
originating from different SQAd batches. Indeed, small variation in the mean size were observed (83
nm/ 99 nm for respectively Ind SQAd and Lab SQAd) while larger differences in the mean zeta potential
were measured (from -25mV to -15mV for respectively Ind SQAd and Lab SQAd) (Figure 1C). Both
formulations were stable for at least 7 days in water as measured by DLS (Figure 1D), while in cell
culture medium Ind SQAd NPs were found to be more stable than Lab SQAd NPs. This difference was
most probably due to their stronger surface charge which inhibited NPs aggregation in a solution of
high ionic strength such as cell medium. In addition, cryo-TEM imaging (Figures 2B-C) showed
substantial differences concerning the NPs internal structure. Lab SQAd demonstrated apparent
internal ordering while Ind SQAd produced NPs with less apparent structure. To further investigate this
observation, the internal structure of SQAd NPs was studied by Small-Angle X-ray Scattering (SAXS)
(Figure 2A). For Lab SQAd, the SAXS pattern displayed three Bragg reflections at q0 = 0.091 Å-1, q1 =

ACCEPTED MANUSCRIPT
0.158 Å-1 (q0 ) and q2 = 0.182 Å-1 (2q0). These peaks indexed as the reflections of an inverse two-
dimensional hexagonal phase with lattice parameter a = 78.0 Å. For Ind SQAd, however, SAXS data
showed only one smooth Bragg peak without high order reflections, in agreement with a lower
ordering of the internal structure, as observed by cryoTEM. The low ordering of Ind SQAd NPs
suggested that some constituents of the SQAd NPs batches may influence NPs colloidal behavior and
warranted further studies to evaluate if these physico-chemical differences may impact the biological
behavior of NPs.
A
B
C
Figure 2: A. Small-angle X-ray diffraction patterns NPs recorded at room temperature. B. Selected Cryo-TEM
image of Lab SQAd NPs obtained by nanoprecipitation of an ethanolic solution in DXT 5%. C. Selected Cryo-TEM
image of Ind SQAd NPs obtained by nanoprecipitation of an ethanolic solution in DXT 5%.

ACCEPTED MANUSCRIPT
c) Cytotoxicity of Lab SQAd and Ind SQAd batches
In this study, HepG2, HUVEC, HL1 and MiaPaca cells lines were chosen as in vitro models of the
influence of SQAd to account for cell subtype-specific variations. The cytotoxicity of different batches
of SQAd NPs was determined after incubation with the various cell lines by performing an MTT assay
(Figure 3). Ind SQAd batches were consistently found to be more toxic than lab SQAd NPs with IC50
values in the 100 µM range while Lab SQAd did not induce a strong cytotoxicity even at concentrations
of 200 µM. This tendency was confirmed on the four tested cell lines and with NPs incubation times
ranging from 24 to 72 hours. Also, the presence of dextrose in the formulations did not to influence the
cytotoxicity of SQAd NPs (Supp. Figure S3).
Figure 3: MTT assays of Lab SQAd and Ind SQAd on HUVEC, HepG2, MiaPaca and HL1 cell lines. Viability was
expressed as % of viability of non treated cells as measured by formazan dye absorbance at 570 nm.
Concentrations ranging from 5 µM to 200 µM were studied over 24 h, 48 h and 72 h of incubation.

ACCEPTED MANUSCRIPT
Squalene is an endogenous precursor of cholesterol which is non toxic to cells. Once functionalized into
squalenylacetic acid 6, it has been shown to induce a moderate inhibition of cell proliferation in certain
models with IC50 concentrations in the 180-200 µM ranges on HUVEC and MiaPaca cells[27].
Adenosine, on the other hand, has a more controversial effect on cell proliferation[28-30]. While it can
induce apoptosis and cell death in certain types of cardiomyocytes[31] and neutrophils[32], adenosine
provides important cytoprotective functions in the heart and brain tissue during ischemia, hypoxia or
ischemia reperfusion[33,34]. A number of studies have reported that cell specific purinergic receptors
contribute to the dual effects of adenosine on cell viability[35-37]. SQAd has been shown not to be a
direct agonist of these purinergic adenosine receptors but rather to act through accumulation of
adenosine intracellularly[38]. In any case, these subtle cytotoxic and cytoprotective effects of
adenosine can hardly explain the gross differences displayed in the cytotoxicity profiles of the different
SQAd batches observed in this study.
d) Plasma protein adsorption
Current understanding of nanoparticles interactions in vivo suggests that the protein corona (ie surface
adsorbed proteins) formed on nanoparticles greatly influences their biological fate[39,40].
Consequently, plasma protein adsorption on Ind SQAd NPs and Lab SQAd NPs was evaluated after
incubation in rat plasma for 1 h at 37 °C. NP-protein complexes were isolated from excess plasma,
washed thoroughly, and the amount of adsorbed proteins was quantified for each formulation. Lab
SQAd formulations afforded an average of 37.4 µg/mL (i.e. 93.5 µg/mg NPs) adsorbed proteins, while
Ind SQAd afforded an average of 12.8 µg/mL (i.e. 32 µg/mg NPs) adsorbed proteins (Figure 4A). These
values refer to those proteins which strongly interacted with the nanoparticles surface, the so called
“hard corona”, which essentially confer the NPs their biological identity. Although previous studies
have shown that strongly negative NPs had a tendency to adsorb higher amount of proteins[41], the
main driving force in protein-NP interactions is not simply electrostatic but revolves also around
complex colloidal and biochemical interactions which characterize the NPs in vivo[42,43]. For instance,
it is possible that the crystal lattice of the Lab SQAd nanoparticles allowed greater access and
interaction with the small proteins of the plasma. To better identify the influence of the industrial

ACCEPTED MANUSCRIPT
scale-up on the nature of the adsorbed proteins at the surface of the NPs, stain-free 1D-PAGE was
performed on the hard corona isolated from NPs of both Lab SQAd and Ind SQAd batches (Figure 4B).
1D-PAGE resolves the corona proteins according to their molecular weight and afforded a basic
evaluation of the different fractions of protein adsorbed to the NPs and their relative amounts.
B
A
50
40
30
20
10
0
d
d
a
A
A
m
Q
Q
s
S
S
a
l
d
b
P
a
n
I
L
C
%
N p F lC
5 0
0
1 0 0
2 5 0 -1 4 0 kD a
1 2 5 -9 5 kD a
L a b S Q Ad
8 5 -7 0 kD a
6 0 -4 5 kD a
4 0 -3 0 kD a
3 0 -2 0 kD a
2 0 -1 5 kD a
In d S Q Ad
Figure 4: A. Concentration of protein corona isolated from different batches of SQAd NPs as measured by Pierce
protein assay. For plasma control, the same procedure as for NPs corona evaluation was used but without the
presence of nanoparticles. B. 1D-PAGE from two independent experiments of Lab SQAd and Ind SQAd protein
corona. Molecular weight ladder was migrated with a pre-stained Biorad Precision Plus sample which is shown
alongside the stain-free gel electrophoresis. C. Normalized percentage of fluorescence count for proteins found
in the hard corona of Lab SQAd and Ind SQAd nanoparticles. Proteins were grouped following molecular weight
expressed in kDa.

ACCEPTED MANUSCRIPT
The relative abundance of each protein fraction on the surface of the nanoparticles was evaluated by
Normalized Percentage of Fluorescence Count (%NpFlC, equivalent to the fraction of total fluorescence
for all proteins found in the corona). The protein corona compositions shown in Figure 4C varied in
subtle ways between the two SQAd batches. Mainly, low molecular weight protein fractions appeared
to be significantly reduced in Ind SQAd and could account for the smaller quantities of protein corona
found on Ind SQAd NPs. On the contrary, proteins of larger molecular weight (above 60 kDa) were
found to be present consistently among the tested nanoparticles samples (lane profiles can be found in
Supp. Figures S4-S5). While the determinants of corona composition are often underappreciated and
hardly characterized, they are critical in determining the subsequent fate and effects of NPs in both in
vitro and in vivo situations. Studies have shown that variations of less than 10 % in the makeup of a
NP’s protein corona could result in an alteration of the pharmacokinetic profiles and
biodistribution[40].

ACCEPTED MANUSCRIPT
II. Identification and characterization of impurities
a) Chromatography studies
Our previous physico-chemical and in-vitro studies seemed to point to the presence of formerly
undetected impurities, especially in the Ind SQAd batches. As such, a more sensitive HPLC-MS
approach of the different SQAd samples was carried out in order to identify and characterize eventual
impurities responsible for batch-to-batch discrepancies. The results, shown in Figure 5, indicate that
other than SQAd (peaks 3 and 4) a small quantity of detectable impurity phase, from now on named
Ph1, corresponding to a mass of elution peaks around 20.8 min (peaks 1,2 and 5), was present in Ind
SQAd batches. UV integration at 270 nm showed Ph1 peaks present at 1-2% in Ind SQAd and hardly
detectable (<0.5%) in Lab SQAd. ES-MS data showed Ph1 as being composed of a mixture of SQAd
analogs with major components having measured masses of 734.48 g/mol and 750.47 g/mol. Two
minor impurities were also found in shoulder 6 for which a MS study can be found in Supp. Figure S6.

ACCEPTED MANUSCRIPT
Figure 5: UV-HPLC spectra of Ind SQAd and Lab SQAd. UV spectra was recorded at 270 nm, corresponding to the
maximum absorption peak of SQAd. At the end of the HPLC system, ES mass spectrometry was performed and
the results reported in Table 1.
Table 1: Spectrometry data for Lab SQAd and Ind SQAd HPLC peaks.
Measured
mass
(m/z)
UV
Integration
(%)
tR
(min)
Peak
no
Proposal
ions
Elemental
composition
Identification
[M-H]+
[M-H]+
[M-H]+
[M-H]+
[M-H]+
1
2
3
4
5
SQAd (OH)2
SQAd OH
SQAd
20.22
20.87
25.00
24.73
20.52
750.47
734.48
718.49
718.49
734.48
C42H63N5O7
C42H63N5O6
C42H63N5O5
C42H63N5O5
C42H63N5O6
0.5
1.5
98
SQAd
99.5
0.5
SQAd OH
This corresponded to SQAd (m/z = 718.49) with one or two additional oxygen atoms which we inferred
to be -OH substituents present on the aliphatic squalene chain of SQAd. To elucidate this further, a
collision induced dissociation (CID) MSn study of these compounds was carried out (Table 2 and Supp.
Figures S7-S9). CID of SQAd in the conditions used for this study usually yields a single major product
ion with m/z = 586.44, corresponding to the fragmentation of a ribose residue from the adenosine
moiety. When CID was carried out on Ph1, it showed that both compounds with m/z = 734.48 and m/z
= 750.47 first demonstrated the usual pathway of parent ion fragmentation observed with secondary
allylic alcohols, corresponding to a β-elimination of the hydroxyl group, resulting in CID products of m/z
= 716.47 and m/z = 732.47 respectively (Figure 6A).
Table 2: Collision induced dissociation of the different SQAd compounds detected by MSn.
MS1 Product
(m/z)
MS2 Product
(m/z)
Parent ion
(m/z)
Compound
SQAd
718.49
586.44

ACCEPTED MANUSCRIPT
SQAd OH
SQAd OH2
734.48
716.47
602.44
602.44
750.47
584.43
732.47
714.47
618.44
618.44
600.43
582.42
A

ACCEPTED MANUSCRIPT
B
Figure 6: A. Presumptive structures for the detected SQAd analogs and their collision-induced product ions are
shown. Upper box shows product ions for single –OH substituted SQAd analogs, lower box corresponds to
double –OH substituted products. Exact masses were calculated with the ChemDraw software and fitted the
experimental data. B. Chromatograms of the different SQAd analogs obtained during HPLC-ESI-MSn
This was followed by the expected product ions induced by SQAd ion fragmentation and loss of the
ribose residue. Of note, chromatograms obtained from the high resolution HPLC-ESI-MSn study
additionally showed that slightly different elution peaks corresponded to compounds having identical
molecular masses (Figure 6B). For single –OH substituted compounds SQAdOH, we found five elution
peaks corresponding to the five possible substitution products over the available isoprenyl subunits,
while double –OH substituted compounds SQAdOH2 gave 10 elution peaks corresponding to the
combination of two substitutions out of a set of five available isoprenyl subunits. This indicated that
the SQAd analogs most probably existed as a mixture of isomers in Ph1, suggesting that the oxidative
side-reaction which led to the additional hydroxyl groups occurred unselectively over all five isoprenyl
subunits. Although, in the absence of an exact structural determination, the mechanism of formation
of these minor byproducts remains speculative, it can be suggested that they result from the ring-
opening reaction with aluminum isopropoxide of internal epoxides produced by over oxidation with

ACCEPTED MANUSCRIPT
NBS. Previous studies have shown that the main side product in the Van Tamelen procedure is the
-bis-bromohydrine[44,45]. During the purification process, using chromatography on silica gel, only
the main side product corresponding to the -bis-bromohydrine could be separated, while minor
less polar internal bromohydrines are much more difficult to eliminate and could remain with the main
product. Once formed, these highly hindered allylic alcohols would not act as substrates in the Claisen-
Johnson transposition and thus would remain as minor components in the bulk main product.
b) Cytotoxicity of hydroxylated SQAd impurities
To elucidate the impact of Ph1 impurities on the observed batch-to-batch discrepancies in NPs
properties, a preparative HPLC separation (HPLC-Prep) was performed to isolate Ph1 from SQAd.
Practically, two series of experiments were performed (Figure 7). First, NPs were prepared from Ph1
only (Figure 7A) and tested as such for their cytotoxicity and physico-chemical properties. When
prepared by nanoprecipitation, Ph1 yielded stable NPs and DLS data showed a strongly negative zeta
potential of -35 mV and sizes of ca. 100 nm (Figure 7B). Further studies focused on the biological
effects of Ph1 NPs which were found to be dramatically cytotoxic, with IC50 values around 5 µM,
compared to isolated SQAd which barely induced cytotoxicity in the tested concentration ranges
(Figures 7D-E). The influence of minor impurities on cytotoxicity was also studied and data reported in
Supp. Figure S10. Later, a series of additional experiments were performed in which pure SQAd
fractions from HPLC prep experiments were spiked with increasing concentrations of Ph1 (from 0.1 %
w/w to 10 % w/w). Remarkably, the amount of Ph1 was proportionally linked to cytotoxicity in SQAd
nanoformulations (Figure 7F).

ACCEPTED MANUSCRIPT
Figure 7: A. Representative scheme of spiking experiments, where isolated Ph1 and pure SQAd were combined
in varying ratios. B. Zeta-potential measurements of SQAd and Ph1 NPs. C. DLS size measurement of SQAd and
Ph1 NPs. For measurements performed in cell medium, NPs were dispersed at 150 g/mL in complete DMEM
medium. D-E. MTT assays of SQAd and Ph1 NPs on MCEC and HepG2 cell lines. Cell viability was measured as %
viability of non treated cells. F. Cell viability after 24 hours incubation with 100 µM of SQAd eluate NPs or NPs of
SQAd spiked with increasing concentrations of Ph1 from 0.1%-10% w/w.
These experiments confirmed that the Ph1 impurity, even when present in small quantity, was
responsible for both the biological and physico-chemical alteration of SQAd NPs properties. It is not
surprising that Ph1 forms a stable colloidal phase when nanoformulated, as other terpenoid derivatives

ACCEPTED MANUSCRIPT
of squalene have been shown to do so[46]. Ph1 yields NPs demonstrating low zeta potential and high
toxicity which can help to explain the properties observed with Ind SQAd batch NPs. We postulate that
the observed cytotoxic effect of Ph1 is likely due to the aberrant squalene chains altering the cell
membrane integrity. Indeed, squalene strongly interacts with lipid bilayers because of its structural
similarity with cholesterol as well as through binding to LDL receptors on the cell surface[21,46,47].
Additionally, it is also possible that hydroxylated SQAd analogs can participate in oxidative stress
signaling through their similitude with lipid peroxidation by-products.
Ultimately, nanoparticle-based delivery systems are unique structures capable of accumulating
bioactive molecules inside tissues and cells. This makes them a class of materials that is especially
sensitive to the presence of residual manufacturing components or variations in purity from batch-to-
batch[48-50]. This has often lead to misjudgments concerning NP toxicities and poor reproducibility of
results[51,52]. Here, even though special precautions were taken in using biocompatible reagents and
formulation strategies, scaling-up the manufacturing process resulted in varying quantities of analogs
of the lipid carrier chain which in turn were responsible for an unexpected cytotoxicity of SQAd NPs.
These findings highlight the complexity of nanotechnology and the escalating level of expertise
required to successfully generate safe and efficient nanomedicines. Not only does it require a thorough
understanding of the chemistry and biocompatibility of compounds used for the initial formulation of
the nanoparticles, but also full comprehension of the synthesis process and its implications on the
aforementioned properties. In a context where the reproducibility of nanoformulations is put into
question[53] and several generics of approved nanoformulated drugs are set to enter the market in the
coming years[54], special care and regulatory oversight over these issues is of outmost importance.
CONCLUSION
In this study, we showed that the scale-up of the bromohydrine synthetic pathway to SQAd can alter
the impurity profile of the squalene bioconjugate which results in critical variations in SQAd NPs
properties. Here, close analogs of the carrier lipid chain were found to be highly cytotoxic and
strenuous to separate from SQAd bulk by conventional chromatographic methods. Surprisingly, these
minor impurities had a strong impact on the colloidal properties, internal structure and biological

ACCEPTED MANUSCRIPT
properties of SQAd NPs. Identifying these issues was crucial in finally obtaining batches of SQAd that
yield nanoparticles with consistent properties, which has now been achieved by the industrial partner.
Failure to properly characterize a material has afflicted the nanotechnology field for many years,
regularly leading to misjudgments about the properties of nanoparticles. The matter is often made
worse in the case of nanocarriers because of the inherent polydispersity of their physico-chemical
properties as well as their plural chemical identity. Identifying which parameters are most important
for establishing the therapeutic and biological equivalence of two nanoparticle formulations is
therefore important. As the field matures and a growing number of products seek regulatory approval,
an especially careful approach in terms of toxicity and pharmaceutical development should be adopted
in the maturation process of these innovative materials. Ultimately, understanding and solving these
issues is a crucial part of the advancement of nanomedicines, and one which cannot be left for the
industry to deal with alone.
Funding:
The authors gratefully acknowledge the financial support from the 7th EuroNanoMed-II call for
proposals, project NanoHeart n°ANR-16-ENM2-0005-01. This work was further supported by la
Fondation pour la Recherche Médicale (FRM) grant number ECO20160736101. Cryo-TEM observations
were made thanks to “Investissements d’Avenir” LabEx PALM (ANR-10-LABX-0039-PALM).
Acknowledgments:
The authors wish to thank Karine Leblanc from the BioCIS platform for her expert technical advice on
running and performing chromatography experiments as well as Audrey Solgadi from UMS-IPSIT
SAMM platform for her help in designing and performing the MSn experiments as well as helpful
corrections in the manuscript. Jéril Degrouard is thanked for his technical help and discussions relating
to cryo-TEM images.
Declaration of Interest: