Discovery of novel berberine derivatives with balanced cholinesterase and prolyl oligopeptidase inhibition profile

Article abstract of DOI:10.1016/j.ejmech.2020.112593

Berberine, a naturally occurring compound, possesses an interesting multipotent pharmacological profile potentially applicable for Alzheimer's disease (AD) treatment. In this study, a series of novel 22 berberine derivatives was developed and tested in vitro. Berberine core was substituted at position 9-O of its aromatic ring region. All the hybrids under the study revealed multi-targeted profile inhibiting prolyl oligopeptidase, acetylcholinesterase and butyrylcholinesterase highlighting 4a, 4g, 4j, 4l and 4s possessing balanced activities in the micromolar range. The top-ranked candidates in terms of the most pronounced potency against POP, AChE and BChE can be classified as 4d, 4u and 4v, bearing 4-methylbenzyl, (naphthalen-2-yl)methylene and 1-phenoxyethyl moieties, respectively. In vitro data were corroborated by detailed kinetic analysis of the selected lead molecules. 4d, 4u and 4v were also inspected for their potential to inhibit aggregation of two abberant proteins in AD, namely amyloid beta and tau, indicating their potential disease-modifying properties. To explain the results of our study, we carried out docking simulation to the active sites of the respective enzyme with the best berberine derivatives, along with QSAR study. We also investigated compounds’ potential permeability through blood-brain barrier by applying parallel artificial membrane permeation assay and addressed their cytotoxicity profile.

Full text of DOI:10.1016/j.ejmech.2020.112593

Journal Pre-proof
Discovery of novel berberine derivatives with balanced cholinesterase and prolyl
oligopeptidase inhibition profile
Katerina Sobolova, Martina Hrabinova, Vendula Hepnarova, Tomas Kucera, Tereza
Kobrlova, Marketa Benkova, Jana Janockova, Rafael Dolezal, Lukas Prchal, Ondrej
Benek, Eva Mezeiova, Daniel Jun, Ondrej Soukup, Jan Korabecny
PII:
S0223-5234(20)30565-1
DOI:
https://doi.org/10.1016/j.ejmech.2020.112593
EJMECH 112593
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 16 May 2020
Revised Date: 15 June 2020
Accepted Date: 15 June 2020
Please cite this article as: K. Sobolova, M. Hrabinova, V. Hepnarova, T. Kucera, T. Kobrlova,
M. Benkova, J. Janockova, R. Dolezal, L. Prchal, O. Benek, E. Mezeiova, D. Jun, O. Soukup,
J. Korabecny, Discovery of novel berberine derivatives with balanced cholinesterase and prolyl
oligopeptidase inhibition profile, European Journal of Medicinal Chemistry, https://doi.org/10.1016/
j.ejmech.2020.112593.
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. Please note that,
during the production process, errors may be discovered which could affect the content, and all legal
disclaimers that apply to the journal pertain.
©
2020 Elsevier Masson SAS. All rights reserved.

BChE
AChE
POP
amyloid beta
tau protein
Berberine
derivatives

Discovery of novel berberine derivatives with balanced cholinesterase and prolyl
oligopeptidase inhibition profile
1,#
1,2,#
1
1
Katerina Sobolova , Martina Hrabinova , Vendula Hepnarova , Tomas Kucera , Tereza
1,2
2
2
2,3
2
Kobrlova , Marketa Benkova , Jana Janockova , Rafael Dolezal , Lukas Prchal , Ondrej
2
2
1,2
1,2,*
1,2,*
Benek , Eva Mezeiova , Daniel Jun , Ondrej Soukup , Jan Korabecny
1
Department of Toxicology and Military Pharmacy, Department of Military Medical Service
Organisation and Management; Faculty of Military Health Sciences, University of Defense,
Trebesska 1575, 500 05 Hradec Kralove, Czech Republic
2
Biomedical Research Centre, University Hospital Hradec Kralove, Sokolska 581, 500 05
Hradec Kralove, Czech Republic
3
Department of Chemistry, Faculty of Science, University of Hradec Kralove, Rokitanskeho 62,
5
00 03 Hradec Kralove, Czech Republic
#
equal contribution
*
corresponding authors: Ondrej Soukup: Phone: +420 495 833 447; E-mail:
ondrej.soukup@fnhk.cz; Jan Korabecny: Phone: +420 495 833 447; E-mail:
jan.korabecny@fnhk.cz

ABSTRACT
Berberine, a naturally occurring compound, possesses an interesting multipotent
pharmacological profile potentially applicable for Alzheimer´s disease (AD) treatment. In this
study, a series of novel 22 berberine derivatives was developed and tested in vitro.
Berberine core was substituted at position 9-O of its aromatic ring region. All the hybrids
under the study revealed multi-targeted profile inhibiting prolyl oligopeptidase,
acetylcholinesterase and butyrylcholinesterase highlighting 4a, 4g, 4j, 4l and 4s possessing
balanced activities in the micromolar range. The top-ranked candidates in terms of the most
pronounced potency against POP, AChE and BChE can be classified as 4d, 4u and 4v, bearing
4
-methylbenzyl, (naphthalen‐2‐yl)methylene and 1‐phenoxyethyl moieties, respectively. In
vitro data were corroborated by detailed kinetic analysis of the selected lead molecules. 4d,
u and 4v were also inspected for their potential to inhibition aggregation of two abberant
4
proteins in AD, namely amyloid beta and tau, indicating their potential disease-modifying
properties. To explain the results of our study, we carried out docking simulation to the
active sites of the respective enzyme with the best berberine derivatives, along with QSAR
study. We also investigated compounds’ potential permeability through blood-brain barrier
by applying parallel artificial membrane permeation assay and addressed their cytotoxicity
profile.
KEYWORDS: acetylcholinesterase; Alzheimer´s disease; amyloid beta; butyrylcholinesterase;
in silico; multi-target directed ligands; prolyl oligopeptidase; quantitative structure-activity
relationship; tau protein

1.
INTRODUCTION
Alzheimer´s disease (AD) is a type of dementia that causes a decline in intellectual
and functional abilities [1]. As of 2019, the global prevalence of AD is estimated to be as high
as 50 million, foreseeing that the number of AD patients can exceed 152 milion by 2050 [2].
One of the greatest known risk factors is increasing age, since the majority of people
suffering from AD are older than 65. From a neurobiological point of view, AD is a chronic
neurodegenerative disorder of the brain characterized by loss of neurons and synapses in
the medial temporal lobe and neocortical association areas. The atrophy of these brain
regions manifests as memory impairment, difficulty in learning, loss of language skills, not
managing the daily self-care, disorientation and personality changes [3].
AD can be considered as a multifactorial and complex disorder with prevailing deficits in
cholinergic system, changes in prolyl oligopeptidase (POP; E.C. 3.4.21.26) activity, elevated
levels of reactive oxygen species (ROS), presence of aberrant proteins like amyloid-beta
peptide (Aβ) forming neuritic plaques and intracellular neurofibrillary tangles composed of
hyperphosphorylated tau (τ) protein. That all leads to neuronal death and clinical
manifestations of this disease [4–6]. One of the typical phenomenon associated with AD is
cellular oxidative and nitrosative stress, which is responsible for dysregulation of nucleic
acids and proteins formation, lipid peroxidation and glycoxidation [7]. Oxidative stress (OS) is
the culprit of metabolic, metal, mitochondrial and cell-cycle abnormalities. Mitochondria are
regulators of cell death, and therefore mutations in their DNA together with OS contribute
to aging [8,9]. The intervention strategies comprise the use of effective antioxidants, up-
regulation of the endogenous antioxidative defense system, and administration of anti-
inflammatory agents as well. Apart from that, it is also well established that monoamine
oxidases are involved in the production of hydrogen peroxide; thus, their inhibitors may
ameliorate damages associated with OS [10,11].
Current palliative treatment is based on the administration of acetylcholinesterase
(AChE; E.C. 3.1.1.7) inhibitors as the most pronounced pathology in AD is associated with
cholinergic neuronal loss. There has also been confirmed interconnection between disrupted
cholinergic system and the formation of intracellular neurofibrillary tangles or senile plaques
[
1,12]. For instance, Inestrosa group found that AChE acts as a pathological chaperone
inducing a conformational transition of Aβ [13]. Generally, the cholinergic hypothesis claims
that the lack of the acetylcholine (ACh) neurotransmitter in the cortex is responsible for

cognitive decline. Both AChE and butyrylcholinesterase (BChE; E.C. 3.1.1.8) are responsible
for the breakdown of ACh in the synapses [14]. The progression of AD goes hand in hand
with AChE activity decrease, which is counterbalanced at the same time by BChE increase
[
14,15]. Therefore inhibition of both cholinesterases (ChEs) can elevate ACh levels and thus
mitigate AD symptoms.
There are a number of experimental targets that popped-up recently as the promising
ones, which are involved in the pathophysiology of the disease, and their targeting might
result in complex action in the disease cascade. One such target is POP enzyme, the cytosolic
serine peptidase. POP is implicated in physiological processes such as maturation and
degradation of peptide hormones and neuropeptides, such as substance P, thymosin β4,
thyrotropin-releasing hormone, vasopressin, and angiotensin. POP inhibitors (POPi) may
improve memory impairment by blocking the metabolism of these substrates. Therefore,
POP has been proposed as a potential therapeutic target for several disorders as,
rheumatoid arthritis [16,17], cystic fibrosis [18], α-synucleinopathies [19], AD [20], autism
spectrum disorder [21], cirrhosis [22], mania and schizophrenia [23], and high POP activity
was observed in proliferating cells [24]. Surprisingly, suppressed expression of POP by an
antisense technique and by specific POPi revealed an inverse correlation between inositol
1
, 4, 5-triphosphate (Ins(1, 4, 5)P ) concentration. Furthermore, decreased activity of POP
3
amplifies substance P mediated stimulation of Ins(1, 4, 5)P3, which is the intracellular second
messenger that binds to its receptor located in the membrane of the endoplasmatic
2+
reticulum and induces release of Ca which may have an impact on cognitive
enhancements. These findings can explain a relationship between an intracellular enzyme
such as POP and the neuropeptides, because Ins(1, 4, 5)P is a central molecule in the
3
signaling cascade of neuropeptides [25]. Promising results were reported for some POPi in
clinical trials. However, these were terminated mostly due to unfavorable pharmacokinetic
profile of the compounds [26].
Searching for new AD therapeutic options remains an elusive task. Given the
complexity of the disorder, applying a multi-target drug discovery approach might deliver
more effective therapeutics [27–29]. Indeed, this strategy foresees to target different
enzymes/receptors at the same time that may ultimately better address the needs to find a
novel and effective cure. One such option represents the chemical modification of natural
compounds to improve their pharmacological activities. Berberine (BBR; Fig. 1), a quaternary

ammonium protoberberine alkaloid, has been used in traditional Chinese and Ayurvedic
medicine possessing a broad spectrum of pharmacological properties. Some of the BBR
analogs have already been investigated in clinical trials [30].
Within this study, we inspected the structure-activity relationship (SAR) in the series
of novel BBR derivatives substituted in position 9-O with different aromatic rings. We tested
the novel compounds for the inhibition of ChEs and POP enzymes, their potency to
counteract OS and we predicted their capacity to cross the blood-brain barrier (BBB). For the
highlighted derivatives, we also performed anti-amyloid and anti-tau studies, and predicted
their binding modes in the active sites of the respective enzymes.
2.
DESIGN
The nature of AD assumes that the therapeutic intervention can be successfully
pursued by small molecules hitting multiple targets relevant to the disorder simultaneously.
Berberine (BBR) is a natural isoquinoline alkaloid with an intrinsic multipotent profile [7].
Indeed, BBR shows antioxidative, anti-proliferative, anti-inflammatory, amyloid-β (Aβ)
peptide level-reducing, monoamine oxidase, and cholinesterase inhibitory activities [31–33].
It has also been described that BBR improves cognitive impairment by inhibiting abnormal
Aβ aggregation and τ hyperphosphorylation [34]. Moreover, it ameliorates OS and
neuroinflammation [35]. Its neuroprotective properties were also confirmed in metal-
induced neurotoxicity animal models of AD [36]. BBR displayed dose-dependent POP
inhibition properties with IC50 = 145 μM [37]. Inhibition of both enzymes (AChE and POP)
potentiates resultant of cognitive effect whereas POP inhibition leads, due to enhanced
Ins(1, 4, 5)P
3
levels, to similar mechanism as signalization at muscarinic ACh receptor M
1
[38].
Figure 1. Chemical structure of berberine with numbering of the core scaffold.

BBR is endowed with a broad pharmacologic profile, on one hand; on the other hand,
it is hampered by relatively low permeability and bioavailability due to its hydrophilic
character that limits its systematic administration [30]. For these reasons, structural
modifications increasing BBR lipophilicity, mostly in position 9-, are pursued in order to
address these drawbacks [39–42]. From the wealth of literature, several SAR were outlined.
For example, SAR analysis in 9-N-alkyl-substituted BBR derivatives indicated higher
antioxidant and Aβ anti-aggregation properties [43]. Furthermore, attachment of lipophilic
substituent to BBR in the region of methylenedioxy substituted aromatic ring, and methoxyl
moiety at C-10 indicated increased hypoglycemic activity [44]. The introduction of
aminomethyl groups into 12-position of berberrubine improved the insulin-resistant reversal
activity and stimulated glucose transport activity, which is potentially applicable for type 2
diabetes mellitus (T2DM) [45]. Indeed, a growing body of epidemiological studies suggest
that people with T2DM are at a higher risk of developing AD [46]. For cholesterol-lowering
and anti-diabetic agents, it have been observed that methylenedioxy functionality in BBR
orchestrates binding to β-cell membranes [47–50]. Accordingly, BBR and its derivatives may
reduce the risk factors associated with AD, i.e. high levels of glucose and cholesterol.
Building on the findings above, in this study, we are exploring SAR in 9-subsitited BBR
derivatives. Into this position, we attached different lipophilic substituents to enhance
permeability through cell membranes plausibly. In parallel, 9-N substituted BBR derivatives
synthesized by Shan and colleagues [43] showed promising results as antioxidants; AChE,
BuChE and Aβ aggregation inhibitors but 9-O substitution, to the best of our knowledge, was
not tested for the possible cognitive enhancing properties. Herewith, we designed and
synthesized a new family of compounds based on nucleophilic substitution at 9-O- position
of BBR core.
3.
RESULTS AND DISCUSSION
3.1
Chemistry
BBR derivatives were synthesized following a chemical procedure depicted in Scheme
1. Initially, BBR (1) chloride underwent demethylation at position 9- in a vacuum (15 mbar)
at 205 °C for 2 hours which was clearly visible by a color change from yellow to dark red
forming berberrubine (2; 77% yield) [51]. Berberrubine was then treated with potassium
carbonate in anhydrous MeCN at 80 °C for 30 min followed by the addition of appropriate

benzyl bromide (3a-3v) and the mixture was refluxed for 1 h at 80 °C. Only reagent 3b was
dissolved in a mixture of anhydrous solvents THF/MeCN (1:1) because of its low solubility in
sole MeCN. To generate 4v, we used 2-(bromomethyl)naphthalene (3v) under the same
conditions. After cooling, the reaction mixture was filtered and washed successively with
MeCN (50 mL) and MeOH (50 mL). Column chromatography was used for purification to
produce 4a-4v in 8-91% yields. All of the final compounds 4a-4v showed analytical and
spectroscopic data in good agreement with their structures. Structural characterization
1
13
1
13
involved melting points, H and C NMR spectra, and LC−MS analysis. All the H and C NMR
spectra can be found at Supplementary information.

Scheme 1. General approach for the synthesis of 9-O-substituted BBR derivatives 4a-4v.
Reaction conditions: i) 205 °C, 15 mbar, 2 h; ii) K
2
CO (5 eq.), anh. MeCN (or THF/MeCN (1:1)
3
for 3b), 80 °C, 1.5 h.
3.2
Cholinesterase inhibition
To determine the therapeutic potential of berberine derivatives (4a-4v) with
particular emphasis for AD, their AChE and BChE inhibitory activities were assayed by the

modified spectrophotometric method described by Ellman et al. using human recombinant
AChE (hAChE) and human serum BChE (hBChE), respectively [52]. Results are expressed as
IC50 values and preference for hAChE is determined by the selectivity index (SI). These data
are summarized in Table 1.
Table 1. In vitro results of BBR derivatives 4a-4v and reference compound BBR
Compounds
POP IC50 ±aSEM
μM)
IC50 hAChE ± SEM IC50 hBChE ± SEM
SI for hAChE b
(hBChE/hAChE)
a
a
(
(μM)
(μM)
BBR
142 ± 21
25.4 ± 2.8
79.1 ± 8.7
71.1 ± 7.9
10.7 ± 1.2
53.3 ± 6.1
58.7 ± 8.4
19.4 ± 4.4
59.6 ± 5.5
38.9 ± 4.9
21.2 ± 3.0
34.0 ± 2.5
24.7 ± 1.3
67.0 ± 3.0
39.4 ± 1.4
52.7 ± 2.2
47.6 ± 1.4
36.5 ± 3.0
40.4 ± 1.0
27.4 ± 1.3
40.5 ± 0.9
28.5 ± 1.7
29.5 ± 1.3
17.4 ± 4.1
5.0 ± 0.1
7.3 ± 0.4
10.4 ± 0.2
75.8 ± 9.1
4.2 ± 0.3
2.5 ± 0.1
3.3 ± 0.4
2.7 ± 0.2
1.7 ± 0.1
0.99 ± 0.04
2.6 ± 0.1
1.45 ± 0.04
1.9 ± 0.1
2.5 ± 0.1
1.22 ± 0.03
2.6 ± 0.1
1.7 ± 0.1
3.5 ± 0.1
1.3 ± 0.1
5.3 ± 0.2
1.5 ± 0.1
0.79 ± 0.02
47.2 ± 3.3
8.0 ± 0.4
14.1 ± 0.8
21.5 ± 1.3
5.8 ± 0.2
2.6 ± 0.1
3.7 ± 0.1
9.6 ± 0.4
4.9 ± 0.1
4.0 ± 0.1
8.0 ± 0.2
2.9 ± 0.1
6.9 ± 0.2
6.1 ± 0.3
6.3 ± 0.2
3.6 ± 0.1
4.9 ± 0.1
11.7 ± 0.3
7.4 ± 0.1
9.4 ± 0.2
4.3 ± 0.1
1.26 ± 0.02
2.7
1.6
1.9
2.1
0.08
0.6
1.5
2.9
1.8
2.3
8.1
1.1
4.8
3.2
2.5
3.0
1.9
6.8
2.1
7.0
0.8
0.8
7.3
4a
4b
4c
4
4
d
e
4f
4g
4h
4i
4j
4k
4l
4m
4
4
4
4
n
o
p
q
4r
4s
4t
4u
4v
5.8 ± 0.1
a
b
Results are expressed as the mean of at least three independent experiments; Selectivity for hAChE is
determined as ratio IC50(hBChE)/IC50(hAChE)
All the novel derivatives 4a-4v displayed hAChE inhibition ability in micromolar to the
sub-micromolar range. BBR, taken as a reference, displayed moderate inhibition ability with
IC50 = 17.4 µM, which is roughly under other literature data [53]. The most active compound
4v displayed high inhibitory potency with IC50 value 0.79 µM, which roughly corresponds to
tacrine, formerly used for the AD treatment [54]. From the SAR analysis, it can be deduced
that derivatives either unsubstituted or bearing methyl in a different positions (4a-4d) of

benzyl ring negatively affected the inhibition power decreasing the IC50 values from a single
unit up-to two-digit micromolar concentration (IC50 = 5 - 76 µM). No other SAR can be drawn
when taking into consideration either i) the position of substituents or ii) electron-
donating/electron-withdrawing group characteristics. Comparing BBR derivates reported
herein with simplified analogues containing core N-benzylphenethylamine moiety reported
by Roselli et al., almost identical IC50 values were found for AChE inhibition [55]. In the study
by Huang et al. berberine derivates turned out to be more potent and more selective for
AChE inhibition [56]. Nevertheless, the data cannot be compared because of the different
origins of AChE and BChE species.
Since the levels of BChE are up-regulated in the advanced stages of AD, and BChE is
also capable of compensating for the decreased activity of AChE, BChE emerged as a backup
therapeutic option [14,57]. Accordingly, all the newly developed BBR derivatives 4a-4v
exerted BChE inhibition ability with IC50 values in the micromolar range (1.26 - 21.5 µM), all
being more potent than parent BBR (IC50 = 47.2 µM). The top-ranked compound was 4u
(IC50 = 1.26 µM) having 37.5 times amplified potency compared to BBR. Similarly to AChE, no
clear SAR could not be established given the nature of attached aromatic moiety.
In general, most of the BBR derivatives are hAChE selective inhibitors having SI values
˃
1.0 (Table 1). The highest SI score revealed 4-nitrophenyl derivative 4j being 8.1 times
more selective in inhibiting hAChE. On the contrary, 4d revealed an intriguing hBChE
selective profile with SI value for hBChE = 13.1.
3.3
Prolyl oligopeptidase inhibition
POP has been shown to participate in several aspects of the central nervous system
(
CNS) like learning, memory and mood, the features that are associated with AD [58]. It has
been discussed that reduced POP activity correlated to the tau pathology and severity of the
disorder [59]. Some of the POP inhibitors have been evaluated in clinical trials to reverse
and/or attenuate memory loss [26]. It has been proposed that the extract of Rhizoma
coptidis containing 23% of BBR as a major alkaloid was a potent POP inhibitor. The study also
disclosed that IC50 value of isolated BBR was 145 ± 19 µM which is in accordance with our
data (Table 1) [60]. All the novel BBR derivatives demonstrated IC50 values ranging from 10
to 79 µM corroborating that the 9-O aryl substitution is essential for activity increment.
Analogue 4d bearing 4-methylbenzyl appendage emerged as the most potent analogue from

the series (IC50 = 10.7 µM). From the obtained data, it can be roughly estimated that more
active derivatives (IC50 < 30 µM) favor position 4- at attached benzyl moiety with electron-
withdrawing groups like NO
2
or CF (4j and 4s, respectively), but also small electron-donating
3
groups like methyl (4d) are well tolerated in this position.
3.4
Kinetic assay
Kinetic study was performed to describe the interactions of the compounds 4s and 4u
with hAChE and hBChE. Inhibition kinetics was determined from velocity curves that were
measured at several concentrations of tested compounds. The type of enzyme inhibition and
corresponding kinetic parameters (K and Ki’) were determined using nonlinear regression
i
analysis. Results for each type model of inhibition (competitive, non-competitive,
uncompetitive and mixed) were compared by sum-of-squares F-test. Statistical analysis
showed noncompetitive inhibition for 4s with hAChE and mixed type of inhibition for 4u with
BChE (p ˂ 0.05), which is in line with Lineweaver–Burk plot, used for better visualization of
obtained data (Graph 1).
Graph 1. Steady state inhibition of cholinesterase substrate hydrolysis by compound 4s (hAChE) and 4u (BChE)
at different concentrations. Lineweaver−Burk plots of initial velocity at increasing substrate concentrations
(
hAChE: 0.1563 - 1.250 mM; BChE: 2.5 - 20.0 mM) are presented. Lines were derived from a linear regression of
the data points.

The intersection of lines is located on the x-axis for 4s and above the axis for 4u,
which corresponds with reversible binding mode to both free enzyme and enzyme-substrate
complex. Apparent Vmax decreased with an increasing concentration of 4s and K
unchanged, indicating that it has an affinity for both the free hAChE and enzyme-substrate
complex. Compound 4s binds with higher affinity to the free BChE (K
was reduced at higher concentration of inhibitor while the K was slightly increased. Both
m
remained
i
< Ki’), apparent Vmax
m
inhibitors interact with enzyme allosteric peripheral anionic site (PAS). This interaction
causes conformational changes of the cholinesterase, yielding the change of its active site. A
K
i
value of 0.456 ± 0.041 nM was measured for 4s on hAChE and a K
i
value of 8.34 ± 1.99 nM
and Ki’ of 75.6 ± 16.7 nM were determined for 4u against BChE, respectively.
3.5
Molecular modeling studies
To further understand the activity enhancing effect of aryl attachment at 9-O-
position of BBR core, we performed molecular modeling studies using X-ray structures of
hAChE (PDB ID: 4EY6), hBChE (PDB ID: 4BDS) and human POP (PDB ID: 3DDU) [61–63]. All the
proteins were downloaded in high resolution solved at 2.4 Å, 2.1 Å and 1.6 Å for AChE, BChE
and POP, respectively.
Both enantiomers, namely (R)-4v and (S)-4v, were selected to elucidate which isomer
presumably contribute more to activity increment, and both of them were also compared
with reference BBR (Fig. 2). BBR spans the cavity gorge of AChE contacting esteratic site
flanked by catalytic triad residues (His447, Ser203 and Glu202) forming hydrogen bond
between hydroxyl Ser203 and oxygen from 1,3-dioxolane moiety (2.4 Å; Fig. 2 A, B). BBR also
established T-shaped π-π interaction with Tyr337 (3.7 Å) and other hydrophobic contacts
with Phe338 and Trp86 in the catalytic active site (CAS) of the enzyme. Quaternary nitrogen
displayed salt bridge with carboxylic oxygen from Asp74. At the entrance of AChE, BBR is
engaged in several hydrophobic interactions including π-π stacking with Phe297, and Tyr341,
van der Waals contact with Trp286, and also hydrogen bond between hydroxyl from Tyr124
and 9-methoxy group from BBR (2.2 Å). (S)-4v (Fig. 2 E, F) revealed similar arrangement to
BBR in the AChE active site; however, interaction with Ser203 is missing. Attachment of (S)‐
1
‐phenoxyethyl moiety yielded in π-alkyl contact with Tyr341, and π-π stacking with Trp286
and Tyr72. (R)-4v adopted close configuration like parent BBR (Fig. 2 C, D). Indeed, typical
hydrogen bond between Ser203 and oxygen from 1,3-dioxolane moiety (2.8 Å), π-π contact

with Tyr337 (3.6 Å), favorable orientation between quaternary charge of BBR and Asp74 and
other hydrophobic contacts in PAS region with Phe297, Tyr124, Phe338 and other amino
acids, are all observed. Additionally, attachment of (R)‐1‐phenoxyethyl provided other
favorable interaction like π-alkyl contact between methyl group and Tyr341, anion-π
interaction with Asp74 and π-π displaced stacking with Tyr72 and Trp286. To conclude,
missing interactions in (S)-4v-hAChE complex like hydrogen bond with Ser203 and also
anion-π contact with Asp74 might be designated as the culprit for higher affinity of (R)-4v
over (S)-4v, however, direct proof-of-concept from in vitro measurement is undoubtedly
required to confirm the docking output.
Figure 2. Docking of BBR (A, B), (R)-4v (C, D), (S)-4v (E, F) to the active site of hAChE (PDB ID: 4EY6)
[61]. Generally to A, C, and E: BBR, (R)-4v and (S)-4v are shown in green, salmon and orange, respectively.

Important amino acid residues are highlighted in dark blue, catalytic triad residues in yellow. Apparent ligand-
enzyme interactions are shown as dashed lines; distance is measured in Å. The rest of the enzyme is displayed
as a grey transparent cartoon. Figures A, C and E were created with the PyMOL Molecular Graphics System,
Version 2.0 Schrödinger, LLC. 2D figures (B, D, F) highlights the interaction of different nature between each
ligand and specific amino acid residues. Figures B, D and F were generated by Discovery Studio 2016 Client
software.
Top-ranked hBChE inhibitor 4u and reference BBR were selected to inspect structural
topology of these two ligands in hBChE active site (Fig. 3). BBR (Fig. 3 A, B) is settled in the
vicinity of catalytic triad residues (Ser198, Glu197 and His438) displaying π-π interaction
between His438 and ring A of BBR. Moreover, there is a strong π-π stacking with Trp82 from
CAS and other aromatic regions of BBR. Other crucial interaction can be classified as either
hydrogen bonding between methoxy oxygen and Gln119, or hydrophobic contacts with
aliphatic amino acids like Asp70, Thr120, Ser198 and Ser287, as well as aromatic residue
Phe329. It is well known, that BChE active site is larger in comparison with AChE and thus
can accommodate bulkier substrates [64]. Bearing this in mind, such knowledge pave the
way for 4u bearing bulky (naphthalen‐2‐yl)methylene moiety. Indeed, 4u revealed
completely different pose to BBR in hBChE active site (Fig. 3 C, D). This accounts for parallel
π-π stacking between Trp82 (3.8 - 4.0 Å) and (naphthalen‐2‐yl)methylene unit, engagement
of Tyr332 by π-π interaction with C and D aromatic rings from BBR core, anion-π attractive
forces with Asp70, and also hydrogen bond formation with hydroxyl group from Ser287.
Thus, all these findings justify our design considerations and rationalize our findings from in
vitro.

Figure 3. Docking of BBR (A, B) and 4u to the active site of hBChE (PDB ID: 4BDS) [62]. Generally to A
and C: BBR and 4u are shown in green and purple, respectively. Important amino acid residues are highlighted
in dark blue, catalytic triad residues in yellow. Apparent ligand-enzyme interactions are shown as dashed lines,
distance is measured in Å. The rest of the enzyme is displayed as grey transparent cartoon. Figures A and C
were created with the PyMOL Molecular Graphics System, Version 2.0 Schrödinger, LLC. 2D figures (B and D)
highlights the interaction of different nature between each ligand and specific amino acid residues. Figures B
and D were generated by Discovery Studio 2016 Client software.
BBR and 4d as the highlighted POP inhibitor were further subjected to molecular
docking studies to reveal their arrangement in POP active site (Fig. 4). The active site of
human POP, defined by a catalytic triad, is composed of the amino acid residues Ser554,
His680 and Asp641, and is located at the interface of the two domains in the cavity gorge
[65]. The active site may be further divided into three regions, named as S1, S2 and S3. The
S1 site of the protein is narrow and can accommodate proline like moieties and the
nucleophilic appendages attached to proline like rings. This substitution is known to enhance
the activity of inhibitors because it enables interaction with Ser554. The S2 site with Asp641
favors positive charge whereas the third site S3 is flanked with amino acids like Phe173,
Met235, Cys255, Ile591, and Ala594 therefore hydrophobic groups as a part of bulky ligands
can be lodged there [66]. BBR (Fig. 4 A, B) is located in the vicinity of catalytic triad displaying

π-π contact with His680 and van der Waals attractive forces with Ser554. Ring A is implicated
in the classical π-π stacking with Phe476 (4.8 Å) and Trp595 (4.3 Å). Cys255 play an
important role in ligand placement by showing π-sulfur and π-alkyl interactions with C and D
rings, respectively. 10-Methoxy moiety is anchored by hydrogen bonds to Arg252 (2.1 and
2.3 Å). The most remarkable finding is that positively charged nitrogen of BBR exerted
unfavorable positive-positive interaction with Arg643 which is presumably responsible for its
relatively low inhibitory potency observed in vitro. On the contrary, 4d is placed outside the
catalytic triad contacting only Ser554 by van der Waals forces (Fig. 4 C, D). 4-Methylbenzyl
attachment revealed several hydrophobic contacts to Cys255, Trp595 and Phe476.
Interestingly, another Arg residue, namely Arg643 displayed hydrogen bond with 10-
methoxy group. This trend was also observed for BBR (Fig. 4 A, B) with Arg252. Phe173 can
be highlighted as the key-mediator residue in 4d-POP complex orchestrating the topology of
BBR core scaffold by anchoring it to A, C and D aromatic regions. Notably, unfavorable
interaction between 4d and Arg643 is missing. Taken also into account all the
aforementioned, these interactions can be mostly responsible for high POP potency of 4d.
Figure 4. Docking of BBR (A, B) and 4d to the active site of human POP (PDB ID: 3DDU) [63]. Generally
to A and C: BBR and 4d are shown in green and grey, respectively. Important amino acid residues are

highlighted in dark blue, catalytic triad residues in yellow. Apparent ligand-enzyme interactions are shown as
dashed lines, distance is measured in Å. The rest of the enzyme is displayed as grey transparent cartoon.
Figures A and C were created with the PyMOL Molecular Graphics System, Version 2.0 Schrödinger, LLC. 2D
figures (B and D) highlights the interaction of different nature between each ligand and specific amino acid
residues. Figures B and D were generated by Discovery Studio 2016 Client software.
3.6 Quantitative structure-activity relationship (QSAR) analyses
QSAR analyses were performed on PM3 geometrically optimized structures of 4a - 4v
in Schrodinger 2020-1 employing its tools for chemoinformatics and machine learning. The
first investigated, deep chemistry convolutional neural network QSAR modeling (DC-CNN
QSAR) based on processing molecular graphs by TensorFlow/Keras libraries, unfortunately,
provided no statistically significant SAR when the input dataset of 22 compounds was
randomly split into training and test sets in 75:25 ratio. The low predictivity of DC-CNN QSAR
2
models, in terms of negative cross-validated coefficients of determination Q for the
prediction in the test set, was mainly caused by the insufficient size of the dataset. No
significant improvement was achieved by the Atom-based QSAR analyses. Better results
were achieved by the AutoQSAR function, which applied a kernel partial least regression
(
kPLS) on a set of various molecular descriptors utilizing an algorithm for selecting optimal
molecular descriptor subsets. As in the kPLS all the input variables are projected in
reproducing kernel Hilbert space by a Gaussian transform (Eq. 1), kPLS is able to recognize
nonlinear features in the data.
ꢉ
ꢆ
ꢇꢈ
ꢅ
ꢉ
ꢊ
ꢇꢈ
ꢃꢄ
ꢁꢂ
(Equation 1)
ꢀ
The kernel function Ki,j replaces the scalar products of X matrix of molecular
descriptors in the PLS algorithm, taking dij argument as Euclidian distance between
descriptors i and j. The nonlinearity of the kernel is additionally modulated by the parameter
ij
σ .
The AutoQSAR program automatically tested hundreds of models and various setting
and selected the best scoring ones. In the final kPLS QSAR models, only one latent factor was
used. The best statistical kPLS QSAR models for the values of log IC50 (POP), log IC50 (hAChE),
and log IC50 (hBChE) are briefly described in Table 2.
Table 2. Statistical overview of kPLS QSAR models found for the studied compounds.

2
a
b
c
2 d
2
e
R
Target
S.D.
RMSE
Q
Q (MW)
POP
0.7160
0.8460
0.7738
0.1170
0.1841
0.1390
0.1015
0.1053
0.1358
0.7425
0.8801
0.5440
-0.0387
0.5204
0.3535
hAChE
hBChE
a
Coefficient of determination for the training set (training and test samples were divided randomly in 75:25
b
c
ratio); standard deviation of prediction for the training set; root-mean-square error for the test set
d
e
prediction; coefficient of determination for the test set prediction; coefficient of determination for the test
set prediction if only molecular weight is used as a predictor (the above statistical criteria definitions were
adopted from the Schrödinger manual).
According to the statistical criteria given in Table 2, all three QSAR models are statistically
significant and can be used for representing the biological properties of the studied
compounds. Thanks to the incorporation of fingerprints in the matrix of molecular
descriptors, the most influencing atoms in the structures can be determined by pseudo-beta
coefficient analysis. Structural features that likely improve or deteriorate the biological
activity of the studied targets are outlined in Fig. 5.
Figure 5. Analysis of kPLS QSAR models with respect to molecular fingerprint contributions. Molecular
features with a positive effect on IC50 are colored in blue, features with a detrimental impact on IC50 (e.g.
lowering its value) are colored in red shadows.

From the analysis of fingerprint contributions in kPLS QSAR models (Fig. 5), it is obvious that
there are only subtle structural differences controlling whether a substructure weakens or
enhances the biological activity. For instance, the change of methyl group position from para
to meta switches the impact of the methylbenzyloxy function on the biological activity
against hAChE dramatically. Noteworthy, the same structural feature that weakens the
biological activity against hAChE (4d) induces substantial enhancement of the biological
activity against POP. More details on fingerprint contributions can be found in
Supplementary Information (S2-S4). In general, it would be very challenging to intuitively
design improved inhibitors of POP and hAChE/hBChE in such variable structure-activity
space, but the developed kPLS QSAR models could serve in this case to predict the biological
activity by ligand-based virtual screening with satisfying ability.
3.7 Cytotoxicity
The effect on cell proliferation of BBR derivatives using the Chinese hamster ovary cell
line (CHO-K1) is presented in Table 3 [67]. MTT (3-[4,5-dimethylthiazol-2-yl]-2,5 diphenyl
tetrazolium bromide) assay was applied to reveal the cytotoxic effect [68]. It is evident that
aryl substitution of BBR scaffold at 9-O- noticeably affected cytotoxicity resulting in higher
toxic species. This trend nicely correlates with lipophilicity increment defined by calculated
logP values (Table 3). Accordingly, 4o as the most lipophilic BBR derivative containing
3
,4-dichlorobenzyl moiety is the most toxic species from the subset. The position of each
specific substituent does not play any significant role, i.e. no cytotoxicity differences were
observed within small subsets like 4b-4d (methyl substitution), 4g-4i (bromine attachment),
and 4r with 4t (methoxy derivatives). Unsubstituted derivative (4a), or analogues containing
methyl (4b-4d) or polar groups (4p, methyl ester; 4q, cyano group) displayed higher cell
viability, i.e. lower cytotoxicity profile. Derivative 4q was the least toxic in the family of
newly developed BBR with IC50 value lying in the same range as BBR.
Table 3. Cytotoxic potential of BBR and novel derivatives 4a-4v determined using CHO-K1 cell lines.
c
Compounds
CHO-K1 cell line
MTT IC50 ± SEM (μM)
336 ± 87
clogP
a
berberine
-1.28
4
a
55 ± 19
58 ± 1
55 ± 8
43 ± 5
17 ± 3
0.44
0.95
0.95
0.95
1.69
4b
4
c
4
4
d
e

4
f
23 ± 1
48 ± 9
32 ± 8
26 ± 7
1.47
1.21
1.21
1.21
0.38
1.07
0.38
1.05
1.05
1.65
0.44
0.30
0.28
1.32
0.28
1.43
0.86
4g
4h
4
4
i
j
b
n.d.
4k
43 ± 6
30 ± 6
30 ± 2
13 ± 1
9 ± 2
50 ± 6
292 ± 37
46 ± 1
21 ± 4
42 ± 3
14 ± 2
42 ± 4
4
l
4m
4
4
4
4
n
o
p
q
4r
4s
4t
4u
4v
a
b
c
Results are expressed as the mean of at least three experiments; Not determined; low solubility profile;
calculated partition coefficient (clogP) between n-octanol and water using MarvinSketch software v. 17.17.0
3.8 In vitro antioxidant properties
To investigate the radical scavenging properties of berberine derivatives (4a-4v)
1
,1-diphenyl-2-picryl-hydrazyl (DPPH) assay was used (Table 4) [69]. The EC50 values of
standards of trolox, N-acetylcysteine, and BBR were compared with published results [70–
3]. Demethylation in position 9- yielding to berberrubine (2) with free phenolic group
7
enabled a remarkable increase in antioxidant activity (EC50 = 87 ± 6.7 μM) [70]. For the novel
derivatives, only compounds 4a-4f possessed higher antioxidant capacity compared to BBR
with top-ranked derivative 4f (EC50 = 990 µM). It can be deduced that slight activity
improvement is associated with small aliphatic substituents (mostly methyl groups), not
depending on their exact position within the attached aryl moiety. These results are,
however, still far from the concentration scale found for trolox or N-acetylcysteine (EC50 = 16
and 28 µM, respectively). Antioxidant activity of novel BBR derivatives can thus be regarded
as improbable.
Table 4. Antioxidant properties of reference compounds (Trolox and N-acetylcysteine), BBR and novel
derivatives 4a-4v established by DPPH assay.
Compounds
DPPH assay ECa50
SEM (µM)
±
4
a
1,900 ± 80
1,200 ± 35
1,200 ± 50
4b
4
c

4
4
4
d
e
f
1,900 ± 47
1,300 ± 110
990 ± 31
4g
˃ 6,000
4
4
h
i
2,200 ± 260
˃ 6,000
4
j
5,000 ± 390
˃ 6,000
4k
4
l
5,600 ± 280
5,400 ± 270
5,200 ± 180
˃ 6,000
5,300 ± 250
˃ 6,000
4m
4
4
4
4
n
o
p
q
4
r
5,300 ± 270
5,300 ± 210
˃ 6,000
4s
4t
4u
˃ 6,000
4
v
5,600 ± 200
5,500 ± 310
16.2 ± 0.4
27.9 ± 1.8
berberine
trolox
N-acetylcysteine
a
Results are expressed as the mean of at least three experiments
3.9 Inhibition of amyloid-β aggregation
Aβ is an intrinsically disordered protein forming insoluble extracellular deposits
[
74,75]. Accumulation of amyloid fibrils and their association with other macromolecules
generate amyloid plaques [4]. Aβ overproduction and its decreased clearance are
responsible for neuroinflammation response, neurodegeneration due to oxidative processes,
disbalance in ion homeostasis, all leading to neuronal dysfunction and death [76]. Since BBR
is capable to reduce levels of Aβ [34,77], we determined anti-amyloid profile of the most
perspective berberine derivatives 4d, 4u and 4v. Doxycycline was selected as a reference
compound known for anti-amyloidogenic property [78]. At 50 µm concentration, all the
tested derivatives revealed high Aβ1-42 inhibition activity compared to BBR, ranging between
82 to 97% (Table 5). The most efficient one was derivative 4u.
Table 5. Inhibition of Aβ1-42 aggregation by the tested compounds at 50 μM concentration.
a
Compound Inhibition of Aβ1-42 aggregates ±SEM (%)
BBR
70.5 ± 6.0
82.6 ± 1.2
97.2 ± 3.4
4d
u
4

4
v
86.7 ± 4.5
~ 100
doxycycline
a
Inhibition ability of the compound to block Aβ1-42 was measured in 1:1 ratio (compound: Aβ1-42) at 50 μM. The
results are the mean of three independent measurements each performed in duplicates.
3
.10 Inhibition of tau protein self-aggregation
Tau (τ) tau phosphorylation is considered to be a major player in pathogenesis of AD.
Abnormally hyperphosphorylated τ generates neurofibrillary tangles, which contain paired
helical filaments [79]. A wealth of evidence proved that τ lesions correlate well with the
severity of dementia [80]. BBR is known to experimentally ameliorate hyperphosphorylation
of τ, and reducing cognitive impairment simultaneously [35]. Building on that fact, 4d, 4u
and 4v were selected as possible inhibitors of τ306-336 peptide aggregation. Results from Table
6
3
demonstrate that all the compounds are potent inhibitors of τ aggregation ranging from
1.9 to 62.0%. 4u yielded the best even surpassing reference compound doxycycline (39.6%)
inhibitory activity (Fig. 6).
Table 6. Inhibition activity of novel berberine derivatives 4d, 4u and 4v on τ self-aggregation. BBR and
doxycycline were used as positive references.
a
Compound Inhibition of τ306-336 peptide self-aggregation ± SEM (%)
BBR
25.0 ± 4.6
31.9 ± 3.6
62.0 ± 4.9
38.3 ± 2.4
39.6 ± 0.5
4
d
u
4
4v
doxycycline
a
Inhibition of τ306-336 self-aggregation was measured in 1:1 ratio (compound:τ306-336) at 25 μM. The
results are the mean of at three independent measurements each performed in duplicates.

1
1
1
1
1
8000
6000
4000
2000
0000
Tau(303-336)_25uM
Tau(303-336) +
berberine_25uM
8
6
4
2
000
000
000
000
0
Tau(303-336) +
4
d_25uM
Tau(303-336) +
u_25uM
Tau(303-336) +
v_25uM
4
4
0
2
4
6
8
10
12
14
16
18
t (h)
Figure 6. Thioflavin (ThT) was selected as fluorescent dye to visualize inhibitory activity of novel BBR
derivatives, BBR and doxycycline as reference compound.
3
.11 In vitro blood-brain barrier permeation
Penetration across the blood-brain barrier (BBB) is an essential property for
compounds targeting the CNS. To predict BBB penetration of chosen BBR derivatives,
parallel artificial membrane permeation assay (PAMPA) was applied (Table 7) [81,82]. For
the estimation of prediction of BBB penetration via passive diffusion, we have selected
analogues with balanced POP and ChE inhibition properties, i.e. 4a, 4g, 4j, 4l, and 4s and 4u
as non-selective ChE inhibitor with highest BChE inhibition profile, top-ranked POP inhibitor
4d, and also the most potent AChE inhibitor 4v. We also measured other drugs serving either
as positive or negative controls, for which CNS availability is known and also reported using
the PAMPA assay (Table 7) [81,82]. According to our methodology using porcine lipid bilayer
only the analogue 4t was predicted to be capable of entering the brain via passive diffusion.
The other analogues, as well as BBR, failed to show such properties. Such findings can be
explained by the presence of permanently charged nitrogen impeding hydrophobic
interactions between BBR analogues with the barrier yielding no or very low permeability by
passive diffusion. Results obtain herein should be taken with precaution since some other
studies displayed that BBR itself is the centrally active drug that can accumulate in
hippocampus to provide a neuroprotective effect after intravenous administration in rat

[83]. Since PAMPA is considered as a powerful screening tool in early drug discovery, it is
however hampered by several limitations i) not taking into account the phospholipid bilayer
difference (porcine vs. human), ii) omitting active transport systems, and iii) excluding
paracellular transport route which is typical for some small hydrophobic compounds [84].
Table 7. Estimation of BBB permeation of BBR, its derivatives and reference compounds
-
6
-1 a
b
Compound
Pe ± SEM (*10 cm s )
Prediction of BBB penetration
BBR
0.2 ± 0.02
0.4 ± 0.06
0.8 ± 0.06
0.6 ± 0.2
1.4 ± 0.4
1.2 ± 0.04
0.8 ± 0.04
4.8 ± 1.4
0.8 ± 0.4
0.4 ± 0.1
0.3 ± 0.09
0.1 ± 0.003
2.5 ± 0.4
9.8 ± 0.4
12.6 ± 3.0
6.0 ± 0.6
CNS –
CNS –
CNS –
CNS –
CNS –
CNS –
CNS –
CNS +
CNS –
CNS –
CNS –
CNS –
CNS +/–
CNS +
CNS +
4
a
d
4
4g
4
j
l
4
4s
4t
4u
4v
cefuroxime
furosemide
piroxicam
donepezil
rivastigmine
tacrine
CNS +
a
b
Results are expressed as the mean of at least three experiments; CNS +: high BBB permeation predicted with
-6
-1
-6
-1
Pe (10 cm.s ) > 4.0, CNS -: low BBB permeation predicted with Pe (10 cm.s ) < 2.0, CNS +/-: BBB permeation
-6
-1
uncertain with Pe (10 cm.s ) from 4.0 to 2.0.
4
CONCLUSION
Phytochemicals represent an indispensable source in drug discovery serving as useful
templates. This is especially important when pursuing neurodegenerative disorders like AD.
BBR is one such successful example of alkaloid with a very low toxicity profile that has been
validated in 77 clinical trials in different areas so far [85]. Being inspired by the success of
BBR itself, we inspected the pharmacological profile of 22 novel BBR derivatives being both
ChE and POP inhibitors. When designing multi-target drugs, the current challenge is to
implement the same degree of modulation for each target [86]. Generally, almost all the

compounds were more profound inhibitors compared to parent BBR. Achieving a balanced
activity towards the aforementioned targets is somehow coined to the structures of 4a, 4g,
4j, 4l, and 4s, all being one-to-two digit micromolar inhibitors of the respective enzymes. For
the specific targets, 4d, 4v, and 4u can be highlighted as the most pronounced POP, AChE
and BChE inhibitors, respectively. Molecular modeling studies between selected ligands and
POP, AChE and BChE, respectively, revealed determinants responsible for the improved
biological profile compared to BBR and justified 9-O aryl substitution. Data for the
cytotoxicity clearly correlated well with the overall lipophilicity increment of each
substituent with derivatives bearing polar groups like cyano (4q) being the least toxic. The
antioxidant profile is presumably associated with the presence of a free phenolic group like
in berberrubine (2). Having said that, novel compounds reported herein revealed only
marginal antioxidant properties, which may be regarded as improbable at physiological
conditions. We also discussed the central availability of BBR derivatives in PAMPA assay
where one derivative (4t) displayed a possibility to be transported into the brain by passive
diffusion. However, direct proof-of-concept by performing in vivo animal studies is
undoubtedly required given the fact that BBR itself that failed in our test, is known to act
centrally. Encouraging results of 4d, 4u and 4v as possible cognitive enhancers prompted us
to validate their disease-modifying properties. All three derivatives showed remarkable
results highlighting 4u. Consistently, 4u outperformed BBR and acted similarly or even better
compared to doxycycline when challenged to Aβ1-42 and τ306-336. To conclude, we believe that
this study may broaden the current knowledge in the development of novel multi-potent
therapeutics derived from BBR with potential implications in neurodegenerative disorders
like AD.
5
5
EXPERIMENTAL SECTION
.1 General Chemical Methods
All of the chemical reagents used were purchased from Sigma-Aldrich (Czech Republic).
Solvents for synthesis were obtained from Penta Chemicals Co (Czech Republic). The course
of the reactions was monitored by thin-layer chromatography (TLC) on aluminum plates pre-
coated with silica gel 60 F254 (Merck, Czech Republic) and then visualized by UV at 254 nm
wavelength. Melting points were determined on a microheating stage PHMK 05 (VEB
Kombinant Nagema, Radebeul, Germany) and are uncorrected. Uncalibrated purity was
ascertained by LC−UV using a reverse phase C18 chromatographic column. All of the

biologically tested compounds exhibited purity 96−99% at a wavelength 254 nm. NMR
spectra of target compounds were recorded on a Varian Mercury VX BB 300 (operating at
1
13
3
00 MHz for H and 75 MHz for C) or on a Varian S500 spectrometer (operating at 500 MHz
1
13
for H and 126 MHz for C; Varian Comp. Palo Alto, CA). Chemical shifts are reported in
parts per million (ppm). Spin multiplicities are given as s (singlet), bs (broad singlet), d
(doublet), dd (doublet of doublets), t (triplet), q (quartet), or m (multiplet). The coupling
1
13
constants (J) are reported in hertz (Hz). All the H and C NMR spectra are listed in
Supplementary Information. High-resolution mass spectra (HRMS) were determined by a Q
Exactive Plus hybrid quadrupole-orbitrap spectrometer.
5
.1.1 Synthesis of berberrubine
The synthesis of berberrubine was carried out according to literature method [87]. Berberine
chloride (5 g, 0.01 M) was placed into the round bottom flask and heated at 205 °C in a
vacuum under reduced pressure (15 mbar) for 2 hours turning yellow BBR to dark red
berberrubine with no need of further purification. Spectral data were in agreement with
literature [45].
5
.1.2 General procedure for the synthesis of 9-O-aryl-berberine bromides (4a-4v)
Berberrubine (0.5 g, 1.4 mM) and potassium carbonate (0.97 g, 2.1 mM) were treated
with anhydrous MeCN (30 mL) at 80 °C for 30 min. Subsequently, the appropriate rate of
arylmethylene bromide (1.5eq) was added. The reaction mixture was refluxed at 80 °C for
1h. After filtration and washing with MeCN (50 mL) and MeOH (50 mL), solid residues were
purified via column chromatography was used for purification using CH Cl/MeOH (5:1, v/v)
3
as eluent. Reagent 3b had to be dissolved in 15 mL THF first because of its low solubility in
MeCN following the aforementioned synthetic pathway.
5
.1.2.1
9-O-benzylberberine bromide (4a)
1
Yield 34%; mp 181-184 °C; H NMR (300 MHz, DMSO-d
6
): δ 9.74 (s, 1H), 8.93 (s, 1H), 8.19 (d,
J = 9.1 Hz, 1H), 7.99 (d, J = 9.1 Hz, 1H), 7.76 (s, 1H), 7.63 – 7.56 (m, 2H), 7.45 – 7.38 (m, 1H),
7
3
1
1
4
.38 – 7.31 (m, 2H), 7.07 (s, 1H), 6.16 (s, 2H), 5.35 (s, 2H), 4.92 (t, J = 6.3 Hz, 2H), 4.07 (s, 3H),
13
.18 (t, J = 6.3 Hz, 2H). C NMR (75 MHz, DMSO-d
6
): δ 150.9, 150.0, 147.9, 145.5, 142.1,
37.6, 136.6, 133.1, 130.8, 129.0, 128.6, 128.5, 126.7, 123.9, 122.0, 120.6, 120.4, 108.6,
+
+
05.6, 102.3, 75.5, 57.2, 55.5, 26.5. HRMS [M+H] : 412.1542 (calculated for: [C26
H
22NO ]
4
12.1543)

5
.1.2.2 9-O-(3-methylbenzyl)berberine bromide (4b)
1
Yield 46%; >300 °C decomposition; H NMR (500 MHz, DMSO-d
6
): δ 9.74 (s, 1H), 9.00 (s, 1H),
8
1
.20 (d, J = 8.9 Hz, 1H), 8.03 (d, J = 8.9 Hz, 1H), 7.79 (s, 1H), 7.43 (s, 1H), 7.35 (d, J = 7.5 Hz,
H), 7.25 (t, J = 7.5 Hz, 1H), 7.14 (d, J = 7.5 Hz, 1H), 7.07 (s, 1H), 6.15 (s, 2H), 5.29 (s, 2H), 4.92
13
(
t, J = 6.4 Hz, 2H), 4.07 (s, 3H), 3.18 (t, J = 6.4 Hz, 2H), 2.31 (s, 3H). C NMR (126 MHz, DMSO-
d
1
2
6
): δ 150.9, 149.9, 147.8, 145.5, 142.2, 137.7, 137.4, 136.5, 133.1, 130.8, 129.6, 129.1,
28.4, 126.7, 126.1, 124.0, 122.0, 120.6, 120.4, 108.6, 105.7, 102.2, 75.6, 57.3, 55.5, 26.5,
+
+
1.1. HRMS [M+H] : 426.1697 (calculated for: [C27
H
24NO
4
] 426.1700)
5
.1.2.3
9-O-(2-methylbenzyl)berberine bromide (4c)
1
Yield 48%; mp 199-202 °C; H NMR (300 MHz, DMSO-d
6
): δ 9.59 (s, 1H), 9.03 – 9.01 (m, 1H),
.22 (d, J = 8.9 Hz, 1H), 8.06 (d, J = 8.9 Hz, 1H), 7.80 (s, 1H), 7.58 (d, J = 7.1 Hz, 1H), 7.26 –
.18 (m, 3H), 7.07 (s, 1H), 6.15 (s, 2H), 5.34 (s, 2H), 4.88 (m, 2H), 4.04 (s, 3H), 3.18 – 3.14 (m,
8
7
2
1
1
13
H), 2.43 (t, 3H). C NMR (75 MHz, DMSO-d ): δ 151.1, 150.0, 147.8, 145.4, 142.4, 137.5,
6
37.0, 134.9, 133.2, 130.8, 130.3, 129.7, 128.6, 126.7, 126.0, 124.1, 121.9, 120.6, 120.5,
+
08.6, 105.7, 102.3, 73.7, 57.2, 55.7, 26.5, 18.8. HRMS [M+H] : 426.1697 (calculated for:
+
[C
27
H
24NO
4
] 426.1700)
5
.1.2.4
9-O-(4-methylbenzyl)berberine bromide (4d)
1
Yield 86%; mp 216-218 °C; H NMR (500 MHz, DMSO-d
J = 9.1 Hz, 1H), 7.99 (d, J = 9.1 Hz, 1H), 7.77 (s, 1H), 7.46 (d, J = 7.5 Hz, 2H), 7.18 (d, J = 7.5 Hz,
H), 7.07 (s, 1H), 6.15 (s, 2H), 5.30 (s, 2H), 4.91 (t, J = 6.4 Hz, 2H), 4.06 (d, J = 8.7 Hz, 3H), 3.18
6
): δ 9.71 (s, 1H), 8.95 (s, 1H), 8.18 (d,
2
13
(
t, J = 6.4 Hz, 2H), 2.28 (s, 3H); C NMR (126 MHz, DMSO-d
6
): δ 150.9, 150.0, 147.8, 145.5,
1
1
42.2, 137.9, 137.5, 133.6, 133.1, 130.8, 129.2, 129.0, 126.7, 123.8, 122.0, 120.6, 120.4,
+
08.6, 105.6, 102.2, 75.4, 57.2, 55.5, 26.5, 21.0. HRMS [M+H] : 426.1697 (calculated for:
+
[C
27
H
24NO
4
] 426.1700)
5
.1.2.5
9-O-(4-isopropylbenzyl)berberine bromide (4e)
1
Yield 65%; mp 207-209 °C; H NMR (500 MHz, DMSO-d
J = 9.1 Hz, 1H), 8.00 (d, J = 9.0 Hz, 1H), 7.78 (s, 1H), 7.49 (d, J = 7.7 Hz, 2H), 7.26 (d, J = 7.7 Hz,
H), 7.08 (s, 1H), 6.16 (s, 2H), 5.29 (s, 2H), 4.90 (t, J = 6.5 Hz, 2H), 4.07 (s, 3H), 3.18 (t, J = 6.5
6
): δ 9.70 (s, 1H), 8.93 (s, 1H), 8.20 (d,
2
13
Hz, 2H), 2.88 (dt, J = 14.7, 7.4 Hz, 1H), 1.17 (d, J = 7.1 Hz, 6H). C NMR (126 MHz, DMSO-d ):
6

δ 150.9, 150.0, 148.9, 147.9, 145.5, 142.3, 137.5, 134.1, 133.1, 130.8, 129.2, 126.8, 126.4,
23.9, 122.0, 120.6, 120.4, 108.6, 105.6, 102.3, 75.5, 57.3, 55.5, 33.4, 26.5, 24.0. HRMS
1
+
+
[
M+H] : 454.2012 (calculated for: [C29
H
28NO
4
] 454.2013)
5
.1.2.6
9-O-(3,5-dimethylbenzyl)berberine bromide (4f)
1
Yield 54%; mp 199-202 °C; H NMR (500 MHz, DMSO-d
6
): δ 9.71 (s, 1H), 8.94 (s, 1H), 8.20 (d,
J = 9.0 Hz, 1H), 8.01 (d, J = 9.0 Hz, 1H), 7.77 (s, 1H), 7.18 (s, 2H), 7.07 (s, 1H), 6.97 (s, 1H), 6.16
13
(
s, 2H), 5.24 (s, 2H), 4.92 (t, 2H), 4.08 (s, 3H), 3.18 (t, J = 6.2 Hz, 2H), 2.27 (s, 6H). C NMR
126 MHz, DMSO-d ): δ 151.0, 150.0, 147.9, 145.5, 142.3, 137.6, 137.5, 136.5, 133.1, 130.8,
29.9, 126.7, 124.0, 122.0, 120.6, 120.4, 108.6, 105.6, 102.3, 75.7, 57.2, 55.5, 26.6, 21.1.
(
6
1
+
HRMS [M+H] : 440.1855 (calculated for: [C28
H
26NO ]+ 440.1856)
4
5
.1.2.7
9-O-(4-brombenzyl)berberine bromide (4g)
1
Yield 7%; mp 222-224 °C; H NMR (500 MHz, DMSO-d
6
): δ 9.79 (s, 1H), 8.94 (s, 1H), 8.18 (d, J
=
2
9.1 Hz, 1H), 8.00 (d, J = 9.1 Hz, 1H), 7.76 (s, 1H), 7.61 – 7.55 (m, 4H), 7.07 (s, 1H), 6.15 (s,
13
H), 5.32 (s, 2H), 4.94 (t, J = 6.2 Hz, 2H), 4.06 (s, 3H), 3.19 (t, J = 6.2 Hz, 2H). C NMR (126
): δ 150.7, 150.0, 147.8, 145.5, 142.0, 137.6, 136.1, 133.1, 131.5, 131.1, 131.0,
30.8, 128.7, 126.6, 124.0, 121.8, 121.7, 120.6, 120.4, 108.6, 105.6, 102.3, 74.6, 57.2, 55.5,
MHz, DMSO-d
6
1
2
+
+
6.5. HRMS [M+H] : 490.0650 (calculated for: [C26
H
21BrNO ] 490.0648)
4
5
.1.2.8
9-O-(3-brombenzyl)berberine bromide (4h)
1
Yield 58%; mp 205-208 °C; H NMR (500 MHz, DMSO-d
6
): δ 9.83 (s, 1H), 8.94 (s, 1H), 8.19 (d,
J = 9.1 Hz, 1H), 8.01 (d, J = 9.1 Hz, 1H), 7.85 (t, J = 1.8 Hz, 1H), 7.76 (s, 1H), 7.59 (d, J = 7.6 Hz,
1
2
d
1
H), 7.55 (dd, J = 7.9, 2.0 Hz, 1H), 7.37 (t, J = 7.9 Hz, 1H), 7.07 (s, 1H), 6.16 (s, 2H), 5.32 (s,
13
H), 4.94 (t, J = 6.3 Hz, 2H), 4.06 (s, 3H), 3.19 (t, J = 6.3 Hz, 2H). C NMR (126 MHz, DMSO-
): δ 150.9, 150.0, 147.8, 145.5, 141.9, 139.4, 137.6, 133.1, 131.4, 131.3, 130.8, 130.8,
27.8, 126.6, 124.2, 121.9, 121.7, 120.6, 120.4, 108.6, 105.6, 102.3, 74.5, 57.2, 55.5, 26.5.
6
+
+
HRMS [M+H] : 490.0649 (calculated for: [C26
H
21BrNO
4
] 490.0648)
5
.1.2.9
9-O-(2-brombenzyl)berberine bromide (4i)
1
Yield 50%; mp 194-197 °C; H NMR (500 MHz, DMSO-d
6
): δ 9.83 (s, 1H), 8.94 (s, 1H), 8.20 (d,
J = 9.2 Hz, 1H), 8.01 (d, J = 9.2 Hz, 1H), 7.85 (t, J = 1.8 Hz, 1H), 7.77 (s, 1H), 7.61 – 7.58 (m,