Synthesis and biological evaluation of novel thiazole- VX-809 hybrid derivatives as F508del correctors by QSAR-based filtering tools

Article abstract of DOI:10.1016/j.ejmech.2017.12.030

The most common CF mutation, F508del, impairs the processing and gating of CFTR protein. This deletion results in the improper folding of the protein and its degradation before it reaches the plasma membrane of epithelial cells. Present correctors, like VX809 only induce a partial rescue of the mutant protein. Our previous studies reported a class of compounds, called aminoarylthiazoles (AATs), featuring an interesting activity as correctors. Some of them show additive effect with VX809 indicating a different mechanism of action. In an attempt to construct more interesting molecules, it was thought to generate chemically hybrid compounds, blending a portion of VX809 merged to the thiazole scaffold. This approach was guided by the development of QSAR analyses, which were performed based on the F508del correctors so far disclosed in the literature. This strategy was aimed at exploring the key requirements turning in the corrector ability of the collected derivatives and allowed us to derive a predictive model guiding for the synthesis of novel hybrids as promising correctors. The new molecules were tested in functional and biochemical assays on bronchial CFBE41o-cells expressing F508del-CFTR showing a promising corrector activity.

Full text of DOI:10.1016/j.ejmech.2017.12.030

Accepted Manuscript
Synthesis and biological evaluation of novel thiazole- VX-809 hybrid derivatives as
F508del correctors by QSAR-based filtering tools
Nara Liessi, Elena Cichero, Emanuela Pesce, Maria Arkel, Annalisa Salis, Valeria
Tomati, Matteo Paccagnella, Gianluca Damonte, Bruno Tasso, Luis J.V. Galietta,
Nicoletta Pedemonte, Paola Fossa, Enrico Millo
PII:
S0223-5234(17)31036-X
10.1016/j.ejmech.2017.12.030
EJMECH 10008
DOI:
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 3 August 2017
Revised Date: 7 December 2017
Accepted Date: 7 December 2017
Please cite this article as: N. Liessi, E. Cichero, E. Pesce, M. Arkel, A. Salis, V. Tomati, M. Paccagnella,
G. Damonte, B. Tasso, L.J.V. Galietta, N. Pedemonte, P. Fossa, E. Millo, Synthesis and biological
evaluation of novel thiazole- VX-809 hybrid derivatives as F508del correctors by QSAR-based filtering
tools, European Journal of Medicinal Chemistry (2018), doi: 10.1016/j.ejmech.2017.12.030.
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

ACCEPTED MANUSCRIPT
Synthesis and biological evaluation of novel thiazole- VX-809 hybrid derivatives as F508del
correctors by QSAR-based filtering tools
1
,2±
3±
4
1
2
Nara Liessi
, Elena Cichero , Emanuela Pesce , Maria Arkel , Annalisa Salis , Valeria
4
1
1,2
3
4
Tomati , Matteo Paccagnella , Gianluca Damonte , Bruno Tasso , Luis J. V. Galietta ,
4
3*
1,2*
Nicoletta Pedemonte , Paola Fossa and Enrico Millo
±
These authors contributed equally to this work
1
Department of Experimental Medicine, Section of Biochemistry, University of Genoa, Viale
Benedetto XV 1, 16132 Genoa, Italy.
2
Center of Excellence for Biomedical Research (CEBR),University of Genoa, Viale Benedetto XV
9, 16132 Genoa, Italy.
3
Department of Pharmacy, Section of Medicinal Chemistry, School of Medical and Pharmaceutical
Sciences, University of Genoa, Viale Benedetto XV, 3, 16132 Genoa, Italy.
4
U.O.C. Genetica Medica, Istituto Giannina Gaslini, Genova, Italy.
*
Corresponding author for E. M. at Center of Excellence for Biomedical Research (CEBR),
University of Genova, Viale Benedetto XV, 9, 16132 Genoa, Italy. Tel.: +39 (0)103533032-33 fax:
39(0)103533024. E-mail address: enrico.millo@unige.it
+
For P.F. at Department of Pharmacy, Section of Medicinal Chemistry, School of Medical and
Pharmaceutical Sciences, University of Genoa, Viale Benedetto XV, 3, 16132 Genoa, Italy.
Tel++39(0)103538238 fax: +39(0)103538357. E-mail address: fossa@difar.unige.it
1

ACCEPTED MANUSCRIPT
Abstract
The most common CF mutation, F508del, impairs the processing and gating of CFTR protein. This
deletion results in the improper folding of the protein and its degradation before it reaches the
plasma membrane of epithelial cells. Present correctors, like VX809 only induce a partial rescue of
the mutant protein. Our previous studies reported a class of compounds, called aminoarylthiazoles
(AATs), featuring an interesting activity as correctors. Some of them show additive effect with
VX809 indicating a different mechanism of action. In an attempt to construct more interesting
molecules, it was thought to generate chemically hybrid compounds, blending a portion of VX809
merged to the thiazole scaffold. This approach was guided by the development of QSAR analyses,
which were performed based on the F508del correctors so far disclosed in the literature. This
strategy was aimed at exploring the key requirements turning in the corrector ability of the collected
derivatives and allowed us to derive a predictive model guiding for the synthesis of novel hybrids as
promising correctors. The new molecules were tested in functional and biochemical assays on
bronchial CFBE41o- cells expressing F508del-CFTR showing a promising corrector activity.
Abbreviations:
CF, cystic fibrosis; CFTR, CF transmembrane conductance regulator; AATs, aminoarylthiazoles;
HS-YFP, halide-sensitive yellow fluorescent protein; QSAR, quantitative-structure activity
relationship; HBTU, O-(Benzotriazol-1-yl)-N,N,N′,N′-tetramethyluronium hexafluorophosphate;
DIPEA, N,N-Diisopropylethylamine.
Keywords:
Aminoarylthiazole, corrector, QSAR, cystic fibrosis, CFTR, VX809.
2

ACCEPTED MANUSCRIPT
1. Introduction
The genetic disease cystic fibrosis (CF) is caused by mutations in the cystic fibrosis transmembrane
conductance regulator gene (CFTR), which encodes a cAMP-regulated chloride channel [1–4].
CFTR protein contain five domains: two nucleotide binding domains (NBD1 and NBD2), two
transmembrane domains (MSD1 and MSD2), and one regulatory domain (R) [5]. So far, a large
number of CF-causing mutations have been identified, which can be classified in six classes
according to the mechanism leading to CFTR loss-of-function [6], such as reduction of protein
expression, function, stability or a combination of these[7-9].
The most common genetic defect is the deletion of phenylalanine 508 (F508del), located in NBD1.
F508del-CFTR is misfolded retained at the endoplasmic reticulum (ER), and rapidly degraded [10-
12]. This kind of mutation leads to distinct defects in channel gating and cellular processing. Cystic
fibrosis results in chronic lung infection, deterioration of lung function, and death.
The primary defects caused by CFTR mutations can be treated with drug-like small molecules,
known as ‘CFTR modulators’, targeting specific defects caused by mutations in the CFTR gene.
They are classified into five main groups including read-through agents, correctors, potentiators,
stabilizers and amplifiers [13]. Even if the development of monotherapies could allow addressing
specific molecular defect related to some CFTR mutations, a combination of treatments could
rectify several multi-defects at the same time, achieving beneficial levels in the patients. Therefore,
small-molecule therapy will likely require compounds that correct at least the two major underlying
problems in CF: (a) CFTR misfolding and ER retention, and (b) defective channel gating [14-15].
In particular, the folding and the channel activity defect of F508del can be managed with the
aforementioned modulators named correctors and potentiators, respectively.
Correctors improve the trafficking of mutant CFTR to the plasma membrane, instead, potentiators
are believed to bind to F508del-CFTR at the cell surface and increase chloride channel gating.
3

ACCEPTED MANUSCRIPT
While potentiators probably act by directly interacting with CFTR protein, the mechanism of action
of correctors is unknown. They could directly bind to CFTR improving its folding and stability or
regulate the expression/function of other proteins involved in CFTR processing and trafficking [16-
17].
Because both a potentiator and corrector are likely required to treat cystic fibrosis [12] caused by
the F508del mutation, CFTR is an attractive target for the development of a multiligand-containing
drug, which represents an emerging paradigm in drug discovery [18].
During the last years, several efforts have been performed to identify new promising chemical
entities able to behave as F508del -CFTR corrector and/or potentiator, relying on virtual screening
protocols (VS) and deepening high-throughput screening strategies (HTS) [18-24].
The most promising data came from HTS approaches, conducted by several research groups and by
Vertex Pharmaceuticals. In particular, Verkman and colleagues paved the way in the discovery of
thiazole-containing correctors through HTS assays, disclosing corrector 4a that exhibited
micromolar potency (EC50 < 3µM) [21]. This hit compound was further optimized, towards the
design of novel series of bioisosteres experienced as CFTR correctors. These compounds included
conformationally-locked bithiazoles, pyrazolylthiazoles and triazolo-bithiazoles, characterized by
improved hydrophilicity and potency values (Figure 1) [18, 21, 23, 24].
4

ACCEPTED MANUSCRIPT
Figure 1. Chemical structure of the HTS-discovered thiazole-containing series of CFTR correctors.
Recurrent hydrophobic and H-bonding features are highlighted in cyan and yellow, respectively.
Then, the same Authors also investigated further chemo-types, bearing a cyanoquinoline core linked
to an aryl amide moiety by a variable tether, which displayed valuable corrector activities [25].
More recently, a screening campaign reported by other research groups enlightened a number of
tetrahydropyridopyrimidines, properly decorated with aromatic substituents and basic moieties, as
novel promising CFTR correctors [26]. Notably, both the two series display recurrent hydrophobic
and H-bonding features if compared with the thiazole-based correctors, thanks to the bicyclic core
and to the related aromatic substituents, maintaining adequate corrector ability (Figure 2).
5

ACCEPTED MANUSCRIPT
Figure 2. Chemical structure of the HTS-discovered thiazole and bioisosteres acting as CFTR
correctors. Recurrent hydrophobic and H-bonding features are highlighted in cyan and yellow,
respectively.
Starting from HTS, Vertex Pharmaceuticals discovered cycloalkyl carboxamide F508del-CFTR
correctors, firstly identifying a thiazole-containing compound (VRT-768) as prototype (Figure 2).
This compound was further optimized towards the well-known corrector VX-809 (Lumacaftor),
bearing a pyrimidine ring as bioisosteres of the previous thiazole nucleus (Figure 2) [27].
In the search of new drugs for CF treatment, in our previous works [28, 29] we have shown how
some aminoarylthiazoles (AATs) exhibit F508del modulator ability, being especially correctors
and/or sometimes potentiators of the channel. Moreover, combined treatment of some of these
correctors and VX809 resulted in additive effect. Indeed, incubation of CFBE41o- cells with some
AATs together with VX-809 generated a five-fold increase in activity relative to vehicle-treated
cells, suggesting for different mechanisms of action by the two chemo-types [29].
6

ACCEPTED MANUSCRIPT
Concerning this issue, a number of in silico docking studies devised VX-809 directly targeting the
full-length F508del-CFTR protein, with a possible binding site at the NBD1-ICL4 interface [30].
In this regard, recent literature from the Forman-Kay group highlights an allosteric coupling
between the binding site of VX809 and the NBD1:ICL4 interface [31].
On the contrary, other hypotheses assert that VX-809 specifically interacts with TMD1 of CFTR
[
32-34], but does not increase the stability of F508del [32].
Nevertheless, the precise mechanism of action supporting the corrector ability featured by VX809
as well as that of AATs is still experimentally unclear, making the drug discovery and lead
optimization process a challenging task.
On the other hand, a perspective of the whole chemical entities so far discovered allows deciphering
common and recurrent pharmacophore features for CFTR correctors, including two aromatic-
lipophilic moieties connected by at least one H-bond acceptor function (see Figures 1-2). Moreover,
this information is in accordance with other studies proposed in the literature and focused on the
development of pharmacological chaperone correctors in CF therapy, displaying aromatic centers
and one H-bond acceptor moiety [35, 36].
On this basis, we deemed interesting to proceed our work about CFTR modulators applying
quantitative-structure activity relationship (QSAR) on the aforementioned different chemical
classes of correctors, in order to point out a limited number of chemical descriptors probably
involved in their variable corrector ability trend. This information allowed providing useful
guidelines to co-orchestrate the rational design of the newly synthesised CFTR correctors here
proposed.
Indeed, the derived pool of descriptors was used as filtering tools guiding for the optimization of
our AAT series. Based on these considerations and with the aim to improve the activity profile of
the AATs, we have now designed and synthesised a library of novel derivatives to investigate their
effect on mutant CFTR.
7

ACCEPTED MANUSCRIPT
In particular, we continued our synthetic efforts exploring the effectiveness of novel derivatives,
namely hybrid compounds, obtained by merging the benzodioxole moiety of VX809 with the
thiazole chemo-type of AATs. This represents the first step in the search of novel correctors
rationally designed by fragment-based criteria, which is highly motivated by the well-known
potency of VX809 so as by the interesting CFTR modulator ability experienced by the thiazole
scaffold (Figure 3).
Figure 3. Chemical structure of the newly synthesised VX-809 and thiazole-containing hybrids.
Merged moieties are shown by grey dot lines. Explored substituents of the hybrid-scaffold are
highlighted in cyan.
The main issue to be addressed was to clarify the role played by the fluorinate benzodioxole portion
when it is coupled with the thiazole-containing motif in maintaining and/or improving the corrector
ability of the new compounds. In addition, the introduction of the carboxamide function linked at
the thiazole core opens the possibility to better fulfil the specific pharmacophore requirements
experienced by the previously mentioned correctors, discussed in the literature.
8

ACCEPTED MANUSCRIPT
The set of compounds drawn and synthesised has been defined “hybrid compounds”, since their
structure is really a hybrid between the AATs and the VX809 scaffolds. The design and synthesis of
new libraries was planned in order to maximize the possibility of identify novel compounds
endowed with improved F508del corrector ability, filtering the most promising compounds based
on the information obtained but the aforementioned QSAR analyses.
The ligand-based guided approach proved to deepen the discovery of VX-809-like hybrid structures
bearing the thiazole core, endowed with a promising corrector activity. Further structural
modifications merging other key moieties of VX809 and of AATs (following series of hybrids) will
represent a rare opportunity to clarify whenever the thiazole scaffold could be endowed with
improved CFTR modulator ability and will be discussed in due course.
2
2
. Results and discussion
.1. Chemistry
The lead structure of our derivatives can be split into three different parts: the benzodioxole
substituted portion, the central cyclopropane moiety and the aminoarylthiazole substituted portion.
To prepare the benzo [1, 3] dioxol-5-yl-cyclopropanecarboxylic acid (1a) we explored the reaction
of arylacetonitriles with ethylene carbonate in K CO . The treatment of arylacetonitriles with
2
3
ethylene carbonate in the presence of potassium carbonate at T=140°C should generate the
cyclopropane ring on the structure but without a satisfactory yield [37].
The cyclopropanation of active methylene compounds is carried out using either ethylene dichloride
or ethylene dibromide as the alkylating agent in the presence of a strong base like 50% aqueous
sodium hydroxide in the presence of a phase transfer catalyst [38].
Initially, we performed the cyclopropanation using 1, 2-dibromoethane and K CO in the presence
2
3
of a catalytic amount of BTEAC as a phase transfer catalyst to furnish 1-(benzo[d] [1, 3] dioxol-5-
yl) cyclopropanecarboxylic acid in very low yield via an intramolecular dialkylation reaction.
9

ACCEPTED MANUSCRIPT
This reaction was repeated with bromochloroethane, and a strong base such as 50 % NaOH in
aqueous media to furnish the product 1a in 65% yield adapting to published procedures [39].
A similar protocol was used to synthesised the 1-(2, 2-difluorobenzo[d] [1, 3] dioxol-5-yl)
cyclopropanecarboxylic acid 1b starting from 2, 2-difluorobenzo[d] [1, 3] dioxole-5-carbaldehyde
in four step as previously published with minor modifications [40].
The substituted aminoarylthiazole were synthesised in a range of variable yields via the classic
Hantzsch thiazole synthesis refluxing the appropriate haloketone with an equimolar amount of the
thiourea in absolute ethanol (Scheme 1) [41, 42].
In the search for a simple and effective means of obtaining aminothiazole derivatives, we used the
selective monobromination of ketones. All the bromoketones that were not commercially available
were readily prepared from the corresponding ketones utilizing N-bromosuccinimide (NBS),
catalyzed by trimethylsilyl trifluoromethanesulfonate (TMSOTf) in acetonitrile [43]. This method is
useful for the side-chain bromination of heteroaromatic carbonyl compounds without the ring
brominations. (Scheme 1).
The aminothiazole of the analogue 5c was synthesised, in a first moment, through the simple
procedure from the precursor 1-(3-hydroxyphenyl) propan-1-one, its bromination and the final
treatment with the thiourea but without success. The very low yield of the procedure the yield is due
to the difficulty of selectively monobromination the ketone and of isolating the respective
intermediate 2-bromo-(4-hydroxyphenyl) propan-1-one. The same synthetic troubles have been
encountered in obtaining the derivative 5d from the precursor 1-(4-hydroxyphenyl) propan-1-one.
To overcome such drawbacks we adopted a different strategy. There was a significant increase in
yield of the condensation of the hydroxyl substituted propiophenone with thiourea and iodine to
obtain 4-(2-amino-5-methylthiazol-4-yl)phenol or 3-(2-amino-5-methylthiazol-4-yl)phenol using
microwave heating under solvent-free conditions as reported in the literature[44] (Scheme 2) .
1
0

ACCEPTED MANUSCRIPT
The synthesis of amide derivatives of 2-aminothiazoles were achieved by condensation reaction of
appropriate 2-amino-thiazole with the carboxylic group of cyclopropanecarboxylic acid derivatives
with uronium salt HBTU /DIPEA activation in anhydrous DMF[29] (Scheme1).
A different synthetic route was used to synthesise 5f. In fact the appropriate thiazole ethyl 4-(2-
amino-5-methylthiazol-4-yl) benzoate was prepared starting from the corresponding derivative
ethyl 4-propionylbenzoate.
Unfortunately we found, in pilot experiments, that the published routes to the carboxylic acid
precursor, 4-carboxypropiophenone, from p-methylpropiophenone gave low and/or irreproducible
yields. The oxidation of the methyl group with KMnO [45] affords mainly a lot of sub-products
4
rather than 4-carboxypropiophenone. To overcome these drawbacks we developed a more reliable
synthesis from the commercially available starting material 4-bromopropiophenone according with
the method of Rosowsky [46]. As shown in Scheme 3, treatment of 4-bromopropiophenone with
CuCN in DMF afforded the nitrile, which on hydrolysis with KOH in acetonitrile was converted to
carboxylic acid. The next acid esterification with ethanol completed the synthesis furnishing the
ethyl 4-propionylbenzoate in good yield. The aminothiazole was thus built through the traditional
procedure from this precursor, its bromination and the final treatment with the thiourea. The
appropriate condensation reaction of ethyl 4-(2-aminothiazol-4-yl) benzoate with the carboxylic
group of cyclopropanecarboxylic acid derivative promotes the formation of the esterified precursor
which gives rise to the final derivative through hydrolysis.
The compound 5e was indeed built starting from ethyl 4-acetylbenzoate. The ketone was then
brominated with NBS as previously described and conjugated with thiourea in refluxing ethanol.
The ethyl 4-(2-aminothiazol-4-yl) benzoate was then conjugated with 1a affording an esterified
derivative that was subjected to alkaline hydrolysis to obtain the final compound with the carboxyl
group (Scheme 4).
1
1

ACCEPTED MANUSCRIPT
The compound 7a and 7b ,that not containing the thiazole scaffold, were obtained by condensation
of the carboxylic group of benzo [1, 3] dioxol-5-yl-cyclopropanecarboxylic acid (1a) with the
amine derivatives 3-(methylthio) aniline and 3-aminoquinoline respectively (Scheme 5) .
2.2. QSAR analyses of F508del CFTR correctors
In the search of novel F508del correctors, we deemed interesting to proceed our work with the
development of quantitative-structure activity relationship (QSAR) analyses, aimed at disclosing
some relevant structural features exhibited by a number of known correctors, as described in the
literature. Indeed, we built in silico a dataset including sixty-three compounds (1-63; see Supporting
information S1-S3), being compounds 1 (pEC50 = 5.59) and 2 (pEC50 = 5.96) the well-known
correctors VX-809 and corrector-4a, respectively. Most of the other derivatives were characterized
by a number of the aforementioned tetrahydropyridopyrimidines (4-30), cyanoquinolines (31-54),
and thiazole-containing molecules (55-63), as shown in Figure 4.
Figure 4. Chemical structure and biological activity range of the investigated F508del CFTR
correctors
1
2

ACCEPTED MANUSCRIPT
The predictive mathematical model was calculated by dividing compounds 1-63 into a training set
including fifty molecules (2-9, 12-19, 21, 23, 27-31, 33, 35, 36, 38-48, 50-56, 58-63) for model
generation, and into a test set with the others derivatives (1, 10, 11, 20, 22, 24-26, 32, 34, 37, 49,
57) to evaluate the reliability of the mathematical relationship. In particular, the obtained statistical
2
2
tool was generated by employing non-cross-validated PLS analysis to give a cross validated r (r )
cv
2
2
2
=
0.80 a non-cross validated r (r ncv) = 0.91, root mean square error (RMSE) = 0.220, a test set r
2
(r
pred) = 0.82. The predicted and experimental corrector ability values of all the compounds,
together with the collected descriptors values, were reported as tables in Supporting information
S4).
(
In order to develop QSAR analyses, three hundred molecular descriptors (including 2D- and 3D-
parameters) were calculated, by means of MOE software. 2D molecular descriptors are classified in
seven clusters, which are related to physical properties (2D-I), subdivided surface areas (2D-II),
atom and bond counts (2D-III), connectivity-based descriptors (2D-IV), partial charges descriptors
(2D-V), pharmacophore features descriptors (2D-VI) and the so-called Adjacency and Distance
Matrix Descriptors (2D-VII). 3D-descriptors consist of five groups, such as potential energy
descriptors (3D-I), MOPAC descriptors (3D-II), Surface Area (3D-III), Volume and Shape
Descriptors (3D-IV) and Conformation-Dependent Charge Descriptors (3D-V).
In this work, a limited cluster of descriptors was selected using the QSAR-Contingency module
implemented in MOE. The top-fifty best ranked parameters, based on the contingency scoring
function, were retained and further filtered by re-iterative partial least square (PLS) analyses, taking
into account only those parameters exhibiting the highest relative importance (RI) values.
Following this method, we selected eight descriptors explaining for the corrector ability range of the
collected library, being the most of them 2D descriptors (see Table 1). Indeed, three descriptors fall
in the 2D-III cluster, while two other ones in the 2D-IV and 2D-VII, respectively. On the other
1
3

ACCEPTED MANUSCRIPT
hand, three 3D-descriptors were chosen, being two of them included in the 3D-III subtype, while
the final one in 3D-I.
Both the two Vsurf descriptors underline a role played by adequate balance in hydrophilic exposed
features and by an overall surface extent. This information is also supported by chi1 and weinerpol
parameters, which suggest for a proper connectivity table and overall dimension of the ligand,
turning in preferred branched and flexible substituents, over rigid, planar and much more extended
groups. As consequence, the presence of single bonds is suggested, being on the other hand limited
the number of hydrogen atoms. This is in harmony with the overall need for a prominent polarity
profile, to be unraveled along a branched chemical structure.
Quantitatively, the corrector ability of the compounds here studied is explained by the descriptors
shown in Table 1 and by the following equation:
Eq. (1): pEC = -0.17522 + 1.54795 * chi1 -0.19794*weinerpol -0.03874*E_nb -0.12813*a_IC -
5
0
0.17006*a_nH +0.24667*b_single +0.10392*vsurf_W8 +0.15617*vsurf_DD12
Based on these data, increased values of chi1, b_single, and vsurf descriptors should improve the
potency of the compounds, being the two 2D descriptors the most effective in tuning potency (the
calculated descriptors within the whole dataset are listed in Supporting information S4). In
particular, most of the more potent derivatives (such as 27, 28; pEC > 6.00) display b_single and
5
0
chi1 values spanning from 50 to 70, and from 15 to 18, while the less effective ones (such as 33-42;
.00 < pEC < 5.00) fall between 30 to 60 and 12 and 17, respectively. This is in agreement with
4
5
0
the lower potency trend experienced by the cyanoquinoline series (31-54; pEC = 4.00-5.82) with
5
0
respect to the other ones, displaying b_single and chi1 values spanning from 28 to 37, and from 12
to 14, respectively.
Interestingly, this turns in more planar and rigid structures and in poor levels of corrector ability.
Conversely, tetrahydropyridopyrimidines (3-30; pEC = 4.00-6.70) and pyrazolylthiazoles such as
50
5
6-63 (pEC = 5.07-6.12) appeared to be well-suited, exhibiting mean values of b_single and chi1
50
descriptors of 54 and 17; and 44 and 16, respectively.
1
4

ACCEPTED MANUSCRIPT
According to our model, too high values in weinerpol, E_nb, a_IC and a_nH descriptors could
impair the potency of these correctors. In particular, lowering of weinerpol descriptor is
accompanied by the introduction of flexible polar chains onto the central core of the molecules,
suggesting for a folded conformation of the derivative, while higher values of this descriptor occurs
when more extended features are chosen. Indeed, the tetrahydropyridopyrimidines 24, 25 (pEC =
5
0
6.70) bearing a flexible basic chain in R2 (see Figure 4) display adequate weinerpol values
(
weinerpol = 46-47) if compared with the analogue 16 (pEC = 4.00; weinerpol = 55), featuring a
5
0
piperazine-containing substituent. Accordingly, at the cyanoquinoline series, the piperazine-based
compounds 44, 45 (pEC = 4.00) exhibit higher weinerpol values (weinerpol = 52) than the related
5
0
linear aliphatic chain-containing analogues 50, 51 (pEC50 = 5.34-5.37; weinerpol = 45), with
expense in corrector potency. As consequence, it could be hypothesized that the proper weinerpol
value moves around 50.
Concerning potential energy, E_nb values decrease in presence of proper chemical fragments linked
to the central scaffold of the corrector, leading to effective intra-molecular contacts, such as intra-
hydrogen bonds and hydrophobic interactions. This kind of conformation often turns in the
aforementioned branched arrangement, improving the corrector ability of the compound. Indeed,
tetrahydropyridopyrimidines bearing flexible chains in R2 together with H-bonding functions (such
as 19; pEC =6.70) could be involved in intramolecular H bonds with the protonated nitrogen atom
5
0
of the tetrahydropyridine ring, displaying low levels in E_nb (19, E_nb = 87). On the contrary, the
more rigid compound 18 (pEC50 = 5.46) exhibits higher E_nb (18, E_nb = 90). Accordingly,
cyanoquinolines featuring basic and flexible ethylene diamine (31-42; pEC50 = 4.00-5.82), or
propylene diamine (49-54; pEC = 4.00-5.52) chains in R1, exhibit E_nb mean values around 55,
50
being able to be protonated and H-bonding the quinolone nitrogen atom. Conversely, the related
rigid analogues with a piperazine ring in R1 (43-48; pEC = 4.00) are characterized by higher E_nb
5
0
values, spanning from 65 to 71. Within the pyrazolylthiazole series, compound 63, bearing an
ethoxy-amine chain in R2, is the one endowed with the lower E_nb parameter (pEC = 5.52, E_nb
5
0
1
5

ACCEPTED MANUSCRIPT
=45). Finally, lowering in a_IC and a_nH parameters proved to be beneficial for the corrector
ability of the derivatives here investigated, promoting the design of poorly planar scaffolds enriched
with polar moieties, within overall limited and bulky conformations.
2.3. Rational design of AAT-VX809 hybrids
Based on one our previous study [29], in which we modified the AAT structure at different
positions, for the identification of derivatives with improved potency and efficacy, we chose to
initially build a series of substituted analogues. The modifications were made in one region at a
time, to determine the contribution of each particular structural portion to bioactivity. Such
modifications led to a variable range of activities that helped us in the generation of a plausible
SAR, as will be described below.
To begin the investigation, information around compound was generated by using the central amide
bond to synthesise analogs, and in this first phase to probe independently the role of both the
benzodioxole and the thiazoles moieties. A range of distinct analogs was designed to examine the
requirements for the action (Table 2).
Despite of the limited number of the (available) collected F508del CFTR correctors, the QSAR
analyses allowed to derive some preliminary information and useful filtering tools to continue our
work in the design and development of novel AAT-containing derivatives, even merging key
moieties featured by VX-809. Among the proposed substitutions, some of them proved to be better
ranked and filtered taking into account the previously mentioned adequate ranges of the b_single,
chi1 and weinerpol descriptors (see Supporting Information S5). Indeed, those derivatives bearing
an ester moiety at the thiazole position 5 (2b-2d) exhibited comparable weinerpol and chi1 mean
values (around 52 and 15, respectively) with respect to the suggested thresholds (weinerpol around
50, chi1 spanning from 15 to 18), while the b_single values was of 36. Accordingly, we considered
1
6

ACCEPTED MANUSCRIPT
them as the most potent within the whole series. In addition, this information has drawn some
guidelines for further optimization, suggesting for more substitutions with non-aromatic rings
and/or with more flexible substituents, in order to increase this parameter to the aforementioned
favorable trend (b_single spanning from 50 to 70).
Notably, 2a is the only one compounds of the newly synthesised derivative to exhibit high chi1
values (around 16), revolving around a bulky branched chemical structure. This data seems to
confirm the information coming from the QSAR study.
According to our model, removing the fluorine-containing benzodioxole moiety of VX-809 such as
in 1-5 series, as well as maintaining the fluorinate motif (compounds 6 series), allowed to derive a
number of congeners sometimes endowed with quite comparable potency trend (see Supporting
Information S5). Indeed, the predicted pEC values of 5a (pred. pEC = 6.59), 5b (pred. pEC =
5
0
50
50
6
.56), and 3c (pred. pEC = 6.17) were comparable with those of 6d (pred. pEC = 6.15), 6c (pred.
50 50
pEC = 6.29) and 6b (pred. pEC = 5.79). Notably, these results were in good agreement with the
50
50
biological assays discussed in the following sections concerning 5a, 5b, 3c (Exp. pEC50 = 5.64-
.74) with respect to 6d, 6c, 6b (Exp. pEC = 5.27-5.89).
5
50
Removing bulky groups by the thiazole position 5 was predicted to be allowed to design
compounds endowed with promising potency values, as well as bearing a methyl group at the same
position, especially in presence of when accompanied by a meta or para-substituted phenyl ring
linked at the thiazole position 4. Accordingly, the methyl-substituted AATs 3a, 3b, 5c, 5d (Pred.
pEC = 5.95-6.47) were predicted to be quite effective as the unsubstituted analogues 4a, 4b, 5b,
50
5
a (Pred. pEC = 6.08-6.36).
50
Removing the thiazole core could in any case be well-tolerated, as shown by the calculated potency
trend for 7a and 7b (Pred. pEC = 4.39-5.40). Conceivably, this kind of scaffold was able to fulfill
50
the pharmacophore requirement we previously mentioned for the collected correctors, which
exhibited two aromatic cores linked by one H-bonding function. As consequence, also compounds
7a and 7b featured modest corrector ability.
1
7

ACCEPTED MANUSCRIPT
2.4. Effect of compounds as correctors of F508del-CFTR trafficking defect
To corroborate our model all designed compounds were then synthesised and tested to investigate
the structure-activity relationships as correctors of F508del-CFTR (Table 2).
By using the YFP functional assay on F508del-CFTR CFBE41o- cells, we tested the compounds
after 24 hours incubation at different concentrations to extrapolate the EC50 values of the
compounds as correctors of mutant CFTR (Table 2). Activity, after incubation with these hybrids,
of F508del-CFTR in the plasma membrane was determined by measuring the rate of HSYFP
quenching caused by iodide influx as previously published [29]. Activity, after incubation with
these hybrids, was compared to that of cells treated with vehicle alone (DMSO) or with the known
corrector VX-809 (1 µM).
All the compounds show activity as correctors except for the compound 7c with a not detectable
EC value. The most active compounds of the series, in according with the predicted pEC (Pred.
50
50
pEC = 6.97-7.21), were 2a (EC =0.09µM; Exp. pEC = 7.06), 2b (EC = 0.3µM; Exp. pEC =
50
50
50
50
50
6
.52), 2c (EC = 0.55µM; Exp. pEC = 6.26) and 2d (EC = 0.90 µM; Exp. pEC = 6.05). All the
50 50 50 50
compounds were characterized by a substitution on position 5 of the thiazole ring with two different
portions like benzoyl group for 2a and ethyl ester for the others derivatives. The presence of
halogens in meta or para-substituted phenyl ring linked at the thiazole position 4 was well-tolerated
and did not change much the activity.
We investigate whether the substitution at the C-5 of the 2-aminothiazole was fundamental for the
activity and, if so, which characteristics these substituents should possess. In fact, the presence of a
bulky substituent (benzoyl or ethyl acetate) at the C-5 might be useful in terms of improving some
pharmacokinetic characteristics. For example, a small group such as the methyl did not improve the
activity compared to the counterpart (see 3a vs 2d or 3b vs 2b, 3c vs 2c and 3d vs 2a), where the
1
8

ACCEPTED MANUSCRIPT
EC50 were higher compared to the precursor, even if is well-tolerated. Nevertheless, the methyl
group allowed an increase in activity in the case of compound 3e (EC = 0.90 µM; Exp. pEC =
5
0
50
6.05), where a different halogen (F) in the para position of the ring was substituted to the chloride
in the most active of this series (3d with EC of 1.60 µM; Exp. pEC = 5.80). It is reasonable to
5
0
50
assume that the introduction of a fluorine atom in the molecule led to an increase in hydrophobicity
with a small bulkiness, a characteristic likely to be important for activity.
Activity was significantly decreased when maintaining the same substituents on the 4-position, the
thiazole ring was unsubstituted at the position 5 (see compounds 4a-c and their EC ). The same
5
0
results were obtained with compound without benzoyl group (data not shown).
We might speculate that the phenyl ring, by virtue of its low flexibility, can accommodate in a
favorable manner if the position 5 is suitably substituted. These data suggest how the thiazole ring
plays a pivotal role in the position of the entire construct. In particular, the substitutions at the C-5
position of the thiazole were optimal when accompanied by a carbonyl group like in the molecule
2a.
A second series of derivatives comprised different substitutions on the position 4 of the thiazole by
changing the substitution pattern of the phenyl ring.
We also investigated some compounds with a common feature like the presence of halogen groups
in the ring. In particular we evaluated two derivatives with chlorine and bromine in para (4a and
4
b; Exp. pEC = 5.54 and 5.31), exhibiting weak corrector ability (EC =2.9 µM and EC =4.9µM
50 50 50
respectively). As the most active contained chlorine in para (4a EC =2.9µM), we decided to insert
5
0
in the same position iodine (4d) leading to a significant increase in activity (EC =1.2 µM; Exp.
5
0
pEC50 = 5.92). Actually, the electronic effects of the halogens were ascribed to their inductive
electron attractive properties and such properties are maximal for chlorine and bromine, less marked
for iodine. Importantly, moving the chlorine group from the para (4a) to the meta position (4c)
impaired the corrector ability (EC = 8.2 µM; Exp. pEC = 5.09).
5
0
50
1
9

ACCEPTED MANUSCRIPT
Surprisingly, the replacement of chlorine with fluorine in the meta position of the phenyl ring (4e
EC50 = 2.1 µM; Exp. pEC50 = 5.68) led to a high decrease in activity compared to the
aforementioned meta chlorine analog 4c. These results were consistent with the notion that at this
position the activity is somehow dependent on the size of the substituent (chlorine is bigger than
fluorine). Conceivably, the introduction of an additional fluorine atom in the molecule led to an
increase in hydrophobicity a characteristic convenient for corrector activity.
We also speculated whenever F508del-CFTR correction activity could be achieved with the
presence polar groups on meta or para-substituted phenyl ring, linked at the thiazole position 4. The
introduction of this moiety was hypothesized in relationship to the good predicted pEC values of
5
0
some of these. First, we investigated the effect of para-substitution with an electron-donating such a
hydroxyl in the compound 5a, where the effect of correction was good (EC =1.8 µM; Exp. pEC
50
5
0
=
5.74). Conversely, the presence in 5b of the same substituent in meta caused a slight decrease in
activity (EC =2.3 µM; Exp. pEC = 5.64).
5
0
50
Interestingly, a small group such as the methyl in position 5 improved the activity, if compared to
the counterpart, only when accompanied by a hydroxyl group in meta (5c vs 5b), where the EC50
value increase from 1.2 µM (Exp. pEC = 5.92) to 2.3 µM (Exp. pEC = 5.64).On the other hand,
5
0
50
the same substitution was detrimental when involved the para position, as shown by 5d (EC = 3.9
5
0
µM; Exp. pEC = 5.41) if compared with the previously mentioned 5a.
50
In agreement with the previous observations, we synthesised two different analogs 5e and 5f
containing electron-withdrawing group like the carboxyl feature in the para position. Only the
compound with a methyl group in position 5, called 5f maintained adequate corrector ability (EC50
=
2.9 µM; Exp. pEC50 = 5.54) while the molecule 5e, where the substitution in position 5 was
absent, was less potent (EC >5 µM; Exp. pEC < 5.30).
50
50
The precise positioning of the hydrogen-bond acceptor on the phenyl ring resulted therefore to be
important in enhancing activity.
2
0

ACCEPTED MANUSCRIPT
Changing the polar moiety with an acetamide one caused a decrease in potency (see 5g; EC >5
5
0
µM).
According to our model, we inserted fluorine atoms in some derivatives, as the typical benzodioxole
moiety of VX-809, to compare our hybrids with respect to the fluorinate analogues. Five different
derivatives were synthesised. Two of these (6a and 6b; Exp. pEC = 5.42 and 5.27), exhibiting the
50
methyl in position 5 and one halogen on the aromatic ring in position 4, have a significantly worse
activity than the non-fluorinated, (6a EC = 3.8 µM against 3b, EC =1.9 µM and 6b EC =5.4
5
0
50
50
µM against 3c EC = 2.3 µM) .
5
0
In the other analogs where the methyl group was not present, like 6c, 6d and 6e (Exp. pEC50
.89,5.77 and 5.40), the introduction of two fluorine led to comparable or more effective
compounds with respect to the not fluorinate analogues 4c, 5a, 5b (Exp. pEC = 5.09, 5.74 and
=
5
50
5.64). Perhaps, the presence of two additional fluorine atoms in the backbone of the hybrids could
increase the hydrophobicity making it similar to the one with the methyl substituents.
To complete the studies, we synthesised a panel of compounds with particular features as the
removal of the thiazole ring. The replacement of the aminothiazole core proved to be tolerated when
the original scaffold was substituted with 3-methylthioaniline portion (7a with EC = 8 µM; Exp.
5
0
pEC50 = 5.10), or overall with quinolin-8-amine portion (7b with EC = 2.2 µM; Exp. pEC =
50
50
5.66). This modification, in fact fulfilled the pharmacophore requirements experienced by the so far
known correctors, such as two aromatic cores linked by one H-bonding function.
On the other hand, this hypothesis was also reinforced by non-activity detected in 7c derivative that
does not contain aromatic portions linked to the thiazole ring but a methyl in position 4 and an
acetyl group in 5.
We then selected ten best potent and effective compounds in terms of CFTR rescue to determine the
dose-response relationships (Figure 5). We treated F508del-CFTR expressing CFBE41o- cells for
24 h with selected compounds and we then determined mutant CFTR activity by using the YFP
functional assay (see Experimental section for details).
2
1

ACCEPTED MANUSCRIPT
Such data were in harmony with the previous information regarding possible substituents in
scaffold obtained by the QSAR analyses. The extrapolated biological activities of all these
derivatives ultimately are not much different from those parameters analyzed in Supporting
Information S5.
Figure 5. Effects of most active compounds on F508del-CFTR activity. The bar graph reports
normalized F508del-CFTR activity in CFBE41o-cells after 24 h treatment with the indicated
compounds at different concentrations. Activity was measured with the HS-YFP assay. The line
indicates the level of activity in cells treated with corrector VX-809 (1 µM), while the dashed line
indicates activity in cells treated with vehicle alone.
Considering that some of the AATs previously described [29] showed additivity with VX-809, we
evaluated whether the new derivatives displayed similar effect. Therefore, we treated F508del-
CFTR expressing CFBE41o- cells for 24 h with selected compounds (at 2, 10 or 25 µM) in
combination with VX-809 (1 µM) and we then evaluated rescue of mutant CFTR activity by using
the YFP functional assay. The results demonstrated that the new hybrid show no additive effect
with VX-809 (data not shown).
2
2

ACCEPTED MANUSCRIPT
We then considered rescue of processing defect biochemically by observing the electrophoretic
mobility of CFTR protein. In Western blots, CFTR protein is detected as two bands, named B and
C, of approximately 150 and 170 kDa, respectively. Band B corresponds to partially glycosylated
CFTR residing in the ER. Band C is instead the mature fully processed CFTR that has passed
through the Golgi. The prevalent form in cells expressing wild-type CFTR is band C. Lysates of
cells expressing F508del-CFTR show primarily band B, consistent with the severe trafficking defect
caused by the mutation (Fig. 6A). To evaluate the effect of hybrid compounds on CFTR
electrophoretic mobility, we treated F508del-CFTR/HS-YFP expressing CFBE41o- cells with
DMSO (vehicle alone) or test compounds or VX-809 (1 µM, as positive control). The following
day, cells were lysed and lysates were subjected to SDS-PAGE followed by western blotting.
Western blot images were analyzed with ImageJ software. For each lane, CFTR bands, analyzed as
ROI, were quantified after normalization for GAPDH to account for total protein loading.
Treatment of F508del-CFTR cells with corrector VX-809 significantly enhanced expression of
mature CFTR (band C), resulting in a change of the C band / B band ratio. Similarly, treatment with
some of the compounds resulted in a significant increase of the C band / B band ratio, although less
marked as compared to VX-809 (Fig. 6B).
2
3

ACCEPTED MANUSCRIPT
Figure 6. Biochemical analysis of the F508del-CFTR expression pattern. A. The figures show the
electrophoretic mobility of F508del-CFTR in CFBE41o- cells after 24 h treatment with indicated
compounds (2a: 2 µM; others: 5 µM) or vehicle alone (DMSO) or VX-809 (1 µM). Arrows indicate
complex-glycosylated (band C) and core-glycosylated (band B) forms of CFTR protein. B.
Quantification of CFTR bands. Data are expressed as C band / B band ratio normalized for the
value observed in cells treated with DMSO. Data are expressed as means ± SEM, n = 3 independent
experiments. Statistical significance was tested by parametric ANOVA followed by the Dunnet
multiple comparisons test (all groups against the control group). Symbols indicate statistical
significance versus DMSO: ** P < 0.01, * P < 0.05.
2.5. Analysis of biological activity of compounds as potentiators of F508del-CFTR gating
defect
In the evaluation of the all molecules we investigated the possibility that some hybrids compounds
could show dual activities of corrector and potentiator as some of thiazolic derivatives demonstrated
in our previous works [28-29].
2
4

ACCEPTED MANUSCRIPT
Therefore, all compounds were tested as potentiators after a 30 min stimulation in the presence of
forskolin (20 µM), on F508del-CFTR / HS-YFP expressing CFBE41o- cells, following rescue of
the mutant channel by low temperature incubation for 24 h. We found that all of these were
ineffective in the potentiator assay, with the exception of 5b, 5c and 7b. However, the efficacy of
these molecules was partial (nearly 50% of the maximal effect elicited by genistein of VX-770) and
their potency low, as the potentiator activity was observed only at high concentration (25µM) while
no activity was observed at the concentration of use of VX-770 (1 µM) (Fig. 7).
Figure 7. Activity of all compounds as potentiators. The bar graphs report F508del-CFTR activity
in CFBE41o- cells after acute treatment with the indicated compounds (2, 10 and 25 µM) in the
presence of forskolin (20 µM). Activity was measured with the HS-YFP assay. For each compound,
the activity was normalized for the activity detected in negative control (cells stimulated with
forskolin alone).The line indicates the level of activity in cells treated with potentiator VX-770 (1
2
5

ACCEPTED MANUSCRIPT
µM) and Genistein (25 µM) while the dashed line indicates activity in cells treated with forskolin
alone.
3
. Conclusion
Herein we reported a ligand-based approach including quantitative-structure activity relationship
QSAR), developed in order to decipher a limited number of chemical descriptors turning in the
rational design and optimization of compounds called VX-809 and thiazole-containing hybrids.
Thus this work showed for the first time, the effectiveness of the ligand based studies on finding
(
,
new CFTR modulators, performing mathematical predictive models on the corrector chemo-types
so far available. The derived model allowed us to guide and prioritize the most promising hybrids to
be synthesised in the following synthetic routes.
The ligand-based hybrid-driven approach demonstrated the effectiveness of VX-809-like chemo-
types, bearing the thiazole nucleus, as endowed with corrector activities. Indeed, we have identified
a number of compounds featuring comparable potency with respect to VX809, revealing those
feature of the thiazole core which could be further optimized for an improvement in the activity.
Among them, we identified compounds 2a-2d (Exp. pEC50 = 7.06, 6.52, 6.26 and 6.05) which
experienced a higher potency than the prototype VX809 (Exp. pEC = 5.59).
50
Unfortunately, no compounds showed additive effect with VX809. We suppose that the common
scaffold of all hybrids and VX-809 i.e. benzo [1, 3] dioxol-5-yl-cyclopropanecarboxylic moiety can
aim molecules over the same target and in that case prevent the combined action of the two
substances.
Notably, the QSAR model efficiently predicted the biological trend of these newly synthesised
derivatives (Pred. pEC50 = 7.21, 7.17, 7.14 and 6.97), being therefore worthy of a further
development and optimization process towards the design of novel correctors.
2
6

ACCEPTED MANUSCRIPT
4
4
. Experimental Protocols
.1. Chemistry
Reagents and solvents were purchased from Sigma Aldrich, Alfa Aesar and VWR and used as
received unless otherwise indicated. Solvent removal was accomplished with a rotary evaporator at
ca.10-50 Torr. The analytical instrument used was an Agilent 1100 high performance liquid
chromatograph (HPLC). The analytical HPLC column was a Waters µBondapack C18, 3.9 id x 300
mm length.
The preparative HPLC Agilent 1260 Infinity preparative HPLC and the column used for preparative
chromatography was a Phenomenex C18 Luna 21.2 x 250 mm length. The analysis of the
intermediates and the raw products was performed by liquid chromatography-electrospray mass
spectrometry (HPLC–ESI–MS) using an Agilent 1100 series LC/MSD ion trap instrument.
HRMS experiment were performed using Q Exactive Orbitrap instrument by Thermo Scientific.
The nuclear magnetic resonance (NMR) spectrometer was a Varian Gemini 200 MHz
The proton spectra were acquired at 200 MHz while carbon spectra were acquired at 50 MHz, at
room temperature. Chemical shifts are reported in δ units (ppm) relative to TMS as an internal
standard. Coupling constants (J) are reported in Hertz (Hz).
All the raw powders obtained were purified with preparative HPLC using the following gradient:
from 0 to 5 min at 20% eluent B, then from 5 min to 40 min to 100% eluent B, from 40 to 45 min at
1
00% eluent B. Eluent A was water + 0.1% formic acid (FOA) and eluent B was acetonitrile +
.1% FOA. All analogues submitted for testing were judged to be of 95% or higher purity based on
0
analytical HPLC/MS analysis.
Compound purity was determined by integrating peak areas of the liquid chromatogram, monitored
at 254 nm.
2
7

ACCEPTED MANUSCRIPT
4.1.1. Synthesis of benzo [1, 3] dioxol-5-yl-cyclopropanecarboxylic acid (1a)
A mixture of benzo [1, 3] dioxole-5- carbonitrile (800 mg, 5 mmol), 1-bromo-2-chloro-ethane (1.4
ml, 17 mmol), and benzyltriethylammonium chloride (23 mg, 0.1 mmol) was heated to 75 °C and
then 50 percent (wt. /wt.) aqueous sodium hydroxide (5 mL) was slowly added. The reaction was
stirred at T= 75 °C for 24 hours. After this time 1-bromo-2-chloro-ethane (700 µl, 8.5 mmol) and 50
percent (wt. /wt.) aqueous sodium hydroxide (1 mL) were added to insure complete formation of
the cyclopropyl moiety (about 12 hours). The reaction was then heated to T=150 °C for about 48
hours to insure complete conversion from the nitrile to the carboxylic acid. When the hydrolysis
was complete (controlled with HPLC-MS) the dark brown mixture was diluted with water (10 ml)
and extracted three times with equal volumes of dichloromethane to remove all the sub-products.
The basic aqueous solution was acidified with concentrated hydrochloric acid to pH=2. The
reaction mixture was centrifuged at 4000g for 5 minutes and the pellet was washed with 1 M
hydrochloric acid (3 x 1 ml). The solid material was dissolved in dichloromethane (5 ml) and
extracted twice with equal volumes of 1 M hydrochloric acid. The organic phase was dried over
Na SO and concentrated under reduced pressure. The product was purified by preparative HPLC to
2
4
afford the title compound (670 mg, 65%).
1
H NMR (200 MHz, CDCl ) 6.94-6.81 (m, 3H), 5.97 (s, 2H), 1.73-1.61 (m, 2H), 1.38-1.19 (m, 2H).
3
1
3
C NMR (50 MHz, CDCl ) δ, 179.9, 146.2, 145.8, 131.4, 122.5, 110.1, 106.9, 100.0, 27.5, 16.6
3
+
HRMS (ESI) calculated for C H O : 207.0657 [M + H] found 207.0651
1
1
11
4
4.1.2. Synthesis of 1-(2, 2-difluorobenzo [1, 3] dioxol-5-yl) cyclopropanecarboxylic acid (1b)
To a solution of 2, 2-difluorobenzo [1, 3] dioxol-5-yl-methanol (930 mg, 5 mmol) in anhydrous
THF (5 ml) was added NaBH (208 mg, 5.5 mmol) in portions at T=0°C in ten minutes. The
4
mixture was then stirred at the same temperature for about 2h (controlled with HPLC). H O (1.5 ml)
2
was added and the mixture was stirred for about 15 min. The mixture was then extracted with
2
8