Enabling Efficient Positron Emission Tomography (PET) Imaging of Synaptic Vesicle Glycoprotein 2A (SV2A) with a Robust and One-Step Radiosynthesis of a Highly Potent 18F-Labeled Ligand ([18F]UCB-H)

Article abstract of DOI:10.1021/acs.jmedchem.6b00905

We herein describe the straightforward synthesis of a stable pyridyl(4-methoxyphenyl)iodonium salt and its [¹⁸F] radiolabeling within a one-step, fully automated and cGMP compliant radiosynthesis of [¹⁸F]UCB-H ([¹⁸F]7), a PET tracer for the imaging of synaptic vesicle glycoprotein 2A (SV2A). Over the course of 1 year, 50 automated productions provided 34 ± 2% of injectable [¹⁸F]7 from up to 285 GBq (7.7 Ci) of [¹⁸F]fluoride in 50 min (uncorrected radiochemical yield, specific activity of 815 ± 185 GBq/μmol). The successful implementation of our synthetic strategy within routine, high-activity, and cGMP productions attests to its practicality and reliability for the production of large doses of [¹⁸F]7. In addition to enabling efficient and cost-effective clinical research on a range of neurological pathologies through the imaging of SV2A, this work further demonstrates the real value of iodonium salts for the cGMP ¹⁸F-PET tracer manufacturing industry, and their ability to fulfill practical and regulatory requirements in that field.

Full text of DOI:10.1021/acs.jmedchem.6b00905

Subscriber access provided by Northern Illinois University
Article
Enabling efficient PET imaging of Synaptic Vesicle
glycoprotein 2A (SV2A) with a robust and one-step
18
18
radiosynthesis of a highly potent F-labelled ligand ([F]UCB-H)
Corentin Warnier, Christian Lemaire, Guillaume Becker, Guillermo Zaragoza, Fabrice Giacomelli, Joël
Aerts, Muhammad Otabashi, Mohamed Ali Bahri, Joël Mercier, Alain Plenevaux, and André Luxen
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.6b00905 • Publication Date (Web): 06 Sep 2016
Downloaded from http://pubs.acs.org on September 9, 2016
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Journal of Medicinal Chemistry is published by the American Chemical Society. 1155
Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 44
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Enabling efficient PET imaging of Synaptic Vesicle
glycoprotein 2A (SV2A) with a robust and one-step
radiosynthesis of a highly potent 18F-labelled ligand
([18F]UCB-H)
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Corentin Warnier†*, Christian Lemaire†, Guillaume Becker†, Guillermo Zaragoza‡, Fabrice
Giacomelli†, Joël Aerts†#, Muhammad Otabashi¤, Mohamed Ali Bahri†,Joël Mercier, Alain
Plenevaux†, and André Luxen†
† GIGA Cyclotron Research Centre In Vivo Imaging, University of Liege (ULg), 4000 Liege,
BELGIUM. ‡ Unidad de RX, Universidade de Santiago de Compostela, 15782 Santiago de
Compostela, SPAIN. # INSERM U1148, 75018 Paris, France. ¤ Trasis SA, 4430 Ans,
BELGIUM.  UCB Pharma, 1420 Braine-l’Alleud, BELGIUM.
Keywords : [18F]UCB-H, fluorine-18, 18F, iodonium, SV2A, PET.
Abstract: We herein describe the straightforward synthesis of a stable pyridyl(4-
methoxyphenyl)iodonium salt and its [18F]radiolabelling within a one-step, fully automated and
cGMP compliant radiosynthesis of [18F]UCB-H ([18F]7), a PET tracer for the imaging of
Synaptic Vesicle glycoprotein 2A (SV2A). Over the course of one year, 50 automated
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 44
1
2
productions provided 34±2% of injectable [18F]7 from up to 285 GBq (7.7 Ci) of [18F]fluoride in
50 minutes (uncorrected radiochemical yield. Specific Activity = 815±185 GBq/µmol). The
successful implementation of our synthetic strategy within routine, high-activity and cGMP
productions attests to its practicality and reliability for the production of large doses of [18F]7. In
addition to enabling efficient and cost-effective clinical research on a range of neurological
pathologies through the imaging of SV2A, this work further demonstrates the real value of
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
18
iodonium salts for the cGMP F-PET tracer manufacturing industry, and their ability to fulfill
practical and regulatory requirements in that field.
INTRODUCTION
[18F]UCB-H1,2 ([18F]7, Figure 1) is a novel Positron Emission Tomography (PET) tracer that
has nanomolar affinity for the Synaptic Vesicle glycoprotein 2A. The pivotal role of SV2A in
neurotransmission processes and its proven implication in epilepsy make it a highly interesting
PET target for the study of various neurotransmission-related diseases, such as epilepsy and
Alzheimer’s disease.3–5
Figure 1. Two SV2A ligands: the anti-epileptic Levetiracetam and the PET tracer [18F]7.
Epilepsy alone is a neurological pathology that affects over 50 million people around the globe
at any given time,6 and yet the cellular processes it originates from still remain vastly unknown
and unexplored. The ubiquity of SV2A in the human brain, however, offers a unique opportunity
ACS Paragon Plus Environment

Page 3 of 44
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
to peek into the very mechanisms of neurotransmission and improve the understanding,
diagnosis and treatment of epilepsy and other neurological conditions using PET technology.
PET has indeed become a corner stone tool for (pre-)clinical research and routine diagnosis over
the past decades;7–10 with its convenient 110 min half-life, almost exclusive + decay and the
significant representation of fluorine in pharmacologically active molecules,11 fluorine-18, in its
no-carrier-added, nucleophilic ([18F]F-) form, is one of the most practical nuclides currently in
use for clinical PET imaging.12
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
To this day, however, designing an efficient and robust automated cGMP production method
for an aromatic [18F]-labelled compound such as [18F]7 remains a challenge, due to the unique
reactivity of fluoride ion. A strong electron withdrawing group in ortho or para of an appropriate
+
leaving group (typically -N(Me)3 or -NO2) is often a sine qua non condition for the efficient
nucleophilic fluorination of an aromatic ring, which severely limits the substrate scope of the
reaction. This approach was used in the first radiosynthesis of [18F]7 (Scheme 1), for its
assessment as an efficient PET imaging agent of SV2A.4 The relative complexity of the
chemistry involved makes the multi-step synthesis very challenging to automate, and restricts its
radiochemical yield and its robustness. Because of this poor synthetic accessibility, the syntheses
of the 11C-derived analogs of [18F]7, [11C]UCB-A and [11C]UCB-J, were recently developed (see
Table 2).13,14 The very short half-life of carbon-11 (~20 min) however implies less flexibility in
terms of synthesis and medium-range delivery to PET centers and hospitals, and [18F]-labelled
analogs are much preferred when possible. Considering the potential of [18F]7 in neurological
research and the increasing demand for it in clinical trials, the development of a robust and
effective synthesis was undertaken.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 44
1
2
Scheme 1. The multi-step radiosynthesis first used for the production of [18F]7, and the novel
3
4
5
6
strategy presented in this work.
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Overcoming the intrinsic limitations of nucleophilic aromatic substitution has been the driving
force of multiple research projects for the past few years. Several publications reported the
development of a variety of novel precursors that allow the late-stage radiolabelling of electron-
rich aromatic rings, thereby potentially enabling considerable upgrades in the synthesis of
[18F]fluoroarene-derived radiotracers. Iodonium salts,15 iodonium ylides,16–18 boronic esters,19
palladium20 or nickel21 complexes have been proposed as versatile precursors for direct
nucleophilic aromatic radiofluorination. However, some of those precursors have, at this stage of
their development, physico-chemical properties that make them less suited than others for
automated and cGMP productions of PET tracers. Iodonium salts, for their part, have repeatedly
been shown to be efficient and stable late-stage radiofluorination precursors from their first
reported 18F-radiolabelling in 1995, which prompted us to develop such an iodonium-based
precursor for the one-step radiosynthesis of [18F]7.22–38 To this day, little information has been
made available on the routine, high-activity [18F]-labelling of iodonium salts in the framework of
cGMP productions; filling that gap was also one of the objectives of this project. We also tackled
what may have represented a major drawback of iodonium [18F]-radiolabellings in a cGMP
ACS Paragon Plus Environment

Page 5 of 44
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
perspective, which is the use of the genotoxic radical scavenger TEMPO ((2,2,6,6-
Tetramethylpiperidin-1-yl)oxy) for controlling the radical side-reactions occurring during the
radiolabelling, and increasing the yield and reproducibility thereof.31,39,40
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
RESULTS & DISCUSSION
Precursor synthesis & characterization
Although a number of reactions allow the synthesis of diaryliodonium salts,30,41–46 few are
compatible with the synthesis of pyridyl(aryl)iodoniums. Recent developments in that field now
allow their one-step synthesis from the corresponding iodinated heterocyle.41,42 The method only
requires common and fairly inexpensive reagents, and no transition metal compound is needed.
In a cGMP perspective, the absence of metallic impurities simplifies the mandatory quality
control of the obtained precursor. Our synthetic strategy for pyridyl(4-methoxyphenyl)iodonium
salt 6 is depicted in Scheme 2.
Scheme 2. Synthesis of 6 from compounds 1 and 4a. The overall yield exceeds 30%.
We started from commercial methyl-3-aminoisonicotinate 1, to form 2 through a modified
Sandmeyer reaction followed by reduction with DIBAL at -96 °C to yield aldehyde 3.47,48
Special attention was dedicated to the temperature control of the 2 → 3 reduction step
(CH2Cl2/N2 cooling bath), as complete reduction to the alcohol analog of 3 and
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 44
1
2
3
4
5
6
7
8
9
hydrodehalogenation by-products became predominant at higher temperatures. Maximal
reproducibility was observed when using methanol to neutralize the excess of DIBAL, which
allowed the quenching to occur at -96 °C in a homogeneous medium. The following three-step,
one-pot reaction converts 3 into 5 through a reductive amination with 4b followed by an
intramolecular cyclisation in basic conditions. The last step (5 → 6) requires strongly oxidizing
conditions (4 eq. of triflic acid on 5 followed by the addition of 2.5 eq. of m-CPBA), but
proceeds with high and reproducible yields (75-80%). The overall synthesis proceeds with a very
good yield (>30% overall) and does not involve any semi-prep HPLC purification, which makes
it as practical as it is effective. It is noteworthy that all the reaction steps were optimized at least
at a 10 mmol scale, except the last one (5 → 6) that yields enough precursor (1.3 g) for more
than 80 radiolabellings at a 2.5 mmol scale. The novel compound 6 was fully characterized
(melting point, CHNS analysis, HRMS, NMR and X-Ray crystallography - Figure 2), and a
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
validated cGMP H-NMR quantification method was also developed for the assessment of its
absolute purity (measured purity: 99.80.5 wt.%).
Figure 2. ORTEP representation of 6 with thermal ellipsoids drawn at the 50% probability level.
Hydrogen atoms and counter-ion molecules were removed for clarity.
ACS Paragon Plus Environment

Page 7 of 44
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
In order to avoid a potentially costly enantiomeric resolution of the final product, the retention
of the initial enantiomeric excess (ee) of 4a during the whole synthetic process of 6 was of prime
importance. By synthesizing 5 and 6 with both enantiomerically pure 4a and racemic (R,S)-4a,
we demonstrated that the critical basic cyclisation step (3 → 5, scheme 2) did not cause any
detectable racemization in compound 5. We also showed that neither the last step (5 → 6), nor
the radiolabelling conditions cause any measurable racemization of the stereogenic center
(comparison of UV and radio chromatograms by chiral HPLC. See SI and Figure 3).
The pyridyl(4-methoxyphenyl)iodonium salt 6 additionally displayed an exemplary stability in
standard atmospheric conditions, which is a highly valuable feature in a routine and cGMP
perspective. The projections of the ongoing stability study tend to confirm a precursor shelf-life
of 2 years (storage conditions: crimp-capped amber vial kept at 5 - 8 °C).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Radiochemistry
In order to probe our precursor’s inclination to undergo the expected radiofluorination, low-
activity labelling tests were carried out, using a Waters® 46 mg QMA carbonate for [18F]fluoride
trapping and a K222/K2CO3-based QMA eluent for the subsequent [18F]fluoride elution and
azeotropic drying in the reactor. A solution of 6 and TEMPO in a polar aprotic solvent was
added to the dry fluoride residue, and the reactor was then sealed and heated up to the given
temperature and during the time given in Table 1. It appeared that usual concentrations of
18
cryptand-2.2.2 (K222) and potassium carbonate in the QMA eluent were ineffective for the F-
labelling of the iodonium precursor 6 (Table 1, Entry 1). The precursor indeed displayed poor
chemical stability in the presence of high concentrations of base (6.6 mg/mL K2CO3), which in
turn led to poor radiochemical yields. We therefore tested the influence of the K2CO3 and K222
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 44
1
2
3
4
concentrations in the QMA eluent on the radiochemical yield and on the stability of the
5
6
precursor.
7
8
9
Table 1. Optimization of the labelling conditions of the pyridyl(4-methoxyphenyl)iodonium
precursor 6.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
QMA eluent
Reaction conditions[b]
RCY
d.c.[c]
Entry
K2CO3 K222 TEMPO Solvent T-time
/mg[a] /mg[a]
/mg
(1 mL)
/°C-min
6.6
1.3
0.6
0.6
0.6
0.6
0.6
20
4
4.3
4.3
4.3
2.3
6.8
0
DMF
DMF
DMF
DMF
DMF
MeCN
MeCN
120 – 10
100 – 5
<1%
5%
1
2
3
4
5
6
7
12
12
12
12
12
125 – 10
125 – 10
125 – 10
125 – 10
125 – 10
20%
16%
16%
<1%
50%
4.3
[a] Amount contained in the QMA eluent (600 µL of a 90/10
MeOH/H2O solution). [b] Using 15.5mg of 6. [c] Decay-
corrected radiochemical yield measured by radio-TLC,
corrected for [18F]fluoride adsorption on the reactor walls.
Using qualitative UV-TLC monitoring and quantitative radio-TLC monitoring, we found that
low concentrations of K2CO3 enabled a better precursor lifetime in the reaction medium, thus
allowing for more efficient radiolabellings (Entries 2-3). Diaryliodonium salts are known to
decompose into radical intermediates upon heating,40 leading to undesired by-products and to
low, fluctuating radiochemical yields. The radical scavenger TEMPO was previously reported to
be a very efficient agent for the enhancement of diaryliodonium [18F]radiolabellings.39 In our
case, it proved necessary for the efficient radiofluorination of 6; 1 to 1.2 equivalent of TEMPO
(with respect to 6) provided the most efficient RCY (Entries 3-7), using MeCN as the solvent of
the reaction (Entry 7). The optimization of the [18F]labelling parameters led to the following
ACS Paragon Plus Environment

Page 9 of 44
Journal of Medicinal Chemistry
1
2
3
4
5
6
conditions, which were subsequently used for the automated productions: 15.5±0.5 mg of
precursor and 4.3±0.2 mg of TEMPO were solubilized in 1 mL of dry acetonitrile and heated for
7
8
9
18
10 minutes at 125 °C in the presence of the azeotropically dried F-fluoride, affording [18F]7 at
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
50% RCY. The UPLC retention time of the obtained labeled compound was compliant with that
of the cold reference [19F]7. LC-MS detection of the cold peak associated with the radioactive
peak (Observed m/z = 325 = M+1) further confirmed the identity of the radiotracer. The
enantiomeric excess of the stereogenic center was left unchanged by the radiolabelling
conditions, as shown in Figure 3. Additionally, although non-symmetric iodonium salts are
known to undergo unselective radiofluorination in many cases,38 we found that in our conditions,
the radiolabelling was strictly regioselective for the pyridyl moiety of the pyridyl(4-
methoxyphenyl)iodonium salt 6 (no [18F]4-Fluoroanisole was detected in the crude reaction
mixture, see Supporting Information).
The purity of the iodonium precursor was found to be determinant of the radiochemical yield.
Especially, the deprotonation step of the iodonium 6 on a basic alumina plug (which is a part of
the work-up of its synthesis, see Experimental Section), should be carried out by strictly
following the reported procedure.42 While the incomplete deprotonation of the obtained
iodonium cannot usually be detected on classical NMR spectra, it nevertheless strongly
undermines the [18F]labelling yield, due to the presence of labile, nucleophilicity-poisoning
protons. For instance, a precursor batch incorrectly submitted to the deprotonation step afforded
1
13
significantly lower radiochemical yields (33% instead of 50%), although its H- and C-NMR
spectra were strictly identical to other functional precursor batches. A second, more careful
deprotonation procedure was performed on the defective batch, and the RCY was restored to its
nominal value. Besides, the crystallization method for the precursor 6 presented in this work has
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 44
1
2
3
4
5
6
7
8
9
been extensively optimized and should also be followed closely, as it ensures obtaining a pure
and highly soluble crystalline form of 6, which in turn facilitates its solubilization during the
automated radiosynthesis.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 3. Chiral HPLC analysis of [19F]7 (blue trace), and of [18F]7 obtained from the
radiolabelling of 6 and (R,S)-6 (green and red traces respectively). Only the expected R
enantiomer of [18F]7 is produced during the high-activity radiolabelling of 6 (no measurable
racemization, green radio trace, n = 3).
Automation & productions at high activity
The radiosynthesis was automated on a Trasis All-In-One synthesizer equipped with an
integrated HPLC purification system (Figure 4). The automated sequence is constituted of the
following main steps: 1) Machine and cassette tests, HPLC and C18 formulation cartridge
conditioning, preparation of the final formulation and precursor solubilization (pre-synthesis, 30
ACS Paragon Plus Environment

Page 11 of 44
Journal of Medicinal Chemistry
1
2
min). 2) 18F-Fluoride recovery, trapping, elution and drying (7 min). 3) Precursor transfer to the
reactor, heating and radiolabelling (13 min). 4) Cooling of the reactor, dilution of the crude and
injection on the semi-preparative HPLC (3 min). 5) HPLC purification and collection of the
purified radiotracer (20 min). 6) Formulation of the radiotracer into an injectable solution (8
min). In the Experimental Section, we describe more precisely the string of events constituting
the radiosynthesis. Extensive description of the cassette and of the materials used throughout the
synthesis may be consulted in the Supporting Information and in the Experimental section.
Figure 4. Graphical representation of the cassette designed for the automated synthesis of [18F]7
on a Trasis AIO synthesizer.
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 44
1
2
3
4
5
6
7
8
9
The automation allowed us to perform syntheses at high levels of activity and assert that the
RCY remained constant at every tested starting activity (tested up to 7.7 Ci, or 285 GBq). A total
of 50 fully automated productions, 40 of which were performed at high activities (typical starting
activity = 110 GBq), yielded a narrow and reproducible 34±2% uncorrected RCY in [18F]7 with
high specific activity (815±185 GBq/µmol, n = 8). The RCY of the automated synthesis was
defined as the ratio of the activity of [18F]7 present in the injectable solution at EOS (End Of the
Synthesis) to the activity trapped on the QMA at SOS (Start Of Synthesis, i.e. 50 minutes
before). The obtained regularity in the automated synthesis RCY’s during a one-year production
campaign with 4 different batches of precursor illustrates the robustness and reliability of this
production method. In addition, the single reaction step renders the automation development
quite straightforward in its general principle; this synthetic strategy may therefore be adapted on
virtually any commercial synthesizer.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The Radiolysis Issue
While the amount of activity present in the reaction medium had no effect on the labelling
yield, it had an important consequence on the radiochemical stability of [18F]7 in the final
formulation. [18F]7 was indeed found to be more sensitive to radiolysis than other PET tracers
routinely synthesized in our production center, such as [18F]FDG, [18F]FDOPA or [18F]FTYR.49–
51
For instance, the formulation of 19 GBq of [18F]7 in a 5% EtOH saline solution promptly led
to the radiolysis of the tracer into multiple, unidentified radiolabelled compounds. After 4 hours,
the radiolytes amounted for almost 70 % of the total activity in the bulk (Figure 5, trace A). A 3
to 5% radiolysis was also observed in similar formulations containing as little as 3.7 – 5.5 GBq
of [18F]7. Taking a 100 µL aliquot from any of the tested formulations at EOS and letting it age
separately from the bulk resulted in a high and constant radiochemical purity (RCP) of [18F]7 in
ACS Paragon Plus Environment

Page 13 of 44
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
the aliquot. This confirmed that the observed degradation was caused by the radioactivity of the
sample, and not only by its chemical composition. It also indicated that the total activity in the
formulation (i.e. the dose rate inside the solution) impacts the radiolysis kinetics more than the
activity concentration alone, as should be expected. As a result, dilution of the formulation alone
could not completely solve the radiolysis problem, so we moved to a chemical stabilization of
the tracer. Ascorbic acid (HAsc) is an efficient and popular additive for the stabilization of
radiotracers at high activity; as summarized in Figure 5, different activity concentrations and
formulation compositions were tested in order to optimize the long-term (> 6 hours) stability of
[18F]7. 1 mg/mL of HAsc added to the formulation allowed [18F]7 to remain completely stable
for almost one hour. Following that short stability period, however, the radiotracer underwent a
fast and nearly quantitative radiolysis (Figure 5, trace B). An increase of the ascorbic acid
concentration, as well as the addition of a citrate buffer to increase the pH of the formulation led
to much more stable RCP’s.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 5. Evolution over time of the RadioChemical Purity (RCP) of [18F]7, at different activity,
stabilizer or buffer concentrations.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 44
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
A short, one-hour period of complete stability was however still followed by a measurable
RCP decrease (Figure 5, traces C and D). Those results brought to light a strong influence of the
pH on the radiolysis profile and kinetics. The replacement of ascorbic acid by sodium ascorbate
allowed the formulation of [18F]7 at nearly neutral pH (pH 6.75) and enabled a better
stabilization of the radiotracer, while suppressing the need for a buffer. For instance, 5 mg/mL of
sodium ascorbate in the bulk solution allowed the formulation of up to 37 GBq of [18F]7 with
RCP’s exceeding 97% over 6.5 hours (Figure 6).
Figure 6. UPLC traces showing the radiochemical purity (Top) and the chemical purity (Bottom)
of [18F]7 in formulations containing 33 GBq of the radiotracer. The peak saturating the UV
detector results from the presence of sodium ascorbate in the formulation.
ACS Paragon Plus Environment

Page 15 of 44
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The cause of the 2-3% radiolysis observed in the 33 – 37 GBq formulations was attributed to
the formulation on the C18 cartridge, during which all the activity is concentrated in a few cubic
millimeters or less. The extremely high dose rates the radiotracer is exposed to during this step
causes inevitable radiolysis that is negligible at low activities but increases with the total
formulated activity. Therefore, although we were able to produce 96 GBq of [18F]7 from 285
GBq of [18F]fluoride, the routine production of [18F]7 is currently limited to 37 GBq.
The TEMPO Issue
This radiosynthesis has an important feature, which is the use of TEMPO (Scheme 3), a free
radical and radical scavenger whose genotoxic properties were pointed out in a 2013 in vitro
toxicology study.52 As the presence of TEMPO is crucial for the efficiency of the
radiofluorination, its removal was ruled out. The presence of a toxic compound is a potentially
major drawback for a synthesis with intended clinical applications, since the synthesis must be
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 44
1
2
3
4
5
6
7
8
9
designed as to deliver a radiotracer formulation compliant with current regulations on the matter.
However, we hereby demonstrate our ability to provide, without any major consequence over the
synthesis design or on the Quality Control (QC) procedures, a formulation with a TEMPO
content 500 to 1000 times below the specification. To the best of our knowledge, the maximal
level of TEMPO tolerated in drugs is not clearly defined. We therefore followed the EMA
(European Medicines Agency) recommendation53 that refers to the ICH-Q3C guideline54 for the
determination of the maximal level of genotoxic impurities in drugs (ICH = International
Council for Harmonization). According to our interpretation of the ICH-Q3C guideline, the
tolerated concentration of TEMPO is 5.8 ppm, for a maximal injection volume of 10 mL in a 50
kg person (limit concentration in a worst-case scenario, see SI for detailed argumentation).
A few precautions were therefore taken during the design of the automation sequence, so that
the probability of any TEMPO contamination reaching the final formulation remained as low as
possible. For that purpose, the synthesis cassette was virtually divided in two parts. The left part
of the cassette, denoted M1 in Figure 4, is where the radiosynthesis takes place. From the
fluoride-18 recovery from the cyclotron through the HPLC injection with syringe S1, M1 is in
direct contact with the solution containing TEMPO. The right part of the cassette, denoted M2, is
where the purified radiotracer is collected from the HPLC and formulated into an injectable
solution. The automated sequence was designed so that no potentially TEMPO-contaminated
liquid was allowed to flow from M1 to M2 during the whole synthesis. An adequate washing
sequence of M1 after the HPLC injection also ensured that the pressure inlet (left side of M1)
could not blow contaminated leftover droplets from M1 to M2 during a necessary line flushing in
the formulation step. Assuming the efficiency of our automation design against TEMPO
contamination, the only remaining possibility for contamination was inefficient HPLC
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment

Page 17 of 44
Journal of Medicinal Chemistry
1
2
purification. The HPLC column and mobile phase were therefore chosen so that [18F]7 was
eluted before TEMPO, in order to avoid any TEMPO contamination due to potential peak tailing.
These two simple precautions proved to be effective, as the residual concentration of TEMPO in
the formulations remained 500 to 1000 times below the 5.8 ppm specification (See Figure 7 and
SI). Another less carefully developed sequence involved the HPLC injection of the crude and the
formulation of the purified compound with the same syringe (S2), featuring a simple ethanol and
saline wash of S2 between the two operations. As expected, that automated sequence resulted in
slightly higher and more scattered contamination with TEMPO (40±25 ppb, n = 5) in the
formulations. The contamination nevertheless remained comfortably below the 5.8 ppm limit,
demonstrating the efficiency of simple lines and syringe washes to almost quantitatively
eliminate TEMPO from the cassette (Figure 7). This synthesis may therefore also be
implemented on synthesizers that are less flexible in terms of cassette design.
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 3. TEMPO free radical and its oxydized and reduced forms TEMPO+ and TEMPO-H.
Figure 7. Residual TEMPO-H in [18F]7 bulk formulations using syringe S1 or S2 for the HPLC
injection of the crude radiolabelling product (see also Figure 4), as determined by LC-MS
analysis.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 44
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Easy GC-FID detection of TEMPO
The problematic of residual TEMPO in drugs has already been the subject of several
publications, and its peculiar red-ox properties are well known.55–58 In the presence of sodium
ascorbate, TEMPO is quantitatively reduced into its TEMPO-H form, which suppresses its UV
absorption band and excludes an otherwise very convenient UV detection.
Although we were very confident in the ability of our automation sequence to provide virtually
TEMPO-free [18F]7 formulations (Figure 7), the very nature of TEMPO’s toxicity imposed the
development of a routine test for a possible excessive TEMPO contamination. The use of LC-
MS is a fast and efficient way to accurately quantify TEMPO in our case. However, such
equipment is not available at every production center, and it adds one more step to the routine
QC. That is why we developed a GC-FID method for the routine compliance assessment of our
formulations of [18F]7. The developed method allowed reaching Limits of Detection (LOD) and
of Quantification (LOQ) that were well below the regulatory maximal concentration of 5.8 ppm:
LOD = 0.6 ppm, LOQ = 2 ppm with RSD’s < 5% (See Figure 8 and SI). The very low
probability of significant contamination with TEMPO in routine batches was an ideal context to
implement a Limit Test for the conformity assessment of the injectables.59
ACS Paragon Plus Environment

Page 19 of 44
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Figure 8. The implemented Limit Test for the assessment of the formulation’s compliance
towards TEMPO(-H) contamination features the GC-FID spectrum of a matrix solution spiked
with 5.8 ppm of TEMPO, and that of an injectable from an automated production of [18F]7. The
spectra are zoomed in on the relevant area for TEMPO-H detection. Result of the test: compliant.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Routine QC (quality control) & GMP compliance
The quality control of the obtained [18F]7 formulation is constituted of classic radio- and UV-
UPLC identification, (radio-)chemical purity assessment and specific activity measurement of
the labelled compound (see SI). The extremely low level of cold impurities in the formulations
(indicated by their UPLC analysis, see Figure 6) further demonstrated the efficiency of the
automated HPLC purification of [18F]7. The QC of the injectable includes radionuclide
identification, radionuclidic purity measurement, GC-FID Limit Test (for residual solvents and
TEMPO), osmolality, pH, bacterial endotoxin testing, and sterility testing, as for any other
cGMP drug. The detail of those quality control tests shall not be exposed here, as they are
common and employed in every cGMP process. Besides the extensive and routine post-
production QC, the rest of the radiosynthesis must also remain under proper control. The use of
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 44
1
2
3
4
5
6
7
8
9
single-use manifolds and of certified USP/Ph. Eur. reagents and solvents, the extensive
characterization of the precursor and the determination of its shelf life are some factors that help
guaranteeing the reproducibility and conformity of the radiosynthesis.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
PET Imaging
Figure 9. Top: representative images of [18F]7 microPET acquisitions. Left panel, image
obtained with the synthesis from this work. Right panel, image obtained with the former
synthesis. Bottom: Time activity curves (expressed as SUV, mean ± SEM) of [18F]7 from this
work (black symbols, n = 5) and from the former synthesis (open symbols, n = 5). Thal:
thalamus, Cb: cerebellum.
The formulation of [18F]7 resulting from this radiosynthesis (5 mg/mL NaAsc in saline, no
buffer) differs from the formulation that was generated with the former synthesis (see Scheme 1).
ACS Paragon Plus Environment

Page 21 of 44
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
The former formulation did not contain any stabilizer (the low activities produced did not impose
any stabilization) and contained a citrate buffer. In order to assess that the difference in
formulation or in cold impurity profile did not negatively impact the quality of the PET images,
we performed test PET scans in rodents that allowed us to compare the two [18F]7 formulations.
The results presented in Figure 9 clearly show that [18F]7, regardless of the formulation
composition, displayed high brain uptake and yielded distribution patterns that were highly
consistent with previous work.4 The time activity curves (TACs) extracted from thalamus and
cerebellum were very close between the “new” and the “old” synthesis.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
CONCLUSION
Primarily focused on the practicality of the developed precursor synthesis and automated
radiosynthesis, this work covers all relevant aspects of the high-activity and cGMP production of
up to 37 GBq of [18F]7. With an entirely new synthetic approach involving the one-step
radiolabelling of a pyridyl(aryl)iodonium salt, our method ensures more than doubling the
formerly possible production yield (Scheme 1) while shortening its duration from 120 to 50
minutes. We also demonstrate for the first time the ability of an iodonium salt to undergo
efficient [18F]radiofluorination at high activity (up to 285 GBq) without any decrease of the
radiochemical yield. Although the production of more than 37 GBq of [18F]7 is challenged by the
sensitivity of this particular tracer towards radiolysis, it still attests that iodonium salts, with their
usually remarkable stability, are more than ever promising precursors for applied late-stage
aromatic radiofluorination of other key PET tracers.
In the context of the necessity to develop a simple and practical automated radiosynthesis, all
the materials and apparatus involved in our production method were chosen for both their
efficiency and their standard usage in routine production of other common PET tracers. We also
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 44
1
2
3
4
5
6
7
8
9
developed a simple QC Limit Test using GC-FID detection in response to the issue raised by the
use of genotoxic TEMPO. In addition, we showed that the residual TEMPO concentrations in the
injectable could be kept 500 to 1000 times below the specification by simply using a thoughtful
and logical design of the automated sequence. This production method will soon replace the
former synthetic strategy (Scheme 1) for the production of [18F]7 for human use in our research
center. The first human PET images acquired thanks to this method shall be published elsewhere.
In this work, multiple chemical, physical and analytical challenges were overcome in order to
make an iodonium-based synthesis concept viable in the context of automated, routine
productions in a cGMP environment. By deliberately exposing all encountered weak and strong
points of iodonium salts in that respect, we provide key solutions that will enable other scientific
teams to develop their own efficient productions of PET tracers of interest for human use using
iodonium-based methods. More specifically, through providing PET centers with a simple, one-
step and highly efficient synthetic method of a SV2A-specific radiotracer (Table 2), we trust that
this work will significantly promote clinical research on most widespread neurological
conditions, such as epilepsy and Alzheimer’s disease.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table 2. Contribution of this work to the synthesis of relevant SV2A-specific PET tracers. Data
extracted from1,13,14
.
ACS Paragon Plus Environment

Page 23 of 44
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
EXPERIMENTAL SECTION
Materials & Methods
Solvents and chemicals were purchased from VWR and Sigma Aldrich and used without
further purification, unless otherwise stated. TEMPO was purchased as “purified by sublimation”
grade. Racemic compound (R,S)-4a was purchased from UOrsY sales EU. Cold reference [19F]7
and 4a were provided by UCB Pharma (Belgium). Methyl-3-aminoisonicatinate 1 was purchased
from KeyOrganics (UK). NMR spectra were acquired on a 400 MHz Bruker spectrometer and
1
processed with MNova 10. The H-NMR cGMP quantification method of 6 was developed on a
500 MHz Bruker spectrometer and maleic acid was used as a reference. The experimental details
18
will be published elsewhere. An IBA Cyclone 18/9 cyclotron was used to produce F-fluoride
18
through the O(p,n)18F reaction in a Niobium target containing 3.5 mL of 18O-enriched water
from Rotem Industries Ltd. The synthesis of [18F]7 was automated on a Trasis All-In-One
synthesizer, and a Phenomenex Luna C18 Column (5 µm, 10 x 250 mm) was used for the semi-
prep purification of the radiotracer. UV and radio chromatograms were recorded on a Waters®
Acquity UPLC, using a BEH C18 Column (1.7 µm, 2.1 x 50 mm) and an Eberline NaI gamma
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 44
1
2
3
4
5
6
7
8
9
detector. The chiral HPLC analyses for the measurement of the radiotracer’s ee were carried out
with an analytical Daicel Chiralpak® AD-RH column, using a 35/65 MeCN/pH 8 phosphate
buffer as eluet at 0.3 mL/min. Mass spectra were acquired on a ESI-MS Waters® TQD coupled
with a Waters® H-Class UPLC system. High-resolution mass spectra were acquired on a Bruker
SolariX ESI-FT-ICR mass spectrometer. Gas Chromatograms were acquired on a
ThermoScientific Trace 1310 gas chromatograph fitted with a PAL autosampler, using a 20-
meter Restek stabilwax crossbond carbowax PEG column (0.18 mm ID, 0.18µm df) and FID
detection. Extensive information on the GC injection parameters is provided in the Supporting
Information document. Melting points were measured on a Büchi B-545 device. The CHNS
analysis of compound 6 was performed twice on an Intersciences Flash EA 1112 apparatus, and
the related weightings made on a Mettler Toledo MX5 microbalance. The Xray diffractogram of
compound 6 was recorded on a Bruker APPEX-II CCD diffractometer. Extensive data and
information on the X-ray diffractogram acquisition are provided in the Supporting Information
document, and the CIF file was deposited with Cambridge Crystallographic Data Centre
(n°1480740). For the sake of graphical consistence and legibility, raw data sets from HPLC and
GC detectors were plotted using Igor Pro. Raw data was not submitted to any transformation.
Chemistry
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Methyl 3-iodoisonicotinate (2)
Procedure adapted from 47. To a round-bottom flask (500 mL) were added acetonitrile (120
mL), methyl 3-aminoisonicotinate 1 (5 g, 33 mmol) and a magnetic stirrer. The stirred mixture
was cooled down to -10 °C. Trifluoromethanesulfonic acid (11.7 mL, 132 mmol, 4 eq.) was then
slowly added, using a glass syringe fitted with an inox needle. A solution of water (12 mL)
containing sodium nitrite (4.78 g, 69 mmol, 2.1 eq.) and potassium iodide (14.24 g, 85 mmol, 2.6
ACS Paragon Plus Environment

Page 25 of 44
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
eq.) was then added dropwise (1.5 mL/min) to the reaction medium. The mixture was then
removed from the cooling bath, and stirred for 1h15 at room temperature. The reaction was
monitored by TLC (80/20 ethyl acetate/hexanes). Water (300 mL) was added to the reaction
mixture, and solid sodium hydrogenocarbonate was then added until a slightly basic pH was
reached. The opaque brown solution became clear-orange upon addition of a sat. solution of
sodium thiosulfate (30 mL). The resulting solution was extracted with cyclohexane (6 x 100
mL), which is a selective solvent for the desired compound. The combined organic fractions
were dried over magnesium sulfate and evaporated under reduced pressure to yield and orange
oil that crystallizes quickly upon cooling. The obtained beige solid was redissolved in a
minimum of cyclohexane and evaporated under reduced pressure to yield the desired compound
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
as beige crystals (needles) (7.3 g, 28 mmol, 85 %). Melting point and H NMR spectrum were
consistent with literature data.60
1H NMR (400 MHz, Chloroform-d)  9.11 (s, 1H), 8.62 (d, J = 4.9 Hz, 1H), 7.64 (d, J = 4.9
Hz, 1H), 3.97 (s, 3H).
13C NMR (101 MHz, Chloroform-d)  165.48, 159.75, 149.21, 142.17, 124.60, 92.61, 53.12.
Melting point: 43 – 44 °C (cyclohexane)
ESI-MS: Observed m/z = 264 (M+1) Calculated: 263.95
3-Iodoisonicotinaldehyde (3)
To a dry round-bottom flask (100 mL) were added a magnetic stirrer, dry toluene (45 mL,
distilled over sodium) and 2 (2.5 g, 9.5 mmol). The reaction mixture was cooled down to -96 °C
(Dichloromethane/liquid nitrogen cooling bath) and a 1 M solution of di-isobutylaluminium
hydride in toluene was added dropwise at a rate of 1.5 mL/min (19 mL, 19 mmol, 1.75 eq). The
reaction mixture was allowed to stir 15 min at -96 °C after the addition was completed. The
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 44
1
2
3
4
5
6
7
8
9
reaction was then quenched (still at -96 °C) by the slow addition of methanol (5 mL). The first
drops were added carefully, as the quenching caused a substantial foaming of the mixture. More
methanol was then added (45 mL), and the reaction medium was removed from the cooling bath
and allowed to stir at RT for 2 h. The crude mixture was poured into a beaker and placed on a 0
°C cooling bath. 4 M HCl (25 mL) was slowly added into the solution under vigorous magnetic
stirring. After the addition was completed, the beaker was removed from the cooling bath and
allowed to stir vigorously at RT for 1 h. Distilled water (30 mL) was added, and a concentrated
solution of K2CO3 was subsequently poured into the mixture until a basic pH was reached in the
aqueous phase. The resulting mixture was extracted with ethyl acetate (3 x 80 mL), and the
combined organic layers were dried over magnesium sulfate and evaporated under reduced
pressure. The obtained crude yellow oil was redissolved in a 5 mL 50/50 mixture of ethyl
acetate/hexanes and purified by flash chromatography on silicagel (80/20 ethyl acetate/hexanes)
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
to yield the desired compound (2.03 g, 8.7 mmol, 93 %, pale yellow solid). H NMR analysis
was consistent with literature data.61
1H NMR (400 MHz, Chloroform-d)  10.05 (s, 1H), 9.11 (s, 1H), 8.68 (d, J = 4.9 Hz, 1H),
7.67 (d, J = 4.8 Hz, 1H).
13C NMR (101 MHz, Chloroform-d)  194.60, 159.60, 150.16, 140.77, 123.13, 97.23.
Melting point: 74 – 75 °C
HRMS: meas. m/z = 233.94068 (M+H) Predict. : 233.94104
(R)-Ethyl-4-amino-3-(3,4,5-trifluorophenyl)butanoate hydrochloride (4b)
In a dry, three-necked round-bottom flask were placed a magnetic stirrer, dry ethanol (50 mL,
distilled over CaH2) and (R)-4-(3,4,5-trifluorophenyl)pyrrolidin-2-one (compound 4a, 2.25 g,
10.4 mmol). The solution was cooled down and kept at 0 °C on an ice bath, and an excess of dry,
ACS Paragon Plus Environment

Page 27 of 44
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
gaseous HCl (generated by the slow addition of 100 g H2SO4 on 100 g NaCl) was bubbled
through the solution over one hour. The flask was then fitted with a condenser and the mixture
was refluxed overnight at 105 °C. The solution was again cooled down to 0 °C prior to the
dropwise addition of SOCl2 (7.5 mL, 103 mmol, 10 eq.). The reaction mixture was allowed to
stir at RT for 2 more hours. The resulting solution was filtered on a sintered glass filter and
concentrated under vacuum, until a beige solid was obtained. The title compound was further
vacuum-dried overnight and stored in the refrigerator (2.91 g, 9.8 mmol, 95%).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1H NMR (400 MHz, DMSO-d6)  7.97 (s, 2H), 7.35 (dd, J = 6.6, 9.2 Hz, 1H), 4.04 – 3.91 (m,
1H), 3.41 (tt, J = 5.5, 9.8 Hz, 1H), 3.16 – 2.96 (m, 1H), 2.93 – 2.65 (m, 1H).
13C NMR (101 MHz, DMSO-d6)  170.51, 150.14 (ddd, J = 3.8, 9.6, 246.7 Hz), 137.90 (dt, J =
15.4, 247.2 Hz), 137.29 (td, J = 4.4, 7.2 Hz), 113.35 – 112.86 (m), 60.05, 42.79, 37.18, 13.94.
ESI-MS: meas. m/z = 262. Predict. m/z = 262.11
(R)-1-((3-Iodopyridin-4-yl)methyl)-4-(3,4,5-trifluorophenyl)pyrrolidin-2-one (5)
To a dry round-bottom flask (100 mL) were added a magnetic stirrer, aldehyde 3 (2.35 g, 10.1
mmol), (R)-ethyl-4-amino-3-(3,4,5-trifluorophenyl)butanoate hydrochloride (compound 4b, 3 g,
10.1 mmol, 1 eq.) and dry ethanol (45 mL, distilled over calcium hydride). The flask was fitted
with a septum, flushed with Argon (or N2) for 5 minutes and then stirred at RT during the
dropwise addition of triethylamine (1.77 mL, 13.1 mmol, 1.3 eq), and further stirred at RT for 2
hours (dark orange solution). The reaction medium was then cooled down to 0 °C, and sodium
cyanoborohydride (764 mg, 12.1 mmol, 1.2 eq) was added to the reaction flask. The mixture was
removed from the bath and stirred for 1 hour at RT. The mixture was again cooled down to 0 °C
prior to the dropwise addition of a freshly prepared solution of sodium methoxide in ethanol. For
that purpose, a clean piece of sodium (580 mg, 25.3 mmol, 2.5 eq) was dissolved in a solution of
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 44
1
2
3
4
5
6
7
8
9
dry methanol (1.5 mL, 37 mmol) in dry ethanol (10 mL). When the addition of sodium
methoxide to the reaction flask was completed, the solution was further stirred at 0 °C for 1.5 h.
The reaction was then quenched by the addition of a saturated aqueous ammonium chloride
solution (50 mL) and vigorously stirred for 10 minutes. The resulting suspension was extracted
with ethyl acetate (3 x 80 mL). The combined organic layers were dried over MgSO4 and
concentrated under reduced pressure. The obtained crude red oil was purified by silicagel
chromatography (ethyl acetate/hexanes/triethylamine 80/20/0.5), to yield a colorless oil (2.5 g,
5.8mmol, 57%), that quickly became slightly yellow upon contact with air. For long-term
storage, the oil was kept at -25 °C [Important note: when starting from racemic compound
(R,S)-4b instead of 4b, the obtained colorless oil could be crystallized: the oil was dissolved in a
minimal amount of a 95/5 CH2Cl2/MeOH mixture, and diethylether (10 times the volume of
CH2Cl2/MeOH used) was then added to the solution. The desired (R,S)-4b then crystallised upon
ultrasound trituration or resting of the solution. The obtained off-white solid was washed with
cold ether and dried under vacuum. Melting point: 132.5 – 134 °C].
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
NMR: 1H NMR (400 MHz, Chloroform-d)  8.92 (s, 1H), 8.49 (d, J = 5.0 Hz, 1H), 7.15 (d, J =
4.9 Hz, 1H), 6.83 (dd, J = 6.2, 8.0 Hz, 2H), 4.55 (dd, J = 15.9, 71.3 Hz, 2H), 3.58 (p, J = 8.2 Hz,
1H), 3.52 (ddd, J = 6.8, 9.8, 165.8 Hz, 2H), 2.73 (ddd, J = 8.6, 17.0, 138.1 Hz, 2H).
13C NMR (101 MHz, Chloroform-d)  173.31, 157.85, 151.39 (ddd, J = 4.1, 10.0, 251.1 Hz),
149.48, 147.22, 138.20 (td, J = 15.1, 251.5 Hz), 137.99 (td, J = 4.7, 7.0 Hz), 123.69, 111.22 –
110.80 (m), 97.89, 53.76, 50.43, 37.94.
HRMS: meas. m/z = 433.00150 (M+H) Predict. : 433.00192
(R)-(4-Methoxyphenyl)(4-((2-oxo-4-(3,4,5-trifluorophenyl)pyrrolidin-1-
yl)methyl)pyridin-3-yl)iodonium trifluoromethanesulfonate (6)
ACS Paragon Plus Environment

Page 29 of 44
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
In a dry round-bottom flask (50 mL) were placed a magnetic stirrer and a solution of
compound 5 (1.1 g, 2.5 mmol) in dichloromethane (15 mL, dried over 4 Å molecular sieves).
The solution was cooled down to -20 °C on an ice/NaCl bath, and trifluoromethanesulfonic acid
(0.9 mL, 10.2 mmol, 4 eq.) was added dropwise. After 10 min of stirring, m-CPBA (purified by 3
pH-controlled extractions, followed by a recrystallization in dichloromethane and the vacuum-
drying the obtained crystals, according to a reported procedure62) (1.1 g, 6.4 mmol, 2.5 eq.) was
added and the mixture was further stirred at room temperature for 3 h, yielding a sticky, beige
solid. The stirring may be restrained by the formed solid as the reaction proceeds. It is better,
although not crucial, to release the magnetic stirrer from time to time with a glass rod. The flask
was then cooled down to 0 °C and water (91 µL, 5 mmol, 2 eq.) was added to the reaction
medium. The reaction flask was thoroughly shaken by hand, in order to efficiently quench the
excess triflic acid. The mixture was allowed to stir for 10 more minutes at 0 °C. A solution of
anisole (0.413 mL, 3.8 mmol, 1.5 eq.) in dichloromethane (5 mL) was then added dropwise to
the reaction medium and the vial was stirred at RT for 1 h, during which the beige solid turned
into a green oil that is the crude, protonated iodonium salt. The solvent was then evaporated
under reduced pressure, and diethylether (30 mL) was added to the obtained crude compound.
The green oil spontaneously crystallized into an off-white solid. The supernatant ether phase was
removed, and the solid washed with Et2O (2 x 30 mL). The obtained protonated pyridyl(4-
methoxyphenyl)iodonium salt (white solid) was dried under vacuum, and deprotonated on a
basic alumina plug according to a reported procedure42. Crystallization: the obtained solution of
pure and deprotonated iodonium was evaporated until it became a pale, yellow oil.
Dichloromethane was added while gently swirling the flask by hand (~15-20 mL). Diethylether
was then added slowly (20 mL) with a Pasteur pipette, and was stopped before any persistent
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment