Candidate PET Radioligand Development for Neurofibrillary Tangles: Two Distinct Radioligand Binding Sites Identified in Postmortem Alzheimer's Disease Brain

Article abstract of DOI:10.1021/acschemneuro.6b00051

[¹⁸F]THK-523 and [¹⁸F]807 are promising radioligands for imaging neurofibrillary tangles (NFTs) with positron emission tomography (PET) in neurodegenerative diseases, such as Alzheimer's disease (AD) and traumatic brain injury. Although [¹⁸F]THK-523 and [¹⁸F]T807 are considered high-affinity selective radioligands for NFTs, uncertainty has existed as to whether PET radioligands for imaging NFTs bind to the same molecular site because in vitro assays for ligands binding to NFTs have been lacking. We labeled THK-523 and T807 with tritium to serve as reference radioligands for in vitro binding assays with AD brain homogenates for newly synthesized ligands. With these radioligands, we identified two distinct binding sites for small molecules, one site with high affinity for THK-523 and the other with high affinity for T807. Moreover, binding assays with [³H]PIB confirmed that the two newly identified binding sites are also distinct from the thioflavin-T binding site where all current clinically useful PET radioligands for imaging β-amyloid plaque bind with high affinity. The two newly identified binding sites are considered to reside on NFTs rather than on β-amyloid plaques. Furthermore, we applied all three binding assays to a set of newly prepared compounds, based on chain modifications to THK-523. Some compounds with high affinity and selectivity for the THK-523 binding site emerged from this set, including one with amenability to labeling with fluorine-18, namely, ligand 10b.

Full text of DOI:10.1021/acschemneuro.6b00051

Subscriber access provided by - Access paid by the | UCSB Libraries
Article
Candidate PET Radioligand Development for Neurofibrillary
Tangles: Two Distinct Radioligand Binding Sites
Identified in Post Mortem Alzheimer’s Disease Brain
Lisheng Cai, Baoxi Qu, Bryan T. Hurtle, Sureshbabu Dadiboyena, Ramon Diaz-Arrastia, and Victor W Pike
ACS Chem. Neurosci., Just Accepted Manuscript • DOI: 10.1021/acschemneuro.6b00051 • Publication Date (Web): 12 May 2016
Downloaded from http://pubs.acs.org on May 14, 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.
ACS Chemical Neuroscience 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 65
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
Candidate PET Radioligand Development for Neurofibrillary Tangles:
Two Distinct Radioligand Binding Sites Identified in Post Mortem
Alzheimer’s Disease Brain
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
2
1
1
Lisheng Cai, † Baoxi Qu, † Bryan T. Hurtle, Sureshbabu Dadiboyena, Ramon Diazꢀ
2
1
Arrastia and Victor W. Pike
1
Molecular Imaging Branch, National Institute of Mental Health, National Institutes of
Health, Bethesda, MD 20892, USA.
2
Center for Neuroscience and Regenerative Medicine, Uniformed Services University of
the Health Sciences, Rockville, MD 20851, USA.
†
These 2 coꢀauthors contributed equally.
1
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 2 of 65
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
18
18
ABSTRACT: [ F]THKꢀ523 and [ F]807 are promising radioligands for imaging
neurofibrillary tangles (NFTs) with PET in neurodegenerative diseases, such as
1
8
Alzheimer’s disease (AD) and traumatic brain injury. Although [ F]THKꢀ523 and
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
18
[
F]T807 are considered highꢀaffinity selective radioligands for NFTs, uncertainty has
existed as to whether PET radioligands for imaging NFTs bind to the same molecular site
because in vitro assays for ligands binding to NFTs have been lacking. We labeled THKꢀ
5
23 and T807 with tritium to serve as reference radioligands for in vitro binding assays
with AD brain homogenates for newly synthesized ligands. With these radioligands, we
identified two distinct binding sites for small molecules, one with high affinity for THKꢀ
3
5
23 and the other with high affinity for T807. Moreover, binding assays with [ H]PIB
confirmed that the two newly identified binding sites are also distinct from the thioflavinꢀ
T binding site where all current clinically useful PET radioligands for imaging βꢀamyloid
plaque bind with high affinity. The two newly identified binding sites are considered to
reside on NFTs rather than on βꢀamyloid plaques. Furthermore, we applied all three
binding assays to a set of newly prepared compounds, based on chain modifications to
THKꢀ523. Some compounds with high affinity and selectivity for the THKꢀ523 binding
site emerged from this set, including one with amenability to labeling with fluorineꢀ18,
namely ligand 10b.
KEYWORDS: Radioligand; in vitro assay; tangle; tau; Alzheimer’s disease; PET.
2
ACS Paragon Plus Environment

Page 3 of 65
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
INTRODUCTION
Alzheimer’s disease (AD) and other tauopathies constitute an increasingly enormous
societal burden in terms of mortality, morbidity and cost, amidst a continuing absence of
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
effective treatment strategies. Brain neurofibrillary tangles (NFTs) and neuritic plaques
1
are two pathological hallmarks of AD.
NFTs are generated from abnormally
hyperphosphorylated tau protein, whereas neuritic plaques are comprised primarily of βꢀ
amyloid. Human postꢀmortem studies indicate that NFT deposition, unlike βꢀamyloid
2–7
plaque deposition, correlates well with neurodegeneration and cognitive impairment.
Furthermore, abundant NFTs are not observed in individuals who are cognitively
5
unimpaired, in contrast to βꢀamyloid plaques which are nonetheless observed in some
8
–11
nonꢀdemented people.
Moreover, tau and phosphoꢀtau (ptau181) residing in CSF are
12–15
known to be useful biomarkers for the diagnosis of AD.
NFTs are also characteristic of many other tauopathies besides AD, such as sporadic
corticobasal degeneration, progressive supranuclear palsy, Pick’s disease, frontotemporal
dementia Parkinsonism linked to chromosome 17, and Guamanian Parkinsonism
1
6
dementiaꢀamyotrophic lateral sclerosis complex.
Tau aggregates are also the
pathological hallmark of chronic traumatic encephalopathy (CTE), a neurodegenerative
1
7
disorder that occurs as a consequence of traumatic brain injury. Aggregation of tau also
arises from numerous forms of neuronal injury, including hypoxia, hyperglycemia, and
heat shock.
PET may detect and quantify a lowꢀdensity protein in living human brain when this
imaging technique is used with an effective radioligand that penetrates brain to provide a
robust and adequately high specific signal. Such radioligands are invariably labeled with
3
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 4 of 65
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
either of two cyclotronꢀproduced shortꢀlived positronꢀemitters, namely carbonꢀ11 (t1/2
=
2
0.4 min) or fluorineꢀ18 (t1/2 = 109.8 min). Over the last decade and a half, significant
progress has been made on developing radioligands for PET imaging of βꢀamyloid
plaques in diseased human brain (Chart 1). The radioligand from Eli Lilly, Amyvid
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
8
18, 19
(Florbetapir; [ F]AVꢀ45),
is the first approved by the US FDA for human use. GE’s
1
8
20
Vizamyl (Flutemetamol; [ F]3’ꢀFꢀPIB; GEꢀ067) is also now approved by the FDA.
18
18
21
18
22
NeuraCeq (Florbetaben; [ F]AVꢀ1; [ F]BAY94ꢀ9172) and [ F]AZD4694 are also
proving clinically useful. Each of these radioligands, and also the progenitor radioligand
11
23
[
C]PIB , binds with high affinity to the soꢀcalled ‘thioflavinꢀT binding site’ on βꢀ
amyloid plaques.
Chart 1. Clinically useful PET radioligands for imaging βꢀamyloid plaques.
By contrast with the PET imaging of βꢀamyloid plaques, the PET imaging of NFTs is
much less advanced, mainly because candidate PET radioligands are still limited in
4
ACS Paragon Plus Environment

Page 5 of 65
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
24, 25
efficacy.
A highꢀperforming PET radioligand for imaging brain NFTs would be
useful not only to identify areas of neuronal damage but also to monitor the effectiveness
of experimental therapies in AD, CTE and other tauopathies. Indeed, several drugs that
block phosphorylation and/or aggregation of tau are being developed for the possible
treatment of these diseases.
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
1
26
18
27
18
28
Recent publications on [ C]PBB3, [ F]T807 (AVꢀ1451) , [ F]T808 , and
18
25
[
F]THKꢀ523 (Chart 2) suggest that these radioligands have at least moderate
selectivity for binding to NFTs over βꢀamyloid plaques, primarily on the basis of
evidence gleaned from in vitro autoradiography of post mortem human brain.
11
26 18
27
18
28
[
C]PBB3, [ F]T807, and [ F]T808 have also shown moderate NFTꢀspecific
1
1
signals in PET imaging of human subjects. However, [ C]PBB3 suffers from photoꢀ
isomerization, which poses difficulties for regular practical application. Based on the
2
5
18
initial report, [ F]THKꢀ523 gives a low PET signal and may suffer some other
undesirable properties, such as low brain entry and radiodefluorination. However,
18
18
29
radioligands that are structurally related to [ F]THKꢀ523, namely [ F]THKꢀ5117 and
18
30
[ F]THKꢀ5351, (Chart 2) have recently entered clinical trials. Therefore, we also
selected THKꢀ523 as a lead compound in our search for more promising PET
radioligands for imaging NFTs.
5
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 6 of 65
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Chart 2. Candidate PET radioligands for imaging NFTs.
Multiple binding sites for PET radioligands that target βꢀamyloid plaques have been
31, 32
identified.
This suggested to us that more than one ligand binding site might also
3
3
exist in NFTs. Therefore, we produced both [ H]T807 and [ H]THKꢀ523 to serve as
reference radioligands for our in vitro assays on new candidate ligands for NFTs. As a
result, we discovered that the binding sites for these two radioligands are distinct from
each other, and that they are also distinct from the βꢀamyloid plaque binding site for
3
11
thioflavins, such as [ H/ C]PIB. In addition, a limited medicinal chemistry effort, based
on chain extension and modification of THKꢀ523, provided ligands with moderate
lipophilicity and with high affinity and selectivity for the identified THKꢀ523 binding site,
including 10b which is amenable to labeling with fluorineꢀ18.
6
ACS Paragon Plus Environment

Page 7 of 65
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
RESULTS
Chemistry. Ligand Synthesis. Specifically, we wished to explore whether the
affinity of THKꢀ523 for binding to NFTs could be enhanced by extension of its pendant
6ꢀ(2ꢀfluoroethoxy) group and by structural variation at its terminus. The use of
oxyethylene units to extend chains has been implemented previously as a promising
strategy in PET radioligand development for imaging βꢀamyloid plaques, as apparent in
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
9
21
the structures of Florbetapir and Florbetaben (Chart 1). Especially, this strategy
allows radioligand lipophilicity to be kept within a desirably moderate range and can also
be used to retain a terminal site for labeling with fluorineꢀ18. We further considered that
variation in the terminal substituent of the pendant chain, especially with a bulkier group
having hydrogen bond acceptors or donors, might enhance binding affinity with NFTs
through extra binding interactions. Therefore, we devised a general strategy to prepare
THKꢀ523 derivatives featuring chains composed of a defined number of oxyethylene
units (1–6) and with a fluoro, hydroxy, or larger substituent as a chain terminus. Our
synthetic strategy was based on commercially available 2ꢀchloroꢀ6ꢀmethoxyquinoline (1)
as starting material.
The first stage in the overall synthesis program was to replace the methoxy group in
1
with one or more oxyethylene units (Scheme 1). The phenol 2 was obtained in
moderate yield (47%) by demethylation of 1 with boron tribromide. Various
poly(oxyethylene)ꢀbisꢀalcohols were monoꢀtosylated to give 3b–f in low to moderate
yields (29–53%). Treatment of 2 with the bromo compound 3a or with one of the
tosylates gave the hydroxyl terminated compounds with the desired number of
oxyethylene units (4a–f) in good to excellent yields, except for 4a (27%). This approach
7
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 8 of 65
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
was also explored for preparing analogous compounds with a fluoro terminus, but
succeeded only for compound 5 having a pendant 6ꢀ(2ꢀfluoroethoxy) chain. Nonetheless,
5
was obtained in virtually quantitative yield (Scheme 1).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Scheme 1. Synthesis of Hydroxy Compounds 4a–f and Fluoro Compound 5
Br
HO
3a
or
TsCl, Py
for n = 2 4
O
O
HO
HO
n Ts
H
n
TsCl, KI, Ag O
2
3b, n = 2
c, n = 3
d, n = 4
for n = 5 and 6
3
3
3e, n = 5
3f, n = 6
O
HO
n
Cs CO
2
3
N
Cl
MeO
HO
4a, n =
BBr3
4b, n = 2
4c, n = 3
4d, n = 4
N
Cl
N
Cl
1
2
4e, n = 5
4f, n = 6
OTs
Cs CO
F
O
F
2
3
N
Cl
5
The next stage aimed to cross couple the prepared chloroquinolines (4a–f and 5)
with 4ꢀ(Bocꢀamino)phenylboronic acid through Suzuki reactions. The preparation of the
methoxy compound 6 from 1 (Scheme 2) was used as a model for establishing effective
coupling conditions. We tried various combinations of phosphine ligand, palladium
compound, and solvent, including 2'ꢀphenylꢀphenyl diꢀtꢀbutyl phosphine plus Pd(OAc)2
in THF or DMF, 1,1'ꢀbis(diisopropylphosphino)ferrocene plus Pd (dba) in toluene, and
2
3
Pd(PPh ) plus K CO in DMEꢀH O. The latter combination was most effective, giving
3
4
2
3
2
virtually quantitative yield of 6. K CO was necessary for the reaction to proceed, and
2
3
the proportion of water had to be kept below 50%. These coupling conditions delivered
the Nꢀprotected hydroxyꢀterminated compounds (7a–f) plus the Nꢀprotected fluoro
8
ACS Paragon Plus Environment

Page 9 of 65
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
compound 8 in good to excellent yields (60–99%). Compounds 7a–f were readily
converted into their tosylates (9a–f) in excellent yields (83–90%) by treatment with tosyl
chloride. Our prior attempts to insert the tosylate group into these compounds before
Suzuki coupling did not succeed. Treatment of 8 with TFA gave THKꢀ523 (10a) in
quantitative yield (Scheme 2).
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 2. Syntheses of Tosylates 9a–f, and THKꢀ523 (10a)
Treatment of the Nꢀprotected tosylates 9b–f with TBAF in THF gave the target
fluoro ligands 10b–f in moderate to high yield (40–92%), without the need for a separate
9
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 10 of 65
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
deprotection reaction (Scheme 3). The small amount of water in the commercial reagent
was found to be critical for the success of these syntheses. If water was rigorously
excluded from the reactions, products were obtained with a lower number of oxyethylene
units.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Scheme 3. Syntheses of THKꢀ523 Analogs with Poly(oxyethylene) Chains and
Various Terminal Groups
10
ACS Paragon Plus Environment

Page 11 of 65
ACS Chemical Neuroscience
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
Treatment of the Bocꢀprotected tosylate 9b with cyclohexaneꢀ1,3ꢀdione,
cyclopentaneꢀ1,3ꢀdione or 2ꢀaminoethanol in acetonitrile with potassium carbonate as
base gave the variously chainꢀterminated compounds 11a (79%), 11b (59%) and 11f
(21%), respectively (Scheme 3). Treatment of 9b with base plus one of the ketones,
oxazolidinꢀ2ꢀone, imidazolidinꢀ2ꢀon or thiazolineꢀ2ꢀone gave the compounds 11c–e with
11
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 12 of 65
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
alternative bulky terminal groups in a similar range of yields (21–59%). The Boc group
in each of the compounds 11a–f was removed readily with BF .OEt to give the target
3
2
ligands 12a–f in 35–72% yields (Scheme 3).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Synthesis of Halo Precursors for Tritiation. Treatment of THKꢀ523 (10a) with
NIS in chloroformꢀmethanol at room temperature gave compound 13 with an iodo group
at each position ortho to the free amino group (Scheme 4). Any protective group at the
amino group, such as an acetyl or formyl group, prevented iodination. We found that the
two iodo substituents in 13 could be removed with hydrogen in the presence of Pd/C as
3
catalyst to regenerate THKꢀ523. Thus, [ H]THKꢀ523 was obtained by treating 13 with
tritium gas as a custom synthesis (Moravek Biochemicals & Radiochemicals, Inc.)
(Scheme 4).
3
3
Scheme 4. Synthesis of [ H]THKꢀ523 ([ H]10a)
Attempt at an analogous iodination of T807 did not generate any tractable
compound, and no further effort was made to synthesize an iodinated derivative of T807.
Bromination of T807 with NBS generated the triꢀbromo substituted derivative 14
(Scheme 5). We found that the three bromo substituents in 14 could be removed with
3
hydrogen in the presence of Pd/C as catalyst to regenerate T807. Thus, [ H]T807 was
12
ACS Paragon Plus Environment

Page 13 of 65
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
obtained by treating 14 with tritium gas as a custom synthesis (Moravek Biochemicals &
Radiochemicals, Inc.) (Scheme 5).
3
Scheme 5. Synthesis of [ H]T807
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
18
FꢀLabeling of THKꢀ523 and Derivatives. Because [ F]10b–f might be
considered candidate PET radioligands, we were interested to test the feasibility of their
1
8
labeling with noꢀcarrierꢀadded (NCA) cyclotronꢀproduced [ F]fluoride ion. The NꢀBoc ꢀ
18
protected tosylates, 9b–f, were used as precursors for trial Fꢀlabeling (Scheme 6). After
1
8
successful substitution of the tosylate group with [ F]fluoride ion, the Boc group was
18
removed under acidic conditions. The labeled THKꢀ523 derivatives [ F]10b–f were
each neutralized and isolated with HPLC in 20–30% decayꢀcorrected radiochemical
yields.
13
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 14 of 65
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
18
Scheme 6. Radiosyntheses of [ F]10b–f
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Binding Assays for Human AD βꢀAmyloid Plaques and NFTs. We used three
tritiumꢀlabeled compounds as reference radioligands (Chart 3) to measure the in vitro
inhibition constants of new compounds with human AD brain homogenate as protein
3
source. Protocols similar to those that have been used previously for assays of [ H]PIB
binding to AD brain tissue did not generate any discernible inhibition curves for
3
3
[ H]THK523 or [ H]T807 binding. We tried a variety of ways to improve the binding
assays, including use of a variety of assay mediums, such as PBS, aq. DMSO (20%), aq.
DMSO (40%), neat DMSO, aq. ethanol (20%), aq. ethanol (40%), neat ethanol, aq. BSA
(
0.1%), or Tweenꢀ20 (0.1%) in PBS, and a variety of washing solutions, such as PBS,
Tris, aq. ethanol (20%), aq. ethanol (40%), or neat ethanol. The combination of
conditions that gave reproducible results was as follows: (a) use of AD brain tissue
homogenate that was reꢀhomogenized after thawing the stock homogenate; (b) treatment
of the reꢀhomogenized suspension with a highꢀpower ultrasonic probe; and (c) use of
Tweenꢀ20 (0.1%) in PBS instead of PBS as the assay medium. After the best conditions
were implemented in, the washing solution was found to have little effect on measured Ki
values (Figure 1).
14
ACS Paragon Plus Environment

Page 15 of 65
ACS Chemical Neuroscience
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
Chart 3. Three reference radioligands used for in vitro binding assays on AD brain
tissue homogenates.
8
000
7000
6
5
000
000
4000
3
2
000
000
neat ethanol
0% ethanol
Tris buffer
2
1000
0
-11
-10
-9
-8
-7
-6
Log(THK-523 conc; M)
Figure 1. Lack of effect of different washing agents on the K measured for THKꢀ523 in
i
3
the in vitro binding assay with [ H]THKꢀ523 as reference radioligand. The data points
are means ± SD of 3 measurements (error bars not shown are within symbol size). The
determined K values are 45 ± 12 nM with neat ethanol wash; 28 ± 6 nM for 20% ethanol
i
wash; and 52 ± 14 nM for Trisꢀbuffer (pH 7.2) wash.
15
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 16 of 65
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
3
Titration of AD Brain Homogenate with [ H]THKꢀ523. The binding of the AD
3
brain homogenate with [ H]THKꢀ523 at different concentrations of THKꢀ523 followed a
typical saturation curve for a radioligand with a large proportion of nonꢀspecific binding
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(
Figure 2). The curve was fitted to the equation [BF] = Bmax[F]/(K + [F]) + (NS×[F]) + c,
D
where [BF] is the concentration of bound radioligand, Bmax the concentration of maximal
binding sites, [F] the concentration of free radioligand plus THKꢀ523, K the equilibrium
D
dissociation constant, NS the coefficient of nonꢀspecific binding, and c the concentration
of bound radioligand in background. The best fit was obtained with K = 3.5 ± 3.4 nM,
D
Bmax = 1576 ± 499 dpm, NS = 321 ± 7, and c = 17.38 ± 173.
16
ACS Paragon Plus Environment

Page 17 of 65
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
35000
30000
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
2
2
1
1
5000
0000
5000
0000
5
000
0
0
10
20
30
40
50
60
70
80
90
100
[3
H]THK-523 plus THK-523 (nM)
3
Figure 2. Binding of [ H]THKꢀ523 to AD brain homogenate versus concentration of
3
[ H]THKꢀ523 plus THKꢀ523. The data points are means of 4 measurements and SDs are
within symbol size. The curve was fitted to the equation [BF] = Bmax[F]/(K + [F]) + (NS
D
×
[F]) + c, as described in the Experimental, giving K = 3.5 ± 3.4 nM.
D
1
8
Titration of AD Brain Homogenate with [ F]THKꢀ523. The binding of the AD
1
8
brain homogenate with [ F]THKꢀ523 at different concentrations of THKꢀ523 also
followed a typical saturation curve for a radioligand with a large proportion of nonꢀ
specific binding (Figure 3). The curve was fitted as in the analogous experiment with
3
[
H]THKꢀ523. The best fit was obtained with K = 4.6 ± 1.8 nM, B = 1639 ± 235 dpm,
D max
NS = 169 ± 3, and c = 57.7 ± 65.2.
17
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 18 of 65
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
2
1
0000
5000
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
10000
5000
0
0
10
20
30
40
50
60
70
80
90
100
1
8
[ F]THK-523 plus THK-523 (nM)
1
8
Figure 3. Binding of [ F]THKꢀ523 to AD brain tissue homogenate versus concentration of
1
8
THKꢀ523 plus [ F]THKꢀ523. The data are means from 4 measurements and SDs are within
symbol size. The curve was fitted to the equation [BF] = Bmax[F]/(K + [F]) + (NS × [F]) + c,
D
as described in the Experimental, giving K = 4.6 ± 1.8 nM.
D
Inhibition of Binding to Human AD βꢀAmyloid Plaques and NFTs. When we used
the three binding assays with AD brain homogenates to screen compounds for inhibition
of radioligand binding at 1 µmol concentration, we found that the reference radioligands
3
occupied distinctly different binding sites. Thus, T807 failed to inhibit [ H]THKꢀ523
3
binding (Figure 4), and THKꢀ523 failed to inhibit [ H]T807 binding (Figure 5). Similar
3
mutually exclusive inhibition behavior was also observed for [ H]PIB (data not shown).
These graphical demonstrations of multiple binding sites were also confirmed by the Ki
values measured for THKꢀ523, T807 and the newly synthesized compounds at the three
1
8
ACS Paragon Plus Environment

Page 19 of 65
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
different binding sites (Table 1). For the new fluoro compounds 10b–f, only the
compound with two oxyethylene units (10b) showed high affinity (K = 28 nM) and
i
selectivity for binding to the ‘THK binding site’. None of these compounds showed high
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
affinity for the ‘T807 binding site’, whereas compound 12e, which has five oxyethylene
units, showed high affinity (K = 43 nM) and selectivity for binding to the ‘PIB binding
i
site’. Some variations in the terminal group were quite well tolerated among other
compounds having two oxyethylene units. Thus, whereas compound 12a and 12b
showed greatly decreased K values, compounds 12c–f retained both high affinity and
i
selectivity for the THK binding site. Notably, 12e and 12f showed higher affinities
(
lower K values) than THKꢀ523 itself in the same assay.
i
19
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 20 of 65
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
1
1
20
00
80
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
6
4
2
0
0
0
0
3
Figure 4. Percentages of specific binding of [ H]THKꢀ523 to AD brain tissue homogenate
inhibited by each of 13 compounds (10a–f, 12a–f and T807) at 1 µM concentration. THKꢀ
523 (10a) gave full inhibition, whereas T807 gave very low inhibition. The data are mean ±
SD from 3 measurements.
20
ACS Paragon Plus Environment

Page 21 of 65
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
1
1
20
00
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
8
6
4
0
0
0
20
0
-20
3
Figure 5. Percentages of specific binding of [ H]T807 to AD brain tissue homogenate
inhibited by each of 13 compounds (10a–f, 12a–f and T807) at 1 µM concentration.
T807 gave full inhibition, whereas THKꢀ523 (10a) gave very low inhibition. The data
are mean ± SD from 3 measurements
21
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 22 of 65
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Table 1. cLogD and K Values for Ligands at Three Binding Sites in AD Brain
Tissue Homogenates
i
K (nM) at
i
Compound Structure
cLogD
THKꢀ523
site
T807 site
PIB site
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
T807
2.91
2.30
> 1000
2.6 ± 0.7
> 1000
> 1000
34 ± 3
PIB
> 1000
51 ± 22
THKꢀ523
3.30
425 ± 383 > 1000
(10a)
1
1
0b
0c
3.74
3.33
2.62
2.70
2.29
28 ±8
> 1000
> 1000
> 1000
N.d.
> 1000
> 1000
> 1000
43 ±39
> 500
> 1000
> 1000
> 1000
> 1000
10d
1
1
0e
0f
N.d.
1
2a
–0.38
1.79
2.87
2.16
358 ± 381
> 1000
> 1000
> 1000
> 1000
> 1000
502 ± 143
> 1000
12b
1
1
2c
35 ± 19
78 ± 7
> 1000
2d
> 1000
22
ACS Paragon Plus Environment

Page 23 of 65
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
12e
12f
3.30
0.38
21 ± 6
> 1000
> 1000
> 1000
> 1000
22 ± 3
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
Note: K values are average of 5 measurements for THKꢀ523, T807, and PIB. K values
i
i
for other compounds are mean ± SD from 3 measurements. N.d. = K , not determined;
i
3
see Figure 5 for percentage inhibition of [ H]T807 binding at 1 µmol concentration.
Computed lipophilicity. Computation indicated that clogD for all the prepared
target ligands was held within a moderate range (1.3–3.74), except for compounds 12a
and 12f.
DISCUSSION
We overcame many obstacles to establish in vitro assays for measuring the relative
binding affinities of small molecules for NFTs. The challenges that we met came from
three main issues: 1) ligand solubility is low in PBS; for example,THKꢀ523 is insoluble
in water or PBS; even its high fluorescence is not observed in water, reflecting its limited
solubility in common media used for binding assays; 2) NFTs from AD brain
homogenate are poorly accessible to small molecules; they exist as aggregates within
neurons, unlike βꢀamyloid plaques, which exist outside neurons; 3) NFTs are highly
heterogeneous, and on standing they precipitate within 5–10 min.
Effective conditions for binding assays on human AD brain tissue were achieved by
taking three measures: 1) substituting PBS with 0.1% Tweenꢀ20 in PBS as the assay
medium; 2) reꢀhomogenizing the brain tissue homogenate before each experiment; and 3)
ultrasound treatment of the tissue after the reꢀhomogenization. With regard to the ligand
23
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 24 of 65
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
solubility issue, we tried adding organic solvent, such as ethanol or DMSO, at various
concentrations, to the binding assay medium. We also tried adding BSA. However, we
did not obtain an appreciable signal for specific binding of radioligand, and the small
dubious signal was also irreproducible. Although a low concentration of Tweenꢀ20
(0.001%) in PBS was ineffective, it was found that a much higher concentration 0.1%
promoted complete dissolution of the ligands in aqueous media. The other two measures
were designed to address the issue of poor NFT accessibility to small molecules. Reꢀ
homogenization before an assay experiment helped to break down the chunk of brain
tissue into smaller pieces, thereby exposing individual neurons to the highest degree.
Subsequent ultrasound treatment then helped to break the neuronal cell membranes,
exposing the NFTs to small molecules.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
During our initial efforts to workꢀup assay conditions, we hypothesized that the
specific radioligand binding signal was obscured by high nonꢀspecific binding to the
brain tissue homogenate. Therefore, various washing agents were tested to reduce
radioligand nonꢀspecific binding, such as PBS or Trisꢀbuffer, or various concentrations of
ethanol, DMSO, acetic acid or formic acid in water. However, it was found that after
deploying the revised combination of conditions for the binding assay, the nature of the
washing agent no longer affected the K measurement (Figure 2).
i
The K value for THKꢀ523 for binding to AD brain homogenate was also measured
D
3
by titration. K was found to be 3.5 ± 3.4 nM and 4.6 ± 1.8 nM with [ H]THKꢀ523
D
1
8
(
Figure 3) and [ F]THKꢀ523 (Figure 4) as radioligands, respectively.
These
measurements are clearly in good agreement. However, the K value measured from the
i
competitive inhibition experiment was much lower at 51 ± 22 nM. This discrepancy may
24
ACS Paragon Plus Environment

Page 25 of 65
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
derive from error that is intrinsic to the titration experiments, where the specific binding
signals are only a small fraction of the total binding.
The principal discovery of this study is the plurality of distinct binding sites on
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
18
NFTs. Both [ F]THKꢀ523 and [ F]T807, and their close analogs, are considered to be
radioligands that bind selectively to NFTs, based on ample evidence from
autoradiography, immunohistochemistry, and the use of brain tissue from different
2
5, 28, 33, 34
disease origins.
Therefore, in this study, we assumed that the binding of THKꢀ
523 analogs to AD tissue homogenate was entirely to binding sites on NFTs, rather than
to other binding sites on βꢀamyloid plaques or to other proteins in AD brain homogenate.
When we analyzed the binding data, we assumed a singleꢀsite model, to which the data fit
quite well. No compelling reason to fit the data to two or more binding sites was found.
3
3
The mutually exclusive binding behaviors of [ H]THKꢀ523, [ H]T807 and of some of the
prepared ligands (e.g., 10b, 12c, 12e; Table 1), showed that they occupy different binding
sites within NFTs. Although multiple binding sites have been observed on βꢀamyloid
3
1, 32, 35
plaques,
only radioligands that bind to the thioflavin T site (Chart 1) have so far
become useful for clinical PET imaging. Nonetheless, for biomedical research on the
various NFTs in tauopathies, radioligands that bind to different sites on NFTs may well
find distinct applications.
From only a small set of synthesized THKꢀ523 derivatives, three compounds (10b,
1
2e, and 12f) emerged with high affinity (< 50 nM) and selectivity for the THKꢀ523
binding site (Table 1). These contained two oxyethylene units. Some fiveꢀmembered
rings were well tolerated as chain terminal groups, as in 12c and 12e. Compounds with
more than two oxyethylene units showed no appreciable binding at the THKꢀ523 site (Ki
25
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 26 of 65
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
>
1000 nM), although 10e with five oxyethylene units showed high affinity and
selectivity for binding to the PIB site (K > 4.3 nM). Compound 10b appears of special
i
interest because of i) its demonstrated amenability to labeling with fluorineꢀ18, ii) high
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
affinity (K = 28 nM) on par with that of THKꢀ523 itself in the same assay (K = 51 nM),
i
i
and iii) a moderate lipophilicity (logD, 3.74) that computes to be only 0.44 clogD units
greater than that of THKꢀ523 (Table 1).
CONCLUSIONS
We developed reliable in vitro binding assays to measure the binding affinities of small
molecules proposed as possible radioligands for PET imaging of NFTs. Through tritiumꢀ
labeling of THKꢀ523 and T807, we identified two different kinds of binding sites on
NFTs. These two binding sites are also different from that of PIB. Use of these three
assays expands the scope for developing new PET radioligands for NFTs by bringing a
variety of binding sites into consideration. Moreover, chain modifications in THKꢀ523
were successful in producing compounds with high affinity and selectivity for the THK
binding site, accompanied by moderate lipophilicity compatible with further development
as PET radioligands.
EXPERIMENTAL SECTION
Materials. Water was purified with a system comprising two filters, one Rio™,
one reservoir, and one MilliꢀQ® synthesis system (Millipore; Bedford, MA). Common
solvents were obtained from Fisher Scientific (Pittsburgh, PA). Literature methods were
2
3
36
37
used to prepare PIB, T807, 2ꢀfluoroethyl 4ꢀmethylbenzenesulfonate, and 2ꢀ
3
8
3
hydroxyethyl 4ꢀmethylbenzenesulfonate,
.
[ H]PIB was purchased from Moravek
Biochemicals Inc. (Brea, CA) as a custom radiosynthesis. 2ꢀChloroꢀ6ꢀmethoxyquinoline
26
ACS Paragon Plus Environment

Page 27 of 65
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
(
1) was obtained from Ark Pharma Inc. (Libertyville, IL). Other chemicals, including 2ꢀ
bromoethanꢀ1ꢀol (3a) were purchased from Aldrich Chemical Co. (Milwaukee, WI),
Fluka Chemical Co. (Milwaukee, WI), Acros (Hampton, NH) or Strem Chemicals
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
(
Newburyport, MA), and were used without further purification.
Instruments and General Methods. Analytical TLC was performed on preꢀ
coated aluminum plates (silica gel 60, F254) and compounds were visualized under UV
light at 254 or 365 nm. Reaction products were purified with flash column
chromatography on silica gel 60 (230–400 mesh), and EtOAcꢀhexaneꢀMeOH was used as
mobile phase. Analytical HPLC was performed with a reversed phase column (Xꢀbridge
C18; 5 ꢀm; 10.0 × 250 mm; Waters Corp.; Milford, MA) eluted with c. NH (0.025%) in
3
MeCNꢀH O at 6.2 mL/min. The chromatography system was fitted with a continuous
2
wavelength UVꢀvis detector (System Gold 168 Detector; Beckman; Fullerton, CA) and
an autosampler (System Gold 508; Beckman). For semiꢀpreparative HPLC, a reversed
phase column (Atlantis C18; 5 ꢀm; 30 × 150 mm; Waters Corp.) or silica gel column (10
ꢀm; 30 × 250 mm; Phenomenex; Torrance, CA) was eluted at 30 mL/min. The HPLC
system was fitted with a manual injector (5ꢀmL loop) and a third delivery pump using
MeCN or EtOAc as mobile phase at 1 mL/min. The purities of compounds were
determined with HPLC monitored for absorbance at 254 nm and expressed as area
percentage of all peaks. All compound purities exceeded 95%.
1
13
The H and C NMR spectra of all compounds were acquired on a DRX 400
1
13
instrument (400 MHz H and 100 MHz C; Bruker; Billerica, MA), using the chemical
shifts of residual deuterated solvent as the internal standard; chemical shift ( ) data for
the proton and carbon resonances are reported in parts per million (ppm) downfield
δ
27
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 28 of 65
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
relative to the signal for TMS (
MS instrument (Thermo Fisher Scientific; Waltham, MA) fitted with a Luna C18 column
5 ꢀm; 2.0 × 150 mm; Phenomenex) eluted at 150 ꢀL/min with a MeOHꢀH O mixture.
δ = 0). Mass spectra were acquired using a Velos Pro LCꢀ
(
2
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Highꢀresolution mass spectra (HRMS) were acquired from the Mass Spectrometry
Laboratory, University of Illinois at UrbanaꢀChampaign (Urbana, IL). Either electron
spray ionization (ESI) or electron ionization (EI) was used. Melting points were
measured with a MelꢀTemp (Electrothermal; Thermo Fisher Scientific) or SMPꢀ30
(Stuart; Staffordshire, UK) manual melting point apparatus and were uncorrected.
cLogD was computed with Pallas for Windows software version 3.8 in default option
(CompuDrug; Bal Harbor, FL). A Discover apparatus (CEM; Matthews, NC) was used
for microwave synthesis.
Brain Tissue. Post mortem brain tissues from an autopsyꢀconfirmed AD case
and a healthy control were obtained from the Human Brain Collection Core (HBCC),
NIH.
Ligand Syntheses. 2ꢀChloroquinolinꢀ6ꢀol (2). 2ꢀChloroꢀ6ꢀmethoxyquinoline (1;
1
0 g; 52 mmol) was dissolved in DCM (200 mL) and BBr (12 mL; 126 mmol) was
3
added through a large needle syringe. Some offꢀwhite precipitate formed quickly and
accumulated during the reaction. The mixture was stirred at RT overnight, and then
quenched slowly by adding MeOH until heat and gas generation stopped. The solution
was basified by adding NaHCO until gas generation stopped. The mixture was then
3
extracted with DCM (200 mL × 3). The collected organic phase was dried (MgSO ), and
4
the solvent was removed to give 2 as a brown solid (4.8 g; 47%) that was pure enough for
1
3
subsequent reactions: H NMR (CD OD) δ 8.11 (d, 1H, J = 8.6 Hz, ArH), 7.78 (d, 1H,
3
HH
28
ACS Paragon Plus Environment

Page 29 of 65
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
3
4
JHH = 9.2 Hz, ArH), 7.38 (m, 2H, ArH), 7.15 (d, J = 2.6 Hz, 1H, ArH); m/z (ESIꢀ
HH
+
+
+
MS): 180.3 (100%, [M+H] ), 182.2 (30%, [M+H] ), 190.1 (22%); HRMS (TOF ): calcd
+
for C H ClNO [M+H] , 180.0216; found, 180.0215. Error (ppm): – 2.5.
9
7
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
2ꢀ(2ꢀHydroxyethoxy)ethyl 4ꢀmethylbenzenesulfonate (3b). 2,2'ꢀOxybis(ethanꢀ1ꢀ
ol) (1.7 g, 16 mmol) was dissolved in DCM (10 mL). pꢀTsCl (1.0 g, 5.3 mmol) and
pyridine (1.2 mL, 16 mmol) were added sequentially. The reaction mixture was stirred at
RT for 2 d, and then partitioned between DCM (100 mL) and brine (100 mL). The aq.
phase was extracted twice with DCM (100 mL). The combined organic phase was dried
(MgSO ), and the solvent was removed to leave a colorless oil, which was dissolved in
4
DCM. Silica gel (4 mL) was added, and the mixture was dried by rotary evaporation.
The dry powder was loaded onto a silica gel column with silica gelꢀhexane (100 mL).
Oxybis(ethaneꢀ2,1ꢀdiyl) bis(4ꢀmethylbenzenesulfonate) eluted in EtOAcꢀhexane (1: 3).
Elution with EtOAcꢀhexane (1: 1) followed by removal of solvent gave 3b as a colorless
1
3
3
oil (0.57 g; 42%): H NMR (CDCl ) δ 7.77 (dm, J = 8.3 Hz, 2H, ArH), 7.32 (dm, J
HH
3
HH
= 8.6 Hz, 2H, ArH), 4.17–4.15 (m, 2H, OCH ), 3.67–3.63 (m, 4H, OCH ), 2.42 (s, 3H,
2
2
13
CH ); C NMR (CDCl ) δ 145.1, 133.0, 130.0, 128.0, 72.7, 69.4, 68.8, 61.8, 21.8 (s, 1C,
3
3
+
CH3); m/z (ESIꢀMS): 371.1 (30%, [TsO(C H O) H+Na] ), 349.1 (30%,
2
4
4
+
+
+
TsO(C H O) H+H] ), 283.1 (100%, [M+Na] ), 278.1 (34%, [M+H O] ), 261.1 (100%,
2
4
4
2
+
+
+
[
M+H] ), 235.1 (30%), 89.1 (40%, [CH CH OCH CH OH] ); HRMS (TOF ) calcd for
2 2 2 2
+
C H O S, 261.0797 [M+H] ; found, 261.0792. Error (ppm): − 1.9.
11
16
5
2
ꢀ(2ꢀ(2ꢀHydroxyethoxy)ethoxy)ethyl 4ꢀmethylbenzenesulfonate (3c). 2,2'ꢀ(Ethaneꢀ
,2ꢀdiylbis(oxy))bis(ethanꢀ1ꢀol) (15.5 g, 103 mmol) was dissolved in DCM (50 mL). pꢀ
TsCl (6.2 g, 32.5 mmol) and pyridine (8.0 mL, 99 mmol) were added sequentially. The
1
29
ACS Paragon Plus Environment