Inhibition of 14-3-3/Tau by hybrid small-molecule-peptides operating via two different binding modes

Article abstract of DOI:10.1021/acschemneuro.8b00118

Current molecular hypotheses have yet to deliver marketable treatments for Alzheimer's disease (AD), arguably due to a lack of basic knowledge of AD biology, and an overreliance on conventional drug modalities. Protein-protein interactions (PPIs) are emerging drug targets, which show promise for the treatment of e.g. cancer, but are still underexploited for treating neurodegenerative diseases. 14-3-3 binding to phosphorylated Tau is a promising PPI drug target based on its reported destabilizing effect on microtubules, leading to enhanced neurofibrillary tangle formation as a potential cause of AD-related neurodegeneration. Inhibition of 14-3-3/Tau may therefore be neuroprotective. Previously, we reported the development of modified peptide inhibitors of 14-3-3/Tau using a novel structure-guided approach. Here, we report further efforts to optimize the binding mode and activity of our modified Tau peptides through a combination of chemical synthesis, biochemical assays, X-ray crystallography and NMR spectroscopy studies. Most notably, we were able to characterize two different high-affinity binding modes, both of which inhibited 14-3-3-binding to full-length PKA-phosphorylated Tau protein in vitro. Our findings, besides producing useful tool inhibitor compounds for studying 14-3-3/Tau, have enhanced our understanding of the molecular parameters for inhibiting 14-3-3/Tau, which are important milestones toward the establishment of our 14-3-3 PPI hypothesis.

Full text of DOI:10.1021/acschemneuro.8b00118

Subscriber access provided by Kaohsiung Medical University
Article
Inhibition of 14-3-3/Tau by hybrid small-molecule-
peptides operating via two different binding modes
Sebastian Alexandru Andrei, Femke Meijer, Joao Neves, Luc Brunsveld,
Isabelle Landrieu, Christian Ottmann, and Lech-Gustav Milroy
ACS Chem. Neurosci., Just Accepted Manuscript • DOI: 10.1021/acschemneuro.8b00118 • Publication Date (Web): 03 May 2018
Downloaded from http://pubs.acs.org on May 4, 2018
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 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 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.
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 57
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
Inhibition of 14ꢀ3ꢀ3/Tau by hybrid smallꢀmoleculeꢀ
peptides operating via two different binding modes.
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
Sebastian A. Andrei,†ξ Femke A. Meijer,† ξ João Filipe Neves,§ Luc Brunsveld,† Isabelle
Landrieu,§* Christian Ottmann,†ζ* and LechꢀGustav Milroy†*
† Laboratory of Chemical Biology, Department of Biomedical Engineering and Institute for
Complex Molecular Systems, Technische Universiteit Eindhoven, Den Dolech 2, 5612 AZ
Eindhoven, The Netherlands
§ UMR 8576 CNRSꢀLille University, 59000 Villeneuve d’Ascq, France
ζ Department of Chemistry, University of DuisburgꢀEssen, Universitätsstrasse 7, 45117 Essen,
Germany
ξ These authors contributed equally to the work.
1
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 2 of 57
1
2
3
4
5
6
7
8
9
Abstract Current molecular hypotheses have yet not delivered marketable treatments for
Alzheimer’s disease (AD), arguably due to a lack of understanding of AD biology, and an
overreliance on conventional drug modalities. Proteinꢀprotein interactions (PPIs) are emerging
drug targets, which show promise for the treatment of e.g. cancer, but are still underexploited for
treating neurodegenerative diseases. 14ꢀ3ꢀ3 binding to phosphorylated Tau is a promising PPI
drug target based on its reported destabilizing effect on microtubules, leading to enhanced
neurofibrillary tangle formation as a potential cause of ADꢀrelated neurodegeneration. Inhibition
of 14ꢀ3ꢀ3/Tau may therefore be neuroprotective. Previously, we reported the structureꢀguided
development of modified peptide inhibitors of 14ꢀ3ꢀ3/Tau. Here, we report further efforts to
optimize the binding mode and activity of our modified Tau peptides through a combination of
chemical synthesis, biochemical assays, Xꢀray crystallography. Most notably, we were able to
characterize two different highꢀaffinity binding modes, both of which inhibited 14ꢀ3ꢀ3ꢀbinding to
fullꢀlength PKAꢀphosphorylated Tau protein in vitro as measured by NMR spectroscopy. Our
findings, besides producing useful tool inhibitor compounds for studying 14ꢀ3ꢀ3/Tau, have
enhanced our understanding of the molecular parameters for inhibiting 14ꢀ3ꢀ3/Tau, which are
important milestones toward the establishment of our 14ꢀ3ꢀ3 PPI hypothesis.
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
Keywords. 14ꢀ3ꢀ3 • Tau • ProteinꢀProtein Interactions • Inhibitors •Drug Discovery • Peptide
Chemistry
2
ACS Paragon Plus Environment

Page 3 of 57
ACS Chemical Neuroscience
1
2
3
4
Introduction.
5
6
7
8
9
Despite the best efforts of academia and pharma, there are currently no marketed drugs for AD,
while the drug development pipeline of agents to treat the underlying pathology of AD – soꢀ
called disease modify therapies (DMTs) – is very active, comprising 70% of the 105 compounds
in clinical trial.1 While the drug mechanisms of current agents in Phases IꢀIII are heterogeneous,
they can nonetheless be broadly classified as targeting either the amyloid cascade or the
downstream pathophysiology, including Tau pathway. Noticeably, more than half of agents in
Phase III are antiꢀamyloids compared to only 4% directly targeting Tau.1 The vast preponderance
of all agents in the pipeline is either monoclonal antibodies or βꢀsite amyloid precursor protein
cleaving enzyme (BACE) inhibitors. Given that no treatment has yet been found, there is
considerable incentive to address the relatively unexplored Tau pathway. The major one being
that, in AD, the severity of cognitive decline is more correlated with the evolution of Tau
neurofibrillary tangles (NFTs) than with that of amyloid deposits.2 However, the effective
targeting of NFTs might require a shift to new molecular modalities,3 the solution to which could
be found in previously intractable molecular targets. Proteinꢀprotein interactions (PPIs) are, by
conventional standards, emerging drug targets,4–8 which show promise for the treatment of other
disease types, such as cancer,9 but are as yet underexploited for treating neurodegenerative
diseases, including AD, despite the clear potential.10 We therefore bring forward the PPI between
14ꢀ3ꢀ3 proteins and Tau, as a potential therapeutic target to treat AD.11–13
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
14ꢀ3ꢀ3 proteins are adapter proteins, existing in seven isoforms – β, ε, γ, η, σ, τ and ζ –14–16
which bind to and modulate the function and folding of phosphorylated proteins.17 While
abundantly expressed in the body, 14ꢀ3ꢀ3 proteins are most abundant in CNS compartments.18
3
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 4 of 57
1
2
3
4
5
6
7
8
9
14ꢀ3ꢀ3 proteins make important PPIs in diverse pathophysiological settings such as cancer,19
metabolic diseases,20 and in neurodegeneration. In the latter case, 14ꢀ3ꢀ3 proteins bind to a
number of client proteins implicated in CNS diseases, among them Tau, αꢀsynuclein, parkin and
LRRK2.21 Therefore, 14ꢀ3ꢀ3 proteins are fundamentally interesting targets for neurodegenerative
drug therapy,13 in which either inhibition or stabilization of 14ꢀ3ꢀ3 PPIs may prove to be viable
therapeutic strategies.22–25
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
Tau protein similarly exerts its multiple neuronal functions by binding a range of partners, the
most wellꢀdocumented being the binding and stabilizing of microtubules.26,27 This interaction, as
is the case for many Tau interactions, is physiologically regulated by phosphorylation.28
However, hyperphosphorylation of Tau, associated with its aggregation inside neurons as PHF,
are wellꢀknown hallmarks of AD.29 The Tau/14ꢀ3ꢀ3 association has been found to impact several
aspects of the Tau pathway in neurodegeneration. First, 14ꢀ3ꢀ3ζ has been described to stimulate
Tau phosphorylation by GSK3β kinase in cell model and in brain,30–32 and cAMPꢀdependent
protein kinase in vitro.30 Additionally, 14ꢀ3ꢀ3ζ is reported to be associated with the
neurofibrillary tangles composed of Tau PHF in AD brain extracts and to stimulate Tau
aggregation in an in vitro assay.33,34
Our group has provided Xꢀray crystallographic evidence for the preferential, bivalent binding
of 14ꢀ3ꢀ3 proteins to phosphoepitope sites, pS214 and pS324, with affinity in the µM range
according to additional biochemical and biophysical data.11 We additionally showed that the
interaction could be decreased through 14ꢀ3ꢀ3σ overexpression in SHꢀSY5Y cells.11 Taken
together, these data led us to hypothesize that 14ꢀ3ꢀ3 binding to phosphorylated Tau (pTau)
4
ACS Paragon Plus Environment

Page 5 of 57
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
occurs at the expense of a stabilizing PPI between Tau and microtubules,11 which is
concomitantly known to be decreased by the phosphorylation of Tau on S214.35 Small molecule
stabilization of microtubules (MTs) is a promising therapeutic strategy to compensate for the loss
of Tauꢀmediated stabilization.36,37 Small molecule inhibition of Tau aggregation, e.g. using cell
permeable Dꢀpeptides,38 may also prove to be a complementary therapeutic strategy.
Alternatively, inhibition of 14ꢀ3ꢀ3/Tau, for example through the targeted development of
modified peptide inhibitors, could potentially exert a neuroprotective effect by decreasing its
phosphorylation level, while increasing the pool of soluble Tau protein for the stabilization of
microtubules and by preventing its aggregation – the 14ꢀ3ꢀ3 PPI hypothesis.
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
Towards testing our 14ꢀ3ꢀ3 PPI hypothesis, we previously demonstrated the potential to inhibit
14ꢀ3ꢀ3/Tau in vitro, with small molecules developed using a structureꢀguided approach.12 With
the Tauꢀderived pS214 phosphopeptide epitope as the chemical starting point, we targeted
chemical modifications specifically at the fusicoccin (FC) binding site of Mode III 14ꢀ3ꢀ3 PPIs39
to produce a potent inhibitor of 14ꢀ3ꢀ3ζ binding to fullꢀlength (fl)ꢀpTau. In this paper, we report
the results of studies in which we attempt structureꢀguided optimization of our Tauꢀderived 14ꢀ3ꢀ
3 inhibitors. Most notably we were able to improve the activity of the lead compound from our
previous study, and characterize two distinct high affinity modes of interaction by fluorescence
polarization (FP) and isothermal calorimetry (ITC) measurements, which we correlate to an
“open” and “closed” state based on seven new Xꢀray coꢀcrystal structures. Both binding modes
were shown to inhibit the binding of 14ꢀ3ꢀ3ζ to flꢀpTau in a concentrationꢀdependent manner,
lending further weight to our 14ꢀ3ꢀ3 PPI hypothesis.
5
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 6 of 57
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
c
R
ProꢀR
H
H
Symmetric S-configured
benzhydryl pyrrolidine
Asymmetrically substituted
benzhydryl pyrrolidine
N (S)
H
(S)
N
H
ProꢀS
Figure 1. a) Zoomed in perspective of deepꢀlying pocket, adjacent to the fusicoccin A (FC)
pocket, present in previously published coꢀcrystal structure of a synthetic Tau peptide, modified
with a benzhydryl pyrrolidine moiety, bound to 14ꢀ3ꢀ3σꢁC (PDB: 5HF3).12 The modified Tau
peptide is depicted in white sticks, the protein surface and residues in orange and the protein
interior surface in dark grey. Pocketꢀforming amino acid residues are labelled. b) The same deepꢀ
lying pocket present in the ternary complex of 14ꢀ3ꢀ3σꢁC bound to the Cꢀterminal ERα
phospopeptide and FC. The ERα peptide is depicted in green sticks, FC is depicted in blue sticks,
the protein surface and residues in orange and the protein interior surface in dark grey. c)
6
ACS Paragon Plus Environment

Page 7 of 57
ACS Chemical Neuroscience
1
2
3
4
Chemical structure of symmetric Sꢀconfigured benzhydryl pyrrolidine & general structure of
5
6
asymmetrically substituted benzhydryl pyrrolidines.
7
8
9
Results and Discussion
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
Rational design of new 14ꢀ3ꢀ3 inhibitors. In a previous study,8 we discovered that a synthetic
derivative of a pS214ꢀTau peptide epitope modified at the Cꢀterminus with a benzhydryl
pyrrolidine moiety bound more strongly to the 14ꢀ3ꢀ3 protein than the unmodified Tau epitope.
In the resulting crystal structure the benzhydryl moiety – specifically, the proꢀR phenyl ring –
can be seen to occupy the fusicoccin (FC) pocket (Figure 1A), thus explaining the improved
activity observed in the biochemical and biophysical assays. Closer inspection of the FC pocket
identified a deepꢀlying pocket proximal to the proꢀR phenyl group lined by seven amino acid
residues derived from the same 14ꢀ3ꢀ3 protomer – Ser45, Tyr48, Lys49, Phe119, Lys122,
Met123, and Tyr 127 (Figure 1A). Interestingly, a similar pocket is also present in other 14ꢀ3ꢀ3
protein–ligand crystal structure complexes (e.g. the 14ꢀ3ꢀ3–ERα complexed stabilized by FC –
Figure 1B).22 The ortho and meta positions of the proꢀR phenyl ring are located closest to the
deepꢀlying pocket (Figure 1C). Based on these observations, we hypothesized that the affinity of
our modified peptides might be further improved through structural variation of the benzhydryl
pyrrolidine group, potentially addressing the deepꢀlying pocket in the process (Figure 1C). We
therefore targeted the synthesis of a structurally diverse collection of monoꢀsubstituted
benzhydryl pyrrolidine analogs (Tables 1ꢀ3), in which the phenyl substituent group differed in
size and the polarity (i.e. H → Cl → Me → OMe → OCH2CH2OCH3 – see Table 1).
7
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 8 of 57
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
Scheme 1. Synthesis of asymmetrically substituted benzhydryl pyrrolidine derivatives 2.6aꢀg
(Table 1).
Synthesis of substituted benzhydryl pyrrolidine derivatives. The synthesis of
asymmetrically substituted benzhydryl pyrrolidine derivatives belonging to the generic structure
depicted in Figure 1C was complicated by the absence of any literature precedent. Therefore, we
elected for a synthesis based on the stereoretentive synthesis of symmetrically substituted (S)ꢀ2ꢀ
diphenylmethylpyrrolidine (Scheme 1),40 starting from enantiopure Lꢀproline ester, though
cognizant of the lack of obvious stereocontrol in the formation of the asymmetric centre at the
benzhydryl carbon, and therefore the likely formation of diastereomers.
In brief, the addition of phenylmagnesium bromide to commercial Weinreb amide 2.1,
followed by acid workꢀup produced benzoylpyrrolidine 2.2 in a 47% yield over the two steps
(Scheme 1). Deprotection of the Boc group yielded the substituted pyrrolidine 2.3, which was
followed by protection of the amine group as the ethyl carbamate 2.4. At this juncture, treatment
of 2.4 with a range of structurally diverse substituted Grignard reagents (Table 1) produced a
small library of pyrrolooxazolones 2.5aꢀg, in isolated yields ranging from 25ꢀ86%. The Grignard
8
ACS Paragon Plus Environment

Page 9 of 57
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
reagents used to make pyrrolooxazolones 2.5c and 2.5d are derived from 1ꢀbromoꢀ2ꢀ(2ꢀ
methoxyethoxy)benzene and 1ꢀbromoꢀ3ꢀ(2ꢀmethoxyethoxy)benzene, respectively, which could
be themselves prepared in one step from commercial compounds (supporting information).
These were converted to the corresponding asymmetrically substituted benzhydryl pyrrolidines
2.6aꢀg using palladium catalyzed hydrogenation conditions, a step which had been shown to
occur without loss of steropurity for the synthesis of (S)ꢀ2ꢀdiphenylmethylpyrrolidine, as
evidenced by chiral HPLC and Xꢀray analysis.40 For the series 2.5aꢀg, we were unable to detect
more than one diastereomer by either NMR or LCꢀMS. For the series 2.6aꢀg by contrast, analogs
2.6cꢀf yielded diastereomers, which were separable by reverseꢀphase (RP) HPLC, with combined
yields in the range of 33ꢀ71% and diastereomeric ratios (drs) in the range 63:37ꢀ86:14 (Table 1).
We were unable to detect the formation of diastereomers in the case of analogs 2.6a, 2.6b and
2.6g.
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
9
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 10 of 57
1
2
3
4
5
6
Table 1. Summary of structures, yields & diastereomeric
ratios (dr) for pyrrolooxazolone (2.5aꢀg) and benzhydryl
derivatives (2.6aꢀg) described in Scheme 1.
7
8
9
2.5
2.6
dr[b]
Substituent
group
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
%
dr
%
Derivative
yield[a]
yield[a]
I
II
77[c]
a
b
c
86
41
57
ꢀ
ꢀ
ꢀ
63:37
76:24
59[c]
4
44 74:26[d]
d
80
ꢀ
55 16 72:28[d]
17 28 63:37[d]
e
f
32
25
86
ꢀ
ꢀ
ꢀ
26
7
86:14[d]
66:34
g
81[c]
[a] isolated yields of diastereomers either combined or
separated by RPꢀHPLC (denoted I or II). [b] dr determined
by comparing integral values in H NMR of crude. [c]
Inseparable diastereromers. [d] diastereomers separable by
RPꢀHPLC (Supporting Information).
1
Synthesis of the first library of modified Tau peptides. With the library of asymmetrically
substituted pyrrolidine derivatives in hand, we proceeded with the synthesis of the corresponding
modified Tau peptides. The synthesis of 3.2a has been described in a previous communication.8
The appearance of the Rꢀepimer in the coꢀcrystal structure of 3.2a with 14ꢀ3ꢀ3σ – likely caused
by racemization of the threonine αꢀstereocentre during the synthesis – was unexpected because
10
ACS Paragon Plus Environment

Page 11 of 57
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
the Sꢀepimer was observed to bind 14ꢀ3ꢀ3σ in coꢀcrystal structures for analogous structures
reported in same study, prepared using the same synthetic strategy.12 An inspection of all
structural data, superposed, suggested that the bulky benzhydryl moiety disfavors binding of the
Sꢀepimer of 3.2a through a steric clash between the benzhydryl group and the threonine
sidechain residue (an effect presumably absent in the case of the Rꢀepimer). To test this
hypothesis, and specifically probe the steric and stereochemical preferences of the modified Tau
peptides for 14ꢀ3ꢀ3 binding, we synthesized a library of analogous modified Tau peptides in
which the Cꢀterminal LꢀThr (3.2a) had been systematically replaced by either Gly (3.2b) or a
short series of other Sꢀ and Rꢀconfigured natural amino acid residues with different sidechains –
LꢀAla (3.2c), DꢀAla (3.2d), LꢀVal (3.2e) and DꢀVal (3.2f) – 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. Synthesis of modified Tau peptides 3.2aꢀf (Table 2) and 4.2aꢀg (Table 3). For the
synthesis of 3.2aꢀf, X = LꢀThr (3.2a), Gly (3.2b), LꢀAla (3.2c), DꢀAla (3.2d), LꢀVal (3.2e), Dꢀ
Val (3.2f) – see Table 2. For the synthesis of 4.2aꢀg, R = see Table 3.
Partially protected Tau peptides 3.1aꢀf were first synthesized as described for a previous
synthesis of 3.1a (refered to as 3b in reference),12 and characterized by LCꢀMS analysis
11
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 12 of 57
1
2
3
4
5
6
7
8
9
(supporting information). Each partially protected peptide was then coupled to (S)ꢀbenzhydryl
pyrrolidine using PyClock as coupling reagent followed by resin cleavage and deprotection of
the sideꢀchain protecting groups (TFA, TIS, H2O) to afford 3.2aꢀf in yields of 8ꢀ36% after
purification by RPꢀHPLC (Scheme 2). While we could not conclusively exclude the possibility
of diastereomer impurities in the final compounds, all modified Tau peptides were purified by
RPꢀHPLC on an optimized gradient (see Methods section). Please refer to the Supporting
Information for LCꢀMS spectra of all compounds after purification.
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
Biochemical evaluation of first library of modified Tau peptides. We next investigated the
activity of our first library of modified Tau peptides in a competitive fluorescence polarization
(FP) assay – Table 2 and Figure S34 (Supporting Information) – and compared their activities to
that of our reference compound, 3.2a. The FP data shows that under the specific assay conditions
used, all new modified peptides (3.2bꢀf) inhibit binding of the competitor FAMꢀlabeled
diphosphorylated Tau competitor peptide to 14ꢀ3ꢀ3ζ, with IC50 values in the same low
micromolar range as the reference, 3.2a (Table 2).
12
ACS Paragon Plus Environment

Page 13 of 57
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
Table 2. Summary of structures, and associated IC50
(FP), and Kd values for the modified Tau peptides 3.2aꢀ
f, the synthesis of which is described in Scheme 2. The
confidence interval (CI) & ± standard error (SE) are
reported.
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
FP
ITC
Kd/
Amino acid
residue, X
(Scheme 2)
Derivative
IC50/ 95%
(±
ꢂM SE)
ꢂM
CI
LꢀThr
Gly
8.1 7.3ꢀ8.9 5.4 0.4
5.9 5.2ꢀ6.8 5.6 0.5
5.3 4.7ꢀ5.8 3.3 0.7
6.2 5.8ꢀ6.6 3.0 0.6
6.2 5.8ꢀ6.5 2.2 0.7
7.6 7.2ꢀ8.1 3.4 0.6
3.2a
3.2b
3.2c
3.2d
3.2e
3.2f
LꢀAla
DꢀAla
LꢀVal
Dꢀval
ITC measurements were next performed in duplicate on modified Tau peptides 3.2aꢀf to
determine their association constant (Ka), stoichiometry (N), and enthalpy (∆H) and entropy
change (∆S) on binding to 14ꢀ3ꢀ3 (Supporting Information). Although the two sets of duplicate
measurements are consistent with one another, one set of data are used here for a quantitative
comparison of the different analogs. Collectively, the calculated Kd values for analogs 3.2aꢀf
(Table 2) are of the same magnitude as the IC50 values determined by FP. The stoichiometries of
binding (N) are also all approximately 1.0, which indicates a 1:1 binding between modified
peptide and protein. Individually, the Kd (2.2ꢀ5.6 ꢂM) & ꢁG (ꢀ7.5 – ꢀ8.1 kcal molꢀ1) values for
each modified peptide are also very similar across the series, which indicates that they all bind
with similar affinity to 14ꢀ3ꢀ3. In the two cases where the activities of Lꢀ and Dꢀisomers could be
directly compared – for Ala and Val analogs –similar binding data was measured e.g. compare
the data for the LꢀAla analog, 3.2c (Kd = 3.3 ꢂM, ꢁG = ꢀ7.8 kcal molꢀ1), and the DꢀAla analog,
3.2d (Kd = 3.0 ꢂM, ꢁG = ꢀ7.8 kcal molꢀ1). There is additional evidence that increasing sterics and
13
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 14 of 57
1
2
3
4
5
6
7
8
9
hydrophobicity at the Cꢀterminal amino acid produce marginal gains in activity in the series
Gly/3.2b (Kd = 5.6 ꢂM, ꢁG = ꢀ7.5 kcal molꢀ1) → LꢀAla/3.2c (Kd = 3.3 ꢂM, ꢁG = ꢀ7.8 kcal molꢀ1)
→ LꢀVal/3.2e (Kd = 2.2 ꢂM, ꢁG = ꢀ8.1 kcal molꢀ1). The thermodynamic binding parameters –
∆H and ꢀT∆S – are also similar for all Tau peptide inhibitors in this library (Supporting
Information), with the ∆H value in the range ꢀ1.7 – ꢀ2.5 kcal molꢀ1, and the T∆S value in the
range 5.0ꢀ6.6 kcal molꢀ1. Notably, the Kd, ∆H, T∆S and ∆G values are near identical bearing
either threonine (3.2a) or glycine (3.2b) at the Cꢀterminus of the modified Tau peptide.
Collectively, the similar FP and ITC data imply that the analogs 3.2aꢀf are binding to 14ꢀ3ꢀ3 with
a similar binding mode, the hypothesis being that the benzhydryl group in each case binds the FC
pocket in a mode similar to the one observed in the crystal structure for 3.2a (Figure 1A).12
Furthermore, increasing the steric bulk and hydrophobicity produces a marginal increase in
activity, while inverting the stereochemistry of the Cꢀterminal amino acid residue importantly
has no significant effect on the modified Tau peptide’s binding and inhibitory properties.
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
Synthesis of the second library of modified Tau peptides. In view of the measured
equipotency and similar thermodynamic profiles of the LꢀThr (3.2a) and Gly analogs (3.2b), we
elected to prepare a second library of modified Tau peptides based on 3.2b. We reasoned that
replacing LꢀThr with Gly would simplify the synthesis while retaining the ability to explore the
FC pocket, potentially addressing the adjacent deepꢀlying pocket (Figure 1A) through the
introduction of asymmetrically substituted benzhydryl pyrrolidine derivatives. A second library
of modified Tau peptides, 4.2aꢀg, was therefore synthesized, of which the details are outlined in
Scheme 2 and the structures summarized in Table 3. For some of the derivatives, two
diastereomers were separated by RPꢀHPLC – denoted I and II – in yields of between 3ꢀ32% for
14
ACS Paragon Plus Environment

Page 15 of 57
ACS Chemical Neuroscience
1
2
3
4
5
6
the individual diastereomers, and overall combined isolated yields of between 34ꢀ57% (see
Methods section).
7
8
9
Table 3. Summary of structures, and associated IC50
(FP), and Kd values for the modified Tau peptides
4.2aꢀg described in Scheme 3. The confidence
interval (CI) & ± standard error (SE) are reported.
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
FP
ITC
Substituent
group
Derivative
IC50/ 95%
Kd/ (±
ꢂM SE)
ꢂM
CI
4.2a[a]
4.2b[a]
4.2cꢀI
6.0 5.7ꢀ6.5 8.9 0.7
5.9 5.6ꢀ6.3 9.3 2.0
2.8 2.2ꢀ3.6 2.7 0.3
4.2 4.0ꢀ4.5 9.0 0.5
5.5 5.0ꢀ6.1 6.3 0.6
4.2cꢀII
4.2dꢀI
12.1 10.7ꢀ 20.6 2.4
13.7
4.2dꢀII
5.0 4.8ꢀ5.3 3.6 0.6
4.2eꢀI
4.2eꢀII
4.2fꢀI
10.4
9.6ꢀ
12.7 0.9
11.2
5.8 5.5ꢀ6.0 4.7 0.6
11.4 10.7ꢀ 16.5 2.0
12.2
4.2fꢀII
4.2g[a]
7.2 6.5ꢀ7.8 6.2 1.2
15
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 16 of 57
1
2
3
4
3.2b]
5.9 5.2ꢀ6.8 5.6 0.5
5
6
7
8
[a] A mixture of diastereomers.
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
Biochemical evaluation of second library of modified Tau peptides. As for the first library,
we used FP and ITC to characterize the 14ꢀ3ꢀ3 binding properties of our second library of
modified Tau peptides. The findings of this short study are summarized in Table 3. The FP data
(Figure S35) shows that all analogs from this library function as competitive inhibitors of 14ꢀ3ꢀ
3σ with IC50 values in the range 2.8ꢀ12.1 ꢂM. Our ITC data show that all analogs bind to 14ꢀ3ꢀ
3σ with Kd values in the range 2.7ꢀ20.6 ꢂM and ꢁG values between ꢀ6.7 – ꢀ7.9 kcal molꢀ1. The
most active analog in the series was the 2ꢀ(methoxyethoxy)ꢀ derived 4.2cꢀI (Kd = 2.7 ꢂM/ ꢁG = ꢀ
7.9 kcal molꢀ1) and the least active, its regioisomer, analog 4.2dꢀII (Kd = 20.6 ꢂM/ ꢁG = ꢀ6.7
kcal molꢀ1). Besides 4.2cꢀI, analog 4.2eꢀI (Kd = 3.6 ꢂM/ ꢁG = ꢀ7.7 kcal molꢀ1) and methoxyꢀ
derivative 4.2fꢀI (Kd = 4.7 ꢂM/ ꢁG = ꢀ7.5 kcal molꢀ1) were also strong binders.
Xꢀray crystallography studies. To provide a molecular explanation for the high affinity binding
of our new modified tau peptide inhibitors, the coꢀcrystal structures of symmetrically substituted
benzhydryl derivatives, 3.2d (1.70 Å), and 3.2e (2.00 Å), and asymmetrically substituted variants
4.2b (1.50 Å), 4.2cꢀI (1.40 Å), 4.2eꢀI (1.25 Å), 4.2fꢀI (1.40 Å) and 4.2fꢀII (1.45 Å) bound to 14ꢀ
3ꢀ3σꢁC were solved to their respective resolutions in parentheses (Figure 2 and supporting
information). Figure 2A depicts a superimposition of the three modified Tau peptides, 3.2a
(white),12 3.2d (purple) and 4.fꢀII (green), in their 14ꢀ3ꢀ3ꢀbound state, in which all three peptides
clearly bind with a similar extended mode within the amphipathic groove of the protein.
However, the structural differences between the three peptides at the Cꢀterminus is observed to
16
ACS Paragon Plus Environment

Page 17 of 57
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
perturb the positioning of the peptide backbone and the leucine sideꢀchain Cꢀterminal to the
phosphoserine residue. This general trend is representative for all coꢀcrystallized modified Tau
peptides. In Figures 2BꢀI, a panel comparing zoomedꢀin perspectives of the FC pocket of the
seven different coꢀcrystallized modified Tau peptides can be seen: the two benzyhydylꢀderived
(3.2d & 3.2e) and five asymmetrically substituted analogs (4.2b, 4.2cꢀI, 4.2eꢀI, 4.2fꢀI & 4.2fꢀII),
compared to 3.2a. In contrast to analogs 3.2d and 3.2e, in which the pyrrolidine group were
logically determined to be Sꢀconfigured (derived from commercial (S)ꢀ2ꢀ
(diphenylmethyl)pyrrolidine), all five asymmetrically substituted analogs were unexpectedly
observed to be Rꢀconfigured at the αꢀcarbon of the pyrrolidine ring, counter to our expectations
(Figure 2EꢀI).40 Interestingly as well, all of the five peptides modified with an asymmetrically
substituted benzhydryl group bound in a similar “open” state, which contrasted with the “closed”
state observed in the case of the three symmetric benzhydryl analogs. A closer examination of
this data reveals that the oxygen of the methoxy substituent on 4.2fꢀI (Figure 2H) is sufficiently
close (3.1 Å) to engage in a hydrogen bond with the side chain of residue Asnꢀ42 – an interaction
witnessed previously between the sugar ringꢀoxygen of the natural product fusicoccin A (FC)
and Asnꢀ42 of 14ꢀ3ꢀ3σ for stabilization of TASKꢀ3 (PDB: 5D3F; 2.7 Å)23 and 14ꢀ3ꢀ3ζ for
stabilization of CFTR (PDB: 3P1Q; 3.2 Å).41 By comparison, an analogous hydrogen bond
interaction is absent from the crystal structure of the less potent diastereomer 4.2fꢀII (Figure 2I).
The ITC data for diastereomers 4.2fꢀI & 4.2fꢀII indicate that their 2ꢀfold difference in affinity by
FP and a 3.5ꢀfold difference by ITC is entropically driven (Figure 3 & Table S2 in the supporting
information). An explanation for the potency of analog 4.2cꢀI – the most potent of the series –
can be observed in the crystal structure in which the methoxyethoxy side chain clearly addresses
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
17
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 18 of 57
1
2
3
4
5
6
the FC pocket. A more detailed discussion of all the coꢀcrystal structures can be found in the
supporting information.
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
Stereochemical outcome of modified Tau peptide synthesis. Logically, the Rꢀconfiguration of
the DꢀAlaꢀderived modified Tau peptide was found in the 14ꢀ3ꢀ3/3.2d coꢀcrystal structure
(Figure 2C). Interestingly, the Rꢀepimer was also observed in the coꢀcrystal structure of the Valꢀ
derived modified Tau peptide 3.2e (Figure 2D), which parallels the result obtained previously12
for the LꢀThrꢀderived modified Tau peptide 3.2a (Figure 2A). While the origin of the epimer
formation in both cases remains unclear (both syntheses began with enantiomerically pure Lꢀ
amino acids), it is hypothesized that the Sꢀepimer of βꢀbranched amino acids is sterically less
favored for binding at the peptide Cꢀterminus than the corresponding Rꢀepimer. Significantly, the
introduction of a glycine residue at the same Cꢀterminal position of our peptides, which will not
be dependent on epimerization, did not significantly affect the affinity of our modified Tau
peptides. The electron densities clearly show that the Rꢀepimer is bound to 14ꢀ3ꢀ3 for each of the
five asymmetrically substituted analogs (4.2b, 4.2cꢀI, 4.2eꢀI, 4.2fꢀI & 4.2fꢀII), which was
unexpected because the synthesis of the corresponding monoꢀsubstituted benzhydryl pyrrolidine
in each case commenced with Sꢀconfigured Lꢀproline, and was based on a enantioretentive
synthesis of (S)ꢀ2ꢀ(diphenylmethyl)pyrrolidine.40
However, we did not monitor the
stereochemical outcome at each step during the synthesis of our monoꢀsubstituted benzhydryl
pyrrolidine derivatives.
18
ACS Paragon Plus Environment

Page 19 of 57
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
j
Figure 2. a) Overlay of the coꢀcrystal structures of the modified peptides 3.2a (white)(PDB:
5HF3)12 3.2d (purple) & 4.2fꢀII (green) bound to 14ꢀ3ꢀ3σꢁC. bꢀi) Zoomꢀin of FC pocket for the
coꢀcrystal structures of the modified peptides 3.2a (b),(PDB: 5HF3)12 3.2d (c),(PDB; 6FI5) 3.2e
(d),(PDB; 6FI4) 4.2b (e),(PDB; 6FBY) 4.2cꢀI (f),(PDB; 6FAW) 4.2eꢀI (g),(PDB; 6FAW) 4.2fꢀI
(h),(PDB; 6FAV) & 4.2fꢀII (i)(PDB; 6FBW) bound to 14ꢀ3ꢀ3σꢁC. The stereochemical
assignment (S and R notation) of the Cꢀterminal amino acid and benzhydryl pyrrolidine solved in
complex with the 14ꢀ3ꢀ3 protein have been added to each panel.
19
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 20 of 57
1
2
3
4
5
6
7
8
9
Analysis of thermodynamic parameters of binding. A clear difference was observed between
the thermodynamic parameters for the asymmetrically substituted and those measured for the
symmetrically substituted benzhydryl pyrrolidineꢀmodified Tau peptides. In the case of the
asymmetrically substituted analogs, the enthalpy change (∆H) was in general more negative and
the entropy change (ꢀT∆S) more positive than for the symmetrically substituted benzhydryl
pyrrolidineꢀmodified Tau peptides. For example, while 3.2e (Kd = 2.2 ꢂM, ꢁG = ꢀ8.1 kcal) and
4.2fꢀI gave similar binding parameters (Kd = 4.7 ꢂM, ꢁG = ꢀ7.5 kcal molꢀ1), their corresponding
thermodynamic components significantly diverge (Figure 4). In the case of 3.2e, the entropic
factor (ꢀTꢁS = ꢀ6.6 kcal molꢀ1) contributed significantly more to the free energy of binding than
the enthalpic factor (ꢁH = ꢀ1.5 kcal molꢀ1), while for analog 4.2fꢀI, the opposite was true (i.e. ꢁH
= ꢀ4.7 kcal molꢀ1 & ꢀTꢁS = ꢀ2.8 kcal molꢀ1). The same divergent behavior was observed across
all Tau peptide inhibitors reported in Tables 2 and 3 such that compound binding could be
classified as being either predominantly entropically driven (i.e. the symmetric benzhydryl
analogs 3.2aꢀf) or enthalpically driven (i.e. the asymmetric analogs 4.2aꢀg).
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 structural explanation for the two classifications of ITC data could be found in the
crystallography data (Figure 4). The predominantly entropically driven binding profile
characterized by ITC could be explained by the “closed” binding mode characterized by Xꢀray
crystallography (illustrated by 3.2a, Figure 4A), in which the Sꢀconfigured symmetric
benzhydryl moiety is observed to fill the hydrophobic FCꢀbinding pocket better. In this mode,
the reduced solventꢀexposed hydrophobic surface area results in fewer ordered water molecules
(as evidenced in Figure 4A), which drives binding through a gain in entropy. By contrast, the
20
ACS Paragon Plus Environment

Page 21 of 57
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
predominantly enthalpically driven binding profile (illustrated by 4.2fꢀI, Figure 4B) could be
explained by a more “open” binding mode in the Xꢀray crystal structure, in which the now Rꢀ
configured asymmetrically substituted benzhydryl moiety is found in a more solventꢀexposed
state, which permits the binding of more ordered water molecules (as evidenced in Figure 4B),
resulting in a comparatively lower entropic contribution on binding.
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 similar correlation between the ITC and the crystallography data for all other solved structures
was observed, which suggests that all derivatives bearing the same, either enthalpic or entropic,
binding profile by ITC (whether coꢀcrystallized or not with the 14ꢀ3ꢀ3 protein) bind via the
aforedescribed “closed” or “open” binding modes by Xꢀray crystallography. The reason for the
two different binding modes is most likely for two reasons: first, the difference in R and S
configuration between the asymmetrically substituted benzhydryl pyrrolidine moieties (Rꢀ
configured), which all bind in the “open” state, and the Sꢀconfigured symmetric benzhydryl
pyrroldines, which bind in the “closed” state. Second, a steric clash between the asymmetrically
substituted benzhydryl group and the protein surface may disfavor a “closed” binding mode,
which the ligand would relieve by adopting the more “open” binding state. In view of the ITC
data for the Glyꢀderivative 3.2b, the suggestion is the benzhydryl group of this derivative binds
in a similar “closed” state as observed for e.g. 3.2a, 3.2d and 3.2e. This result would suggest the
added conformational freedom introduced by the Cꢀterminal glycine residue would apparently
not significantly influence the binding mode of 3.2b, and by extension analogs from within the
4.2 series. While the modifications at the Cꢀterminus of our modified Tau peptides do not
address the targeted deepꢀlying pocket of the 14ꢀ3ꢀ3 protein (Figure 1C), as hypothesized, they
have enabled the potent targeting of chemically distinct sites within the adjacent FC pocket.
21
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 22 of 57
1
2
3
4
5
6
7
8
9
0
ꢀ2
ꢀ4
ꢀ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
∆H
ꢀT∆S
∆G
ꢀ8
ꢀ10
Tau peptide inhibitor
Figure 3. Comparison of enthalpic (ꢁH, blue) and entropic contributions (ꢀTꢁS, red) to the
Gibbs free energy of binding (ꢁG, green) at 310 K for modified Tau peptides 3.2a, 3.2b, 3.2e &
4.2fꢀI. For a comprehensive comparison of thermodynamic parameters for all modified Tau
peptides see the supporting information (Table S2).
Figure 4. Crystal structure of a) 3.2a at 1.8 Å resolution (PDB: 5HF3) and b) 4.2fꢀI(PDB;
6FAV) at 1.4 Å resolution, in complex with 14ꢀ3ꢀ3σ. A difference in the amount of water
molecules (red spheres) in the pocket can be observed, which may explain the difference in
entropy observed in the ITC data.
Inhibition of the interaction of 14ꢀ3ꢀ3 with fullꢀlength phosphorylated Tau protein. We next
investigated the potential of the three most potent analogs towards the inhibition of the
22
ACS Paragon Plus Environment

Page 23 of 57
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
interaction between 14ꢀ3ꢀ3ζ and fullꢀlength PKAꢀphosphorylated Tau (pTau). From the analogs
tested, one (3.2e), binds to 14ꢀ3ꢀ3 via an “open” binding mode, while the other two, (4.2cꢀI and
4.2eꢀI), via the “closed” binding mode. NMR spectroscopy was used to assess the modulation of
this PPI in solution, using chemical shift perturbation mapping, based on the assigned resonances
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
of the phosphorylated Tau Hꢀ15N 2D spectrum.42–45 The 2D 15Nꢀ1H HSQC spectra of 15Nꢀ
labelled pTau in the presence of unlabelled 14ꢀ3ꢀ3ζ were acquired with or without each of the
inhibitors. The intensities of the correlation peaks corresponding to specific amino acid residues
along pTau sequence (I) were monitored in each experiment and compared to the intensities of
the corresponding correlation peaks in the spectrum of pTau alone (I0). The binding of 14ꢀ3ꢀ3ζ
to pTau led to peak broadening and consequently, to the decrease of the (I/I0) ratio for
resonances corresponding specifically to residues located in the binding region of 14ꢀ3ꢀ3 (Figure
5A and 5B). The addition of 4.2eꢀI to the solution containing 15N pTau/14ꢀ3ꢀ3ζ led to a doseꢀ
dependent recovery of the intensity of these same resonances (Figure 5A and 5B). This effect is
well illustrated by the resonances corresponding to the amide groups of the PKAꢀphosphorylated
serines of Tau that showed intensity recovery with addition of each analog (Figure 5B, Figures
S36). The same set of 2D experiments was performed for inhibitors 3.2e and 4.2cꢀI resulting in a
similar I/I0 profile (Figure 5, Figures S37 and S38), which confirmed the capacity of these
analogs to decrease the formation of the complex 14ꢀ3ꢀ3ζ/pTau. To get further insights on the
inhibitory effect of these analogs, a series of 1H spectra of 4.2eꢀI was recorded in the presence of
increasing concentrations of 14ꢀ3ꢀ3ζ (Figure 5C). By monitoring one wellꢀisolated resonance of
the spectrum of the smallꢀmolecule, a concentration dependent intensity decrease upon the
addition of 14ꢀ3ꢀ3ζ was observed, which can be attributed to the interaction, as the sharp signal
of the free ligand get broadened when complexed with the protein. Additionally, the spectrum of
23
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 24 of 57
1
2
3
4
5
6
7
8
9
4.2eꢀI was recorded in the presence and absence of pTau, which did not reveal any sign of
interaction (Figure S39) as both spectra were identical. Based on these results, it can be
concluded that 3.2e, 4.2cꢀI and 4.2eꢀI inhibited the interaction pTau/14ꢀ3ꢀ3ζ in a concentration
dependent manner, by binding to 14ꢀ3ꢀ3ζ and competing with pTau.
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
24
ACS Paragon Plus Environment

Page 25 of 57
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
Figure 5. 4.2eꢀI inhibits the binding of pTau to 14ꢀ3ꢀ3ζ in a concentrationꢀdependent manner. A)
15
Selected enlarged regions of the overlaid Nꢀ1H HSQC spectra showing the intensity recovery
25
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 26 of 57
1
2
3
4
5
6
7
8
9
of the correlation peaks correspondent to the amide groups of the PKAꢀphosphorylated Serines
after the addition of 4.2eꢀI. The spectra are shown superimposed to the pTau spectrum, which is
colored in black. With the addition of a 3ꢀfold excess of 4.2eꢀI (considering 14ꢀ3ꢀ3ζ
concentration) it is possible to remark the intensity recovery of the weakest epitopes (pS208,
pS356 and pS416) and finally, with the addition of a 10ꢀfold, it is possible to detect pS214 and
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
pS324. B) Plot of the ratios of the bound (I)/free (I0) Hꢀ15N correlation peak intensities of full
length pTau 60 ꢂM (y axis) versus the amino acid sequence (x axis) in the presence of 14ꢀ3ꢀ3ζ
120 ꢂM (red plot); 14ꢀ3ꢀ3ζ 120 ꢂM + 4.2eꢀI 360 ꢂM (green plot) and 14ꢀ3ꢀ3ζ 120 ꢂM + 4.2eꢀI
1200 ꢂM (blue plot). A total of 155 correlation peak intensities are shown. The x axis is not
proportional. The domains of fullꢀlength pTau (Nꢀter for Nꢀterminal; PRD for ProlineꢀRich
Domain; MTBR for Microtubule Binding Region; Cꢀter for Cꢀterminal) and the phosphorylation
sites (S208, S214, S324, S356 and S416) are identified below the x axis. C) Section of overlaid
1H spectra of 4.2eꢀI 150 ꢂM alone (blue) and in the presence of 14ꢀ3ꢀ3ζ 10 ꢂM (red), 30 ꢂM
(green), 75 ꢂM (purple) and 150 ꢂM (gold). The enlarged region shows a wellꢀisolated
resonance correspondent to a 4.2eꢀI proton.
Conclusions.
Here, we describe the synthesis of a novel seventeenꢀmembered collection of modified Tau
peptide inhibitors, targeting increase binding affinity at the fusiccocin A (FC) pocket and
proximal deepꢀlying pocket. In the case of the peptides bearing a symmetric, Sꢀconfigured
benzhydryl pyrroldine group, exchanging the Cꢀterminal Thr residue during the synthesis with a
number of other Lꢀ and Dꢀamino acid residues at the Cꢀterminus of the Tau peptide, including
nonꢀchiral Gly, did not produce a significant change in the IC50 and Kd and associated
26
ACS Paragon Plus Environment

Page 27 of 57
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
thermodynamic parameters. This result suggested that the Cꢀterminal amino acid in the modified
Tau peptide can be flexibly replaced by amino acids bearing different side chains and
stereochemistry. All resulting Tau peptide analogs (Table 2 & 3) were active inhibitors of 14ꢀ3ꢀ
3/Tau. Interestingly, using a combination of FP, ITC and Xꢀray crystallography data, we
characterized two binding modes for our modified Tau peptides – one “closed” entropicallyꢀ
driven state, the other an “open” enthalpic state at the peptide Cꢀterminus. The different binding
modes is most likely caused by the inverted stereochemistry at the αꢀcarbon between the
commercial Sꢀconfigured symmetric benzhydryl pyrrolidine (“closed”) and the synthetic Rꢀ
configured asymmetric benzhydryl pyrrolidine (“open”) – an unexpected outcome of the
synthesis – and steric factors induced by monoꢀsubstitution of the benzhydryl group. While
neither of these two modes were capable of addressing the deepꢀlying pocket (Figure 1C), both
inhibited 14ꢀ3ꢀ3ꢀbinding to fullꢀlength PKAꢀphosphorylated Tau protein in vitro. Considering
the manner in which the Tauꢀderived phosphoepitope has been studied presently, it could also be
envisaged as a nonꢀcovalent tether to investigate new chemotypes to address the FC pocket, thus
potentially opening the door to new nonꢀphosphorylated inhibitors or novel stabilizers of Mode
III 14ꢀ3ꢀ3 PPIs.
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
Methods
Synthesis of substituted benzhydryl pyrrolidine derivatives
General methods
Unless otherwise stated, all solvents employed were commercially available and used without
purification. Water was purified using a Millipore purification train. Dry solvents were obtained
27
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 28 of 57
1
2
3
4
5
6
7
8
9
from a MBRAUN Solvent Purification System (MBꢀSPSꢀ800). Deuterated solvents were
obtained from Cambridge Isotope laboratories. All reagents were commercially available,
supplied by SigmaꢀAldrich, and used without purification. NMR data were recorded on a Bruker
Cryomagnet for NMR spectroscopy (400 MHz for 1HꢀNMR and 100 MHz for 13CꢀNMR). Proton
experiments were reported in parts per million (ppm) downfield of TMS. All 13C spectra were
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
relative to the residual chloroform signal (77.16 ppm). HꢀNMR spectra are reported as follows:
chemical shift, multiplicity (s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet, dd =
doublet of doublets, td = triplet of doublets), integration, and coupling constant (J) in Hertz (Hz).
Analytical Liquid Chromatography coupled with Mass Spectrometry (LCꢀMS) was performed on
a C4 Jupiter SuC4300A 150 x 2.0 mm column using H2O with 0.1% Formic Acid (F.A.) and
Acetonitrile with 0.1% F.A., in general with a gradient of 5% to 100% Acetonitrile in H2O in 15
min (connected to a Thermo Fischer LCQ Fleet Ion Trap Mass Spectrometer). Preparative High
Pressure Liquid Chromatography (HPꢀLC) was performed on a Gemini S4 110A 150 x 21.20
mm column using H2O with 0.1% F.A. and Acetonitrile with 0.1% F.A. Silica column
chromatography was performed manually using silica gel with particle size 60–200 µm (60 Å).
Reaction progress was monitored by thinꢀlayer chromatography using Merck TLC silica gel 60
F254 plates.
tertꢀbutyl (S)ꢀ2ꢀbenzoylpyrrolidineꢀ1ꢀcarboxylate (2.2). To an oven dried 100 mL twoꢀnecked
flask was added magnesium turnings (1.18 g, 48.4 mmol) and anhydrous THF (10 mL) under
argon pressure. A small amount of iodine was added followed by the slow addition of
bromobenzene (3.80 g, 24.2 mmol). The reaction was slowly heated using a heat gun to start the
reaction and it was then stirred at room temperature for 30 min. Subsequently, the reaction was
28
ACS Paragon Plus Environment

Page 29 of 57
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
cooled down to 0 °C, at which time the Weinreb amide 2.1 (2.50 g, 9.7 mmol) in THF (6 mL)
was added slowly. The resultant mixture was stirred at 0 °C for 3 h, then quenched with saturated
NH4Cl (15 mL) and extracted with EtOAc (3 x 30 mL). The combined organic layers were
washed with brine, dried over Na2SO4, filtered and concentrated in vacuo. The product was
purified by column chromatography, eluting with heptane/EtOAc 72:28 v/v to yield tertꢀbutyl
(S)ꢀ2ꢀbenzoylpyrrolidineꢀ1ꢀcarboxylate (2.2) as a white powder (1.24 g, 4.50 mmol, 47%).
Silica gel TLC Rf = 0.23 (Heptane/EtOAc 72:28 v/v); LCꢀMS (ESI): calc. for C16H21NO3 [M+Hꢀ
BOC]+: 176.10, observed 176.25, LC, Rt = 6.76 min; 1H NMR (400 MHz, CDCl3): δ (ppm) 8.02
– 7.91 (m, 2H), 7.62 – 7.52 (m, 1H), 7.51 – 7.41 (m, 2H), 5.37 – 5.16 (m, 1H), 3.73 – 3.42 (m,
2H), 2.40 – 2.24 (m, 1H), 2.02 – 1.85 (m, 3H), 1.47 (s, 9H); 13C NMR (100 MHz, CDCl3): δ
198.92, 154.45, 135.29, 133.19, 128.70, 128.69, 128.52, 128.51, 79.78, 61.36, 46.81, 29.82,
28.21, 28.20, 28.19, 24.18.
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
(S)ꢀphenyl(pyrrolidinꢀ2ꢀyl)methanone (2.3). The tꢀBocꢀprotected amine 2.2 (1.14 g, 4.1 mmol)
was dissolved in dichloromethane (25 mL) and a solution of trifluoroacetic acid (10 mL) (70:30
v/v) was added. The reaction was stirred at room temperature for 2 hours. The solvent was
evaporated, yielding (S)ꢀphenyl(pyrrolidinꢀ2ꢀyl)methanone (2.3) (718 mg, 4.1 mmol, 99%),
which was directly used without purification. LCꢀMS (ESI): calc. for C11H13NO [M+H]+:
176.10, observed 176.25, LC, Rt = 2.38 min; 1H NMR (400 MHz, CDCl3): δ (ppm) 7.98 (d, J =
7.6 Hz, 2H), 7.72 (t, J = 7.6 Hz, 1H), 7.60 – 7.52 (m, 2H), 5.56 – 5.47 (m, 1H), 3.69 – 3.52 (m,
2H), 2.80 – 2.68 (m, 1H), 2.30 – 2.18 (m, 1H), 2.16 – 1.97 (m, 2H); 13C NMR (100 MHz,
CDCl3): δ 193.67, 135.79, 131.26, 129.33, 129.32, 129.21, 129.19, 63.18, 47.47, 30.22, 24.32.
29
ACS Paragon Plus Environment