Targeted Covalent Inhibition of Prolyl Oligopeptidase (POP): Discovery of Sulfonylfluoride Peptidomimetics

Article abstract of DOI:10.1016/j.chembiol.2018.04.013

Prolyl oligopeptidase (POP), a serine protease highly expressed in the brain, has recently emerged as an enticing therapeutic target for the treatment of cognitive and neurodegenerative disorders. However, most reported inhibitors suffer from short durati

Full text of DOI:10.1016/j.chembiol.2018.04.013

Brief Communication
Targeted Covalent Inhibition of Prolyl Oligopeptidase
(POP): Discovery of Sulfonylfluoride
Peptidomimetics
Graphical Abstract
Authors
Salvador Guardiola, Roger Prades,
Laura Mendieta, ..., Teresa Tarrag o´ ,
Rob M.J. Liskamp, Ernest Giralt
Correspondence
robert.liskamp@glasgow.ac.uk
(
R.M.J.L.),
ernest.giralt@irbbarcelona.org (E.G.)
In Brief
Prolyl oligopeptidase (POP) is a neuronal
enzyme involved in cognitive disorders.
By combining a peptidomimetic scaffold
with a mild electrophile, Guardiola et al.
designed highly active covalent inhibitors
that are not only potent and selective but
display optimal properties for accessing
the brain by passive diffusion.
Highlights
d
d
d
d
Targeted covalent POP inhibitors are built on a prolylsulfonyl
fluoride warhead
Binding is fast, irreversible, and specific for the catalytic Ser
residue
The best hit (18) shows low-nanomolar cellular potency
and >1,0003 selectivity
Their high membrane permeability grants access to the CNS
Guardiola et al., 2018, Cell Chemical Biology 25, 1–7
July 19, 2018 ª 2018 Published by Elsevier Ltd.
https://doi.org/10.1016/j.chembiol.2018.04.013

Please cite this article in press as: Guardiola et al., Targeted Covalent Inhibition of Prolyl Oligopeptidase (POP): Discovery of Sulfonylfluoride Peptido-
mimetics, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.04.013
Cell Chemical Biology
Brief Communication
Targeted Covalent Inhibition
of Prolyl Oligopeptidase (POP):
Discovery of Sulfonylfluoride Peptidomimetics
1
2
2
3
3
1
Salvador Guardiola, Roger Prades, Laura Mendieta, Arwin J. Brouwer, Jelle Streefkerk, Laura Nevola,
2
3,4,
1,5,6,
Teresa Tarrag o´ , Rob M.J. Liskamp, * and Ernest Giralt *
Institute for Research in Biomedicine (IRB Barcelona), The Barcelona Institute of Science and Technology, Baldiri Reixac, 10, 08028
1
Barcelona, Spain
2Iproteos, S.L., Barcelona Science Park, Baldiri Reixac 10, 08028 Barcelona, Spain
3Department of Chemical Biology and Drug Discovery, Utrecht Institute for Pharmaceutical Sciences, Faculty of Science, Utrecht University,
3
508 TB Utrecht, the Netherlands
4School of Chemistry, University of Glasgow, University Avenue, Glasgow G12 8QQ, UK
5Department of Inorganic and Organic Chemistry, University of Barcelona, Barcelona, Spain
6Lead contact
*Correspondence: robert.liskamp@glasgow.ac.uk (R.M.J.L.), ernest.giralt@irbbarcelona.org (E.G.)
https://doi.org/10.1016/j.chembiol.2018.04.013
SUMMARY
recognition of the target; and (2) high affinity, by incorpor-
ating mild electrophiles and transforming them into covalent
Prolyl oligopeptidase (POP), a serine protease highly inhibitors.
expressed in the brain, has recently emerged as an
enticing therapeutic target for the treatment of cogni-
tive and neurodegenerative disorders. However,
most reported inhibitors suffer from short duration
of action, poor protease selectivity, and low blood-
brain barrier (BBB) permeability, which altogether
limit their potential as drugs. Here, we describe the
structure-based design of the first irreversible,
selective, and brain-permeable POP inhibitors. At
Here, we focused this approach on the inhibition of prolyl
oligopeptidase (POP), an 81-kDa cytosolic protease that is
prominently expressed in the brain with the capacity to hydro-
lyze post-proline bonds of small peptides (Yoshimoto et al.,
1
983). Besides its enzymatic function, POP plays important
roles in neurogenesis, brain plasticity, and memory formation
H o¨ fling et al., 2016), although the exact mechanisms are not
fully known. Being a highly dynamic protein (L o´ pez et al.,
016), recent studies point to the involvement of POP in
protein-protein interactions (PPIs) with other partners such
(
2
low-nanomolar concentrations, these covalent pep- as GAP43, a protein involved in neuronal function (Di Daniel
tidomimetics produce a fast, specific, and sustained et al., 2009), and a-synuclein, an intrinsically disordered pro-
inactivation of POP, both in vitro and in human cells. tein associated with Parkinson disease (PD) (Brandt et al.,
2
008). In agreement with these findings, POP was shown to
accelerate a-synuclein aggregation in vitro and in cell cultures
Savolainen et al., 2015), thus providing an explanation for the
co-localization of both proteins in the brain of patients with PD
More importantly, they are >1,000-fold selective
against two family-related proteases (DPPIV and
FAP) and display high BBB permeability, as shown
in both lipid membranes and MDCK cells.
(
(
Hannula et al., 2013).
Due to its therapeutic potential, numerous compounds have
been developed as POP inhibitors and tested in preclinical set-
tings. Notably, in mouse models of PD, they have been able to
accelerate the clearance of aggregated a-synuclein in the brain
INTRODUCTION
As tools for chemical biology, peptides often show low and, more importantly, to restore the motor behavior that is
permeability across biological barriers and rapid dissociation impaired as a result of the disease (Svarcbahs et al., 2016).
rates, two factors that limit their potential to target intracellular Also recently, POP inhibitors have been shown to significantly
proteins (Huhn et al., 2016). On the other hand, small improve the cognitive symptoms affected by schizophrenia in
molecules can achieve higher potency and permeability, but mice (Prades et al., 2017). Despite these promising results,
they have sometimes proved insufficient for the selective POP inhibitors have failed to succeed in clinical trials, two
manipulation of proteins that share high analogy with other main factors limiting their success: poor selectivity against
members of the family, such as kinases and proteases. related prolyl peptidases (causing side effects) and lack of
These issues have stimulated research (Baillie, 2016) on the permeability through the blood-brain barrier (BBB), which limits
use of modified peptides (peptidomimetics) as an alternative their distribution inside the central nervous system (CNS) (L o´ pez
class of covalent inhibitors able to encompass two main et al., 2013). Here, we address these key issues through the
components: (1) better selectivity, by retaining the main design and characterization of a new class of covalent POP
features of peptides—i.e., topological diversity and selective inhibitors.
Cell Chemical Biology 25, 1–7, July 19, 2018 ª 2018 Published by Elsevier Ltd.
1

Please cite this article in press as: Guardiola et al., Targeted Covalent Inhibition of Prolyl Oligopeptidase (POP): Discovery of Sulfonylfluoride Peptido-
mimetics, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.04.013
A
B
F476
P3
P2
P1
C255
Y473
S2
S3
S1
M235
W595
R643
S554
F173
C
Figure 1. Structure-Guided Design of Irreversible POP Inhibitors
(
(
(
A) Crystal structure of KYP-2047 bound to the POP active site (PDB: 4AN0), highlighting the main enzyme cavities (S1, S2, and S3) and contiguous residues.
B) Representative examples of covalent POP inhibitors.
C) General structure of the family of shape-complementary POP inhibitors. Selected P2 and P3 groups are combined with a sulfonyl fluoride electrophile to
generate irreversible binding peptidomimetics.
STRUCTURE-GUIDED DESIGN OF IRREVERSIBLE POP
INHIBITORS
that fit tightly around the substrate proline rings and account
for the high specificity of the enzyme (F u€ l o¨ p et al., 1998). The
S3 pocket, in contrast, is a larger cavity flanked by hydrophobic
In recent years, numerous POP inhibitors have been reported residues Phe173, Met235, Cys255, Ile591, and Ala594. For the
from a variety of scaffolds such as natural products (Tarrago design of our shape-complementary inhibitors, we have divided
et al., 2007), small molecules (Mariaule et al., 2016), peptides, the molecules into three parts (P1, P2, and P3) and followed a
and peptidomimetics (L o´ pez et al., 2013). Some of them feature rational strategy, considering some of the best pharmacophores
C-terminal reactive functionalities, such as aldehyde, hydroxy- for potency and selectivity, together with a new electrophilic
methyl ketone, or nitrile, which covalently bind to the hydroxyl proline derivative.
group of the catalytic serine (Ser554). Structure-activity studies
At position P3, POP inhibitors often present a single phenyl
have shown that these covalent inhibitors are more potent and ring, such as the Cbz group of ZPP (Yoshimoto et al., 1985).
effective than their non-covalent analogs (Juillerat-Jeanneret, Instead of the carbamate of the Cbz, a three-carbon aliphatic
2
008). However, these molecules form a transient covalent chain—as in KYP-2047 (Ven a€ l a€ inen et al., 2006)—has been
bond with the enzyme that is hydrolyzed after a short time, found optimal for POP inhibition (Figures 1A and 1B). However,
thus regaining enzymatic activity. Thus, a main goal in our struc- careful structural analysis reveals that the S3 pocket is large
ture-based design was to endow our molecules with longer enough to accommodate bulkier aromatic residues, such as
target residence times, as well as improved selectivity and mem- Fmoc (Li et al., 1996), which can provide selectivity against other
brane permeability.
prolyl peptidases. For this purpose, we have included in our
The active site of POP comprises three well-defined cavities: designs a fluorinated gallic acid derivative (4-(benzyloxy)-3,
S1, S2, and S3 (Schechter and Berger, 1967) that accommodate 5-difluorobenzoyl [BdFB]), which in our hands provided not
the peptide substrate and place the C-terminal prolyl carbox- only high selectivity but also extended microsomal stability and
ylate in close proximity to the catalytic Ser554 (Figure 1A). high lipid permeability (Giralt et al., 2014). At position P2,
From a structural perspective, S1 and S2 are narrow cavities L-proline or close analogs are optimal for activity (L o´ pez et al.,
2
Cell Chemical Biology 25, 1–7, July 19, 2018

Please cite this article in press as: Guardiola et al., Targeted Covalent Inhibition of Prolyl Oligopeptidase (POP): Discovery of Sulfonylfluoride Peptido-
mimetics, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.04.013
Table 1. Structure of Designed POP Inhibitors, Their Activity, and Permeability Profile
1
2
a
b
b
ꢀ6
ꢀ1 b
Id
7
R
R
Yield (%)
75
POP IC50 (nM)
Transport (%)
Retention (%)
37.1
19.7
2.1
P
e
(10 cm s )
H
F
H
1.3
4.5
21.2
7.9
0.8
3.2
1.9
2.3
2.1
1.9
1.8
0.9
1.7
1.2
3.9
19.1
11.4
11.2
16.3
7.8
8
H
51
18.8
9
F
F
41
24.8
1
1
1
1
1
1
1
1
1
1
2
0
H
H
H
F
CF
3
48
14.0
40.7
14.2
29.3
23.6
20.4
51.2
57.8
26.7
34.0
62.3
51.4
0.6
1
2
3
4
5
6
7
8
9
0
CH
H
3
62
23.9
15.4
13.3
12.2
16.8
13.3
12.4
17.2
17.9
8.7
59
21.5
H
43
20.2
F
F
28
25.4
H
H
H
F
CF
CH
H
3
40
21.5
3
46
20.3
39
25.4
H
26
26.5
F
F
18
15.3
H
CH
3
33
19.6
11.8
7.3
KYP-2047
a
Isolated yields of the coupling reactions.
13.3
b
Transport, retention, and permeability parameters obtained in the PAMPA. Data represent mean values (n = 3).
2011). Here, with the aim of enhancing the drug-like properties of similar pose to that reported for reversible inhibitors ZPP and
our inhibitors, we have further explored the introduction of small KYP-2047 (Figure 3A) (Kaszuba et al., 2012). As a differential
substituents (-F, -F
ring (Figure 1C).
2
, -CH
3
, or -CF
3
) at position 4 of the proline trait, the larger fluorinated BdFB moiety is able to occupy most
of the S3 cavity, resulting in extended p-p contacts with F173
Finally, at position P1, we have conveniently replaced the and W595 of POP. On the other hand, the sulfonyl group at
C-terminal carboxylate of the natural substrate with a mildly position P1, despite being larger than the natural amide, is well
reactive electrophile, such as a sulfonyl fluoride moiety. Upon fitted between Asn555 and His680, and provides an optimal
binding, this group is intended to form a stable covalent bond binding geometry toward Ser554. The flexible methylene group
with the catalytic Ser554 of the enzyme, thus blocking its activity. connecting the sulfur atom to the pyrrolidine ring also facilitates
In contrast to other more reactive warheads, sulfonyl fluorides the accommodation of the electrophile in front of the Ser554 side
have an adequate balance of biocompatibility, aqueous stability, chain, thus enabling the covalent binding to occur.
and protein reactivity (Narayanan and Jones, 2015), and have
found significant use as probes in chemical biology. Peptidyl sul- SEQUENTIAL SYNTHESIS YIELDS HIGHLY POTENT POP
fonyl fluorides, in particular, have been developed as chymo- INHIBITORS
trypsin (Brouwer et al., 2011) and proteasome (Dubiella et al.,
2
014) inhibitors, among other applications, showing therapeutic The synthesis of the proline warhead was accomplished by start-
efficacy and safety when administered in mice (Tschan et al., ing from the natural precursor L-prolinol (Figure S1A). The
013). All in all, with the triple objective of achieving high po- hydroxyl group was oxidized in two steps to sodium sulfonate,
2
tency, selectivity, and CNS permeability, we sought to condense which was fluorinated to the corresponding sulfonyl fluoride.
this intricate structure-activity data into a peptidomimetic For this reaction, the use of XtalFluor-M (L’heureux et al., 2010)
scaffold that combines optimal P2 and P3 moieties with a unique provided a cleaner crude product and a higher yield (76%)
prolylsulfonyl fluoride warhead, leading to a total of 15 potential than other fluorinating agents such as DAST and XtalFluor-E
inhibitors (Table 1).
(13% and 69% yield, respectively). Once the warhead was
Special attention was devoted to those inhibitors bearing the obtained, a divergent strategy was followed to sequentially
BdFB P3 group, as this moiety is much bulkier than the gold couple the respective P2 and P3 groups (Figure S1B). A single
standards Cbz and phenylpropyl. However, covalent docking purification step, at the end of the synthesis, afforded the final
calculations revealed that these compounds can adopt a very compounds on a milligram scale.
Cell Chemical Biology 25, 1–7, July 19, 2018
3

Please cite this article in press as: Guardiola et al., Targeted Covalent Inhibition of Prolyl Oligopeptidase (POP): Discovery of Sulfonylfluoride Peptido-
mimetics, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.04.013
ring (R1, R2), meant to enhance the selectivity and membrane
permeability of the inhibitors, was well tolerated (inhibitors 11
and 18). Notably, the BdFB moiety used at P3 resulted in highly
potent compounds (17–20).
SULFONYL FLUORIDE PEPTIDOMIMETICS SHOW HIGH
BBB PERMEABILITY
In addition to potency, it is imperative that POP inhibitors
cross the BBB to enter the brain tissue, a criterion that peptido-
mimetic drugs often fail to meet (L o´ pez et al., 2013). This is not
surprising given that the BBB—a highly specialized and com-
plex vascular system—prevents most (>98%) drugs from
reaching the CNS (Pardridge, 2005). To address this issue,
we studied the transport of our compounds in the parallel arti-
ficial membrane permeability assay (PAMPA), in which a lipid
brain extract mimics the composition of the BBB, resulting in
a convenient method to evaluate the transport of compounds
by passive diffusion (Di et al., 2003). Reassuringly, high perme-
ability values were obtained for all three families of compounds
ꢀ ꢀ1
5
(P > 10 cm s , see Table 1), in the same range as propran-
e
olol, a highly permeable CNS-active drug used as positive
control. In particular, transport was increased upon fluorination
of the P2 proline ring; in contrast, the introduction of lipophilic
3 3
groups (CH and CF ) led to membrane retention and a slight
decrease in permeability. To further validate these results,
also taking the effect of active transport and efflux pumps
into account, we evaluated the absorptive transport through a
monolayer of MDCK cells, which is a more relevant model for
the prediction of BBB permeability (Wang et al., 2005). Again,
all inhibitors achieved excellent transport rates, in some cases
higher than metoprolol, a drug approved by the US Food and
Drug Administration (Figure 2A). Only the BdFB compounds
showed slightly lower values due to retention in lipids, as also
shown in the PAMPA.
HIGH-AFFINITY COVALENT BINDING IS NOT
COMPROMISED BY OFF-TARGET EFFECTS
Figure 2. Sulfonyl Fluoride Peptidomimetics Show High BBB Perme-
ability and Specific and Irreversible Binding to the POP Catalytic Site
(
A) Apparent permeability (Papp) of compounds 7–20 and metoprolol (control)
in MDCK cells. Error bars represent standard deviations of triplicate samples.
B) Kinetic constants that define a covalent binding process (top). The
In addition to POP, the S9 family of serine proteases holds other
proline-specific peptidases, such as fibroblast activation protein
(FAP) and dipeptidylpeptidase IV (DPPIV), which also hydrolyze
proline-containing peptide hormones and neuropeptides. While
DPPIV is exclusively an exopeptidase, FAP also shares equiva-
lent endopeptidase activity with POP. As well as being function-
ally related, the active sites of these three enzymes share a high
structural identity, with conserved Arg, Phe, and Tyr residues
surrounding the catalytic Ser (Juillerat-Jeanneret, 2008). Thus,
(
exponential decay of POP activity, at several concentrations of inhibitor 18, is
represented versus the preincubation time (left). The kobs plot (right) yields the
kinetic parameters that govern irreversible binding (K
i 2
, k ).
(
C) Mass spectra (right) and deconvoluted mass (left) of the CNBr-
digested catalytic peptide of POP, showing selective covalent modification of
Ser554 by 18.
For evaluation and quantification of enzyme inhibition, a fluo- the design of inhibitors with a good degree of selectivity
rometric assay was set up using recombinant human POP. To between these enzymes has proved difficult, and it is of utmost
allow for comparison, half maximal inhibitory concentration importance to avoid cross-reactivity and undesired side effects
(IC50) values were determined at regular time points during the (Poplawski et al., 2013).
linear phase of the inhibition process (Figure S2). Although To assess the off-target reactivity of our compounds, three
dependent on the preincubation time, IC50 values are a useful class-representative inhibitors (8, 13, and 18) were screened
metric to compare the potency of a given set of inhibitors, since against FAP and DPPIV at concentrations up to 50 mM
their full kinetic characterization (see below) is more laborious
ꢁ
( 50,000-fold the POP IC50). In this assay, a remarkable
and time consuming. As expected from the rational design, level of selectivity was found for all three families (Table S1).
all molecules had strong potencies in the low-nanomolar Even at the highest concentration of inhibitors (50 mM), catalytic
range (Table 1). Derivatization of position 4 of the P2 proline activities >70% were observed in all cases for FAP and DPPIV.
4
Cell Chemical Biology 25, 1–7, July 19, 2018

Please cite this article in press as: Guardiola et al., Targeted Covalent Inhibition of Prolyl Oligopeptidase (POP): Discovery of Sulfonylfluoride Peptido-
mimetics, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.04.013
In this experiment, an exponential decay in enzyme activity,
dependent on the preincubation time, was observed, thus con-
firming the irreversible nature of POP inhibition. In the context
of irreversible binding, inhibitor potency and selectivity can be
governed by the initial non-covalent binding to the enzyme (K
by the subsequent bonding step (k ), or by both of them (k /K
For compound 18, the k /K
resents inhibitory potency—was 2 3 10 M
potent and fast inactivation of the enzyme (Miyahisa
i
),
2
2
i
).
2
i
ratio—the parameter that best rep-
6
ꢀ1 ꢀ1
s , thus indicating
a
et al., 2015).
In addition, mass spectrometry analysis of a BrCN-digested
POP sample was performed to confirm the specific binding of
18 to the active site. In this assay, the mass of the peptide con-
taining the catalytic Ser554 was increased by 508 Da versus a
buffer-treated reference sample, which indicated single sulfony-
lation of the Ser side chain, consistent with a 1:1 inhibitor/protein
ratio (Figure 2C). Therefore, the combination of kinetic and mass
spectrometry POP binding assays reinforced our initial assump-
tion of covalent and irreversible modification of the enzyme
through the prolylsulfonyl warhead.
LOW-NANOMOLAR POP INHIBITION IN INTACT
HUMAN CELLS
In order to assess the activity of our hit compound in a
biologically more relevant environment, human neuroblastoma
SH-SY5Y cells were treated with inhibitor 18 at concentrations
ranging from 0.2 pM to 200 nM. Subsequently, POP activity
was measured on intact cells by recording the release of fluores-
cence from a specific substrate (Z-Gly-Pro-AMC). Under these
conditions, compound 18 was able to reach the cytosol and pro-
duce a fast—10-min incubation time—and strong inhibition of
the POP enzyme, showing an IC50 of 1.4 nM (Figure 3B).
Figure 3. Inhibitor 18 Covalently Binds to the POP Catalytic Site and
Displays Low-Nanomolar Activity in Human Cells
(A) Predicted binding of inhibitor 18 (structure shown bottom right) to the POP
catalytic pocket, highlighting the main cavities (S1, S2, and S3) and interacting
residues. Hydrogen bonds are shown with yellow dashes.
(B) POP activity in intact human SH-SY5Y cells at different concentrations of
18. Error bars represent standard deviations of triplicate samples.
Collectively, our findings show how the distinct strengths of
In particular, inhibitor 18 had a marginal inhibitory effect (below reversible and irreversible mechanisms can be combined to
0%) on the closely related FAP protease. At lower concentra- yield drug candidates with carefully tuned reactivity and
2
tions—in the POP IC50 range—these compounds were able to specific target engagement. As pursued with the structure-
fully discriminate between the different proline-specific pepti- based design, the optimal selection of P2-P3 groups, com-
dases. This high specificity fulfills a main requirement for the bined with a novel prolylsulfonyl fluoride warhead, yielded
selective and covalent modulation of the POP enzyme.
POP inhibitors that were not only highly potent and selective,
but also CNS permeable. By analogy with other covalent inhib-
itors such as ZPP, the binding of these compounds to the POP
active site may produce a prolonged modulation of the highly
dynamic protein conformational equilibrium, effectively stabiliz-
BINDING IS FAST, IRREVERSIBLE, AND SPECIFIC FOR
THE CATALYTIC SER OF POP
Considering the previous results, we selected inhibitor 18—hav- ing the closed or inactive state of the enzyme (L o´ pez et al.,
ing a suitable pharmacological profile and a more innovative 2016). If this was the case, our inhibitors would ultimately
scaffold—for further characterization. First, we studied the inhi- abolish the deleterious interactions of POP with pathogenic
bition mechanism of POP by the sulfonyl fluoride warhead. proteins such as a-synuclein, thus opening the door for new
From a kinetic perspective, covalent drugs block their targets treatments of CNS pathologies.
in a two-step fashion (Singh et al., 2011). Initially, the inhibitor as-
sociates via non-covalent interactions to the target enzyme—a SIGNIFICANCE
process defined by K
by the protein takes place, giving rise to the inhibited com- POP is a proteolytic enzyme and partner in intracellular
plex—defined by k (Figure 2B). In those cases in which the inhi- PPIs that is actively involved in the pathogenesis of
i
—and afterward the nucleophilic attack
2
bition is effectively irreversible, kꢀ2 will essentially be zero, and, if several cognitive disorders ranging from Parkinson dis-
allowed sufficient time, the binding reaction will proceed to ease to schizophrenia. Thus, designing efficient inhibitors
completion rather to equilibrium. In order to confirm that the able to reach the brain and discriminate POP from
inhibitory mechanism was irreversible, compound 18 was sub- other proteases has been one of the main objectives of
jected to complete kinetic profiling (Figure 2B).
researchers in the field. However, most reported drugs
Cell Chemical Biology 25, 1–7, July 19, 2018
5

Please cite this article in press as: Guardiola et al., Targeted Covalent Inhibition of Prolyl Oligopeptidase (POP): Discovery of Sulfonylfluoride Peptido-
mimetics, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.04.013
are reversible inhibitors that have found limited in vivo
application, mostly due to low membrane permeability,
lack of selectivity, and short residence times. Here, we
report the design of shape-complementary targeted
inhibitors that exert sustained POP inhibition at very low
concentrations. In our bifunctional molecules, the peptide
backbone and side chains form a template that selectively
recognizes the active site of the enzyme—thus providing
selectivity—while the sulfonyl fluoride electrophile irre-
versibly binds to the catalytic Ser. In addition to their
low-nanomolar potency, these compounds were >1,0003
selective for POP over two closely related proteases
AUTHOR CONTRIBUTIONS
S.G. wrote the manuscript, designed the research, performed the research, and
analyzed the data; R.P. wrote the manuscript, designed the research, and
analyzed the data; L.M. provided useful support for the in vitro studies; A.J.B.
and J.S. performed the chemical synthesis and analyzed the data; L.N., T.T.,
R.M.J.L., and E.G. designed the research and supervised the experiments.
All authors commented on the manuscript.
DECLARATION OF INTERESTS
E.G., R.P., and T.T. have filed a patent (EP2917209B1). The other authors
declare no competing interests.
Received: February 1, 2018
Revised: March 14, 2018
Accepted: April 12, 2018
Published: May 17, 2018
(
DPPIV and FAP), thus ensuring low cross-reactivity.
Notably, high permeability values were obtained in both
PAMPA and MDCK assays, a property that translates
into potent cellular activity, as was shown in SH-SY5Y
human cells. Beyond their application in the study of the
function of POP in CNS disorders, these compounds
pave the way for the design of selective irreversible inhib-
itors of other serine proteases.
REFERENCES
Baillie, T.A. (2016). Targeted covalent inhibitors for drug design. Angew Chem.
Int. Ed. 55, 13408–13421.
Brandt, I., G e´ rard, M., Sergeant, K., Devreese, B., Baekelandt, V., Augustyns,
K., Scharp e´ , S., Engelborghs, Y., and Lambeir, A.-M. (2008). Prolyl oligopep-
tidase stimulates the aggregation of alpha-synuclein. Peptides 29, 1472–1478.
STAR+METHODS
Brouwer, A.J., Ceylan, T., Jonker, A.M., van der Linden, T., and Liskamp,
R.M.J. (2011). Synthesis and biological evaluation of novel irreversible serine
protease inhibitors using amino acid based sulfonyl fluorides as an electro-
philic trap. Bioorg. Med. Chem. 19, 2397–2406.
Detailed methods are provided in the online version of this paper
and include the following:
d KEY RESOURCES TABLE
d CONTACT FOR REAGENT AND RESOURCE SHARING
d EXPERIMENTAL MODEL AND SUBJECT DETAILS
d METHOD DETAILS
Checler, F., Vincent, J.P., and Kitabgi, P. (1985). Inactivation of neurotensin
by rat brain synaptic membranes partly occurs through cleavage at the
Arg8 -Arg9 peptide bond by a metalloendopeptidase. J. Neurochem. 45,
1
509–1513.
Di Daniel, E., Glover, C.P., Grot, E., Chan, M.K., Sanderson, T.H., White, J.H.,
Ellis, C.L., Gallagher, K.T., Uney, J., Thomas, J., et al. (2009). Prolyl oligopep-
tidase binds to GAP-43 and functions without its peptidase activity. Mol. Cell.
Neurosci. 41, 373–382.
B General
B Analytical and Preparative HPLC
B TLC
B NMR
Di, L., Kerns, E.H., Fan, K., McConnell, O.J., and Carter, G.T. (2003). High
throughput artificial membrane permeability assay for blood-brain barrier.
Eur. J. Med. Chem. 38, 223–232.
B Chemical Syntheses
B Covalent Docking of the POP Inhibitors
B POP Activity Assays
Dubiella, C., Cui, H., Gersch, M., Brouwer, A.J., Sieber, S.A., Kruger, A.,
Liskamp, R.M.J., and Groll, M. (2014). Selective inhibition of the immunopro-
teasome by ligand-induced crosslinking of the active site. Angew Chem. Int.
Ed. 53, 11969–11973.
€
B Mass Spectrometry Assays
B Cell Activity Assays
B Selectivity Versus FAP
B Selectivity Versus DPPIV
F u€ l o¨ p, V., B o¨ cskei, Z., and Polg a´ r, L. (1998). Prolyl oligopeptidase: an unusual
beta-propeller domain regulates proteolysis. Cell 94, 161–170.
B Parallel
Assay (PAMPA)
B MDCK Permeability Assay
Artificial
Membrane
Permeability
Giralt, E., Prades, R., Tarrago, T., and Royo, S. (2014). 1-[1-(Benzoyl)-pyrroli-
dine-2-carbonyl]-pyrrolidine-2-carbonitrile derivatives. Patent EP2917209B1,
filed November 11, 2013, and published February 8, 2017.
d QUANTIFICATION AND STATISTICAL ANALYSIS
Hannula, M.J., My o¨ h a€ nen, T.T., Tenorio-Laranga, J., M a€ nnist o¨ , P.T., and
Garcia-Horsman, J.A. (2013). Prolyl oligopeptidase colocalizes with a-synuclein,
b-amyloid, tau protein and astroglia in the post-mortem brain samples with
Parkinson’s and Alzheimer’s diseases. Neuroscience 242, 140–150.
SUPPLEMENTAL INFORMATION
Supplemental Information includes two figures, one table, and one data file
can be found with this article online at https://doi.org/10.1016/j.chembiol.
H o¨ fling, C., Kulesskaya, N., Jaako, K., Peltonen, I., M a€ nnist o¨ , P.T., Nurmi, A.,
Vartiainen, N., Morawski, M., Zharkovsky, A., V o˜ ikar, V., et al. (2016).
Deficiency of prolyl oligopeptidase in mice disturbs synaptic plasticity and
reduces anxiety-like behaviour, body weight, and brain volume. Eur.
Neuropsychopharmacol. 26, 1048–1061.
2018.04.013.
ACKNOWLEDGMENTS
Huhn, A.J., Guerra, R.M., Harvey, E.P., Bird, G.H., and Walensky, L.D. (2016).
Selective covalent targeting of anti-apoptotic BFL-1 by cysteine-reactive
stapled peptide inhibitors. Cell Chem. Biol. 23, 1123–1134.
This study was funded by MINECO-FEDER (BIO 2016-75327-R) and
Generalitat de Catalunya (XRB and 2017SGR-998). We also thank the
NMR facility from Scientific and Technological Centre of the University of
Barcelona (CCiT UB), the Mass Spectrometry Core Facility (IRB Barcelona),
and the Barcelona Supercomputing Center (BSC) for their technical support.
IRB Barcelona is the recipient of a Severo Ochoa Award of Excellence from
MINECO (Government of Spain).
Juillerat-Jeanneret, L. (2008). Prolyl-specific peptidases and their inhibitors in
biological processes. Curr. Chem. Biol. 2, 97–109.
Kaszuba, K., R o´ g, T., Danne, R., Canning, P., F u€ l o¨ p, V., Juh a´ sz, T., Szeltner, Z.,
St. Pierre, J.-F., Garc ı´ a-Horsman, A., M a€ nnist o¨ , P.T., et al. (2012). Molecular
6
Cell Chemical Biology 25, 1–7, July 19, 2018

Please cite this article in press as: Guardiola et al., Targeted Covalent Inhibition of Prolyl Oligopeptidase (POP): Discovery of Sulfonylfluoride Peptido-
mimetics, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.04.013
dynamics, crystallography and mutagenesis studies on the substrate gating
Savolainen, M.H., Yan, X., My o¨ h a€ nen, T.T., and Huttunen, H.J. (2015). Prolyl
oligopeptidase enhances a-synuclein dimerization via direct protein-protein
interaction. J. Biol. Chem. 290, 5117–5126.
mechanism of prolyl oligopeptidase. Biochimie 94, 1398–1411.
L’heureux, A., Beaulieu, F., Bennett, C., Bill, D.R., Clayton, S., Laflamme, F.,
Mirmehrabi, M., Tadayon, S., Tovell, D., and Couturier, M. (2010).
Aminodifluorosulfinium salts: selective fluorination reagents with enhanced
thermal stability and ease of handling. J. Org. Chem. 75, 3401–3411.
Schechter, I., and Berger, A. (1967). On the size of the active site in proteases.
I. Papain. Biochem. Biophys. Res. Commun. 27, 157–162.
Singh, J., Petter, R.C., Baillie, T.A., and Whitty, A. (2011). The resurgence of
Li, J., Wilk, E., and Wilk, S. (1996). Inhibition of prolyl oligopeptidase by
covalent drugs. Nat. Rev. Drug Discov. 10, 307–317.
Fmoc-aminoacylpyrrolidine-2-nitriles. J. Neurochem. 66, 2105–2112.
Svarcbahs, R., Julku, U.H., and My o¨ h a€ nen, T.T. (2016). Inhibition of prolyl
oligopeptidase restores spontaneous motor behavior in the a-synuclein virus
vector–based Parkinson’s disease mouse model by decreasing a-synuclein
oligomeric species in mouse brain. J. Neurosci. 36, 12485–12497.
L o´ pez, A., Tarrag o´ , T., and Giralt, E. (2011). Low molecular weight inhibitors of
prolyl oligopeptidase: a review of compounds patented from 2003 to 2010.
Expert Opin. Ther. Pat. 21, 1023–1044.
L o´ pez, A., Mendieta, L., Prades, R., Royo, S., Tarrag o´ , T., and Giralt, E. (2013).
Peptide POP inhibitors for the treatment of the cognitive symptoms of
schizophrenia. Future Med. Chem. 5, 1509–1523.
Tarrago, T., Kichik, N., Segu ı´ , J., and Giralt, E. (2007). The natural product
berberine is a human prolyl oligopeptidase inhibitor. ChemMedChem 2,
3
54–359.
L o´ pez, A., Herranz-Trillo, F., Kotev, M., Gair ı´ , M., Guallar, V., Bernad o´ , P.,
Millet, O., Tarrag o´ , T., and Giralt, E. (2016). Active-site-directed inhibitors of
prolyl oligopeptidase abolish its conformational dynamics. ChemBioChem
Tarrag o´ , T., Frutos, S., Rodriguez-Mias, R.A., and Giralt, E. (2006).
Identification by 19F NMR of traditional Chinese medicinal plants possessing
prolyl oligopeptidase inhibitory activity. ChemBioChem 7, 827–833.
17, 913–917.
Toide, K., Iwamoto, Y., Fujiwara, T., and Abe, H. (1995). JTP-4819: a novel
Mariaule, G., De Cesco, S., Airaghi, F., Kurian, J., Schiavini, P., Rocheleau, S.,
Huski cꢀ , I., Auclair, K., Mittermaier, A., and Moitessier, N. (2016). 3-oxo-
prolyl endopeptidase inhibitor with potential as
a cognitive enhancer.
J. Pharmacol. Exp. Ther. 274, 1370–1378.
hexahydro-1H-isoindole-4-carboxylic acid as a drug chiral bicyclic scaffold:
structure-based design and preparation of conformationally constrained
covalent and noncovalent prolyl oligopeptidase inhibitors. J. Med. Chem.
Tschan, S., Brouwer, A.J., Werkhoven, P.R., Jonker, A.M., Wagner, L., Knittel,
S., Aminake, M.N., Pradel, G., Joanny, F., Liskamp, R.M.J., et al. (2013).
Broad-spectrum antimalarial activity of peptido sulfonyl fluorides, a new class
of proteasome inhibitors. Antimicrob. Agents Chemother. 57, 3576–3584.
59, 4221–4234.
Miyahisa, I., Sameshima, T., and Hixon, M.S. (2015). Rapid determination of
the specificity constant of irreversible inhibitors (kinact/KI) by means of an
endpoint competition assay. Angew Chem. Int. Ed. 54, 14099–14102.
Ven a€ l a€ inen, J.I., Garcia-Horsman, J.A., Forsberg, M.M., Jalkanen, A., Wall e´ n,
E.A.A., Jarho, E.M., Christiaans, J.A.M., Gynther, J., and M a€ nnist o¨ , P.T. (2006).
Binding kinetics and duration of in vivo action of novel prolyl oligopeptidase in-
hibitors. Biochem. Pharmacol. 71, 683–692.
Narayanan, A., and Jones, L.H. (2015). Sulfonyl fluorides as privileged
warheads in chemical biology. Chem. Sci. 6, 2650–2659.
Wang, Q., Rager, J.D., Weinstein, K., Kardos, P.S., Dobson, G.L., Li, J., and
Hidalgo, I.J. (2005). Evaluation of the MDR-MDCK cell line as a permeability
screen for the blood–brain barrier. Int. J. Pharm. 288, 349–359.
Pardridge, W.M. (2005). The blood-brain barrier: bottleneck in brain drug
development. NeuroRx 2, 3–14.
Poplawski, S.E., Lai, J.H., Li, Y., Jin, Z., Liu, Y., Wu, W., Wu, Y., Zhou, Y.,
Sudmeier, J.L., Sanford, D.G., et al. (2013). Identification of selective and
potent inhibitors of fibroblast activation protein and prolyl oligopeptidase.
J. Med. Chem. 56, 3467–3477.
Yoshimoto, T., Nishimura, T., Kita, T., and Tsuru, D. (1983). Post-proline
cleaving enzyme (prolyl endopeptidase) from bovine brain. J. Biochem. 94,
1179–1190.
Prades, R., Munarriz-Cuezva, E., Urig u€ en, L., Gil-Pisa, I., G o´ mez, L., Mendieta,
L., Royo, S., Giralt, E., Tarrag o´ , T., and Meana, J.J. (2017). The prolyl oligopep-
tidase inhibitor IPR19 ameliorates cognitive deficits in mouse models of
schizophrenia. Eur. Neuropsychopharmacol. 27, 180–191.
Yoshimoto, T., Kawahara, K., Matsubara, F., Kado, K., and Tsuru, D. (1985).
Comparison of inhibitory effects of prolinal-containing peptide derivatives on
prolyl endopeptidases from bovine brain and Flavobacterium. J. Biochem.
98, 975–979.
Cell Chemical Biology 25, 1–7, July 19, 2018
7

Please cite this article in press as: Guardiola et al., Targeted Covalent Inhibition of Prolyl Oligopeptidase (POP): Discovery of Sulfonylfluoride Peptido-
mimetics, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.04.013
STAR+METHODS
KEY RESOURCES TABLE
REAGENT or RESOURCE
Chemicals, Peptides, and Recombinant Proteins
Cbz-L-prolinol
SOURCE
IDENTIFIER
Sigma
Cat#512966
Cat#471259
Cat#719447
Cat#B3021
Cat#P21005
Cat#852017
Cat#11435
Methanesulfonyl chloride
XtalFluor-Mꢀ
Sigma
Sigma
Benzyl chloroformate
TCI Chemicals
Sigma
4-phenylbutanoic acid
Fmoc-L-Pro-OH
Merck
(
(
(
(
2S,4R)-4-Fluoro-pyrrolidine-2-carboxylic acid
2S)-Fmoc-4,4-difluoro-pyrrolidine-2-carboxylic acid
4S)-4-Methyl-L-proline
ChemImpex
ChemImpex
Fluorochem
Enamine
Cat#26718
Cat#303270
Cat#300-60590
2S,4S)-1-[(tert-butoxy)carbonyl]-4-(trifluoromethyl)
pyrrolidine-2-carboxylic acid
S)-pyrrolidin-2-ylmethanesulfonyl fluoride
(
This study
Bachem
Bachem
Sigma
N/A
N-benzyloxycarbonyl-Gly-Pro-methylcoumarinyl-7-amide
H-Gly-Pro-methylcoumarinyl-7-amide
Fibroblast Activation Protein a human
Critical Commercial Assays
CACO and MDCK permeability assay
PAMPA buffer
Cat#I-1145
Cat#I-1225
Cat#F7182
ReadyCell
pION
CACO-Ready
PAMPA Evolution
Cat#141101P
PAMPA brain polar lipid extract
Experimental Models: Cell Lines
SH-SY5Y cells
Avanti
ATCC
Cat#CRL-2266
Software and Algorithms
Xcalibur 2.0
Thermo Scientific
http://www.thermofisher.com/order/catalog/
product/OPTON-30487
GraphPad 6
CovDock
Prism
https://www.graphpad.com/scientific-software/prism/
https://www.schrodinger.com/covdock
Schrodinger LLC
Waters
MassLynx
http://www.waters.com/waters/en_US/
MassLynx-Mass-Spectrometry-Software-/
nav.htm?cid=513164&locale=en_US
CONTACT FOR REAGENT AND RESOURCE SHARING
Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Ernest
Giralt (ernest.giralt@irbbarcelona.org)
EXPERIMENTAL MODEL AND SUBJECT DETAILS
Human SH-SY5Y cells were obtained from the American Type Culture Collection (ATCC). Cells were cultured in DMEM-high-glucose
medium containing 10% fetal bovine serum (FBS), 1% glutamine and 1% penicillin-streptomycin and grown in a humidified incubator
ꢂ
at 37 C, 5% CO2.
METHOD DETAILS
General
Unless otherwise stated, all solvents and chemicals were used as received. All solvent mixtures (eluents) are given in v/v.
e1 Cell Chemical Biology 25, 1–7.e1–e4, July 19, 2018

Please cite this article in press as: Guardiola et al., Targeted Covalent Inhibition of Prolyl Oligopeptidase (POP): Discovery of Sulfonylfluoride Peptido-
mimetics, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.04.013
Analytical and Preparative HPLC
All reactions were monitored by HPLC (Waters Alliance 2695 equipped with 2487 photodiode array detector, Sunfire C18 column (2.1 x
˚
1
000 mm, 3.5 mm, 100 A, Waters); flow rate = 1 mL/min; solvents A=0.036% trifluoroacetic acid in water, and B=0.045% trifluoroacetic
acid in acetonitrile. The reaction crude products were purified by semi-preparative HPLC on a Waters 2700 sample manager equipped
with a Waters 2487 dual wavelength absorbance detector, a Waters 600 controller, a Waters fraction collector; using a symmetry C18
column (100 mm x 19 mm, 5 mm; Waters). Column flow was set at 15 ml/min; solvents A=0.1% trifluoroacetic acid in water, and
B=0.1% trifluoroacetic acid in acetonitrile. The purity of all final compounds was 95% or higher, as measured by HPLC.
TLC
Reactions for compounds 5-10 were monitored by TLC analysis using Merck pre-coated silica gel 60 F-254 (0.25 mm) plates.
Spots were visualized with UV light, ninhydrin, Cl2-TDM or sulfuric acid. Solvents were evaporated under reduced pressure
ꢂ
at 40 C. Column chromatography was performed on Siliaflash P60 (40-63 mm) from Silicycle (Canada).
NMR
NMR spectra were recorded on Varian Mercury 400, Agilent 400, or Varian Inova 500 MHz spectrometers, using CDCl
3
as solvent at
ꢂ
1
5 C. Variable temperature H NMR spectra (400 MHz) were recorded on a Varian S400 spectrometer. Chemical shifts are given in
2
1
13
parts per million (ppm) (d relative to residual solvent peak for H and C, or relative to TMS (0.00 ppm) for compounds 5-10). Some of
1
3
1
13
the C NMR spectra were recorded using the attached proton test (apt) pulse sequence. Most of the H and C NMR spectra
contained additional peaks, due to the presence of rotamers and fluorine atoms.
Chemical Syntheses
Synthesis of the Prolylsulfonyl Fluoride Electrophile (6)
Cbz-Pro-j[CH
and the resulting solution was cooled in an icebath for 1 h under N
added dropwise and the mixture was stirred overnight at r.t. Additional CH
KHSO (1.0 M, 3 x 1.0 L), water (1.0 L) and brine. The organic layer was dried with Na
2
S]-Ac (2). To a solution of Cbz-Prolinol (1, 51.8 g, 220 mmol) in CH
atmosphere. Methanesulfonyl chloride (20.4 ml, 264 mmol) was
Cl (1.0 L) was added and the mixture was washed with
SO and concentrated in vacuo, affording the
2 2 3
Cl (1.0 L) was added Et N (36.8 ml, 262 mmol)
2
2
2
4
2
4
mesylate as a yellow oil (69.6 g, 220 mmol, quantitative). The mesylate was not stored, but used directly in the reaction with in situ
generated cesium thioacetate. To this end, thioacetic acid (23.4 mL, 332 mmol, 1.5 eq.) was added to a mixture of cesium carbonate
(
53.3 g, 164 mmol) in DMF (1.1 L) under N
The flask was covered with aluminium foil and the mixture was stirred over 3 nights at r.t. After completion, EtOAc (2.5 L) and water
1.0 L) were added and the water layer was extracted with EtOAc (0.5 L). The combined organic layer was washed with saturated
NaHCO solution (1.0 L, 2x), KHSO (1.0 M, 1.0 L, 2x) and brine (300 ml). After drying (Na SO ) and concentration in vacuo, the crude
product was purified by column chromatography (EtOAc: Hex, 1:9), affording 3 as a yellowish solid (43.9 g, 150 mmol, 68%).
Cbz-Pro-j[CH SO ]-ONa (3). To a solution of thioacetate 2 (43.9 g, 150 mmol) in acetic acid (450 mL) was added an aqueous
hydrogen peroxide solution (150 mL, 30% w/w) and the mixture was stirred overnight at r.t. The color of the reaction changed over-
night from clear brown to clear yellow. NaOAc.3H O (20.4 g, 150 mmol) was added and stirring was continued for 15 min. DMF
500 mL) was added and the solution was concentrated in vacuo to approximately half of its volume. Another portion of DMF
500 mL) was added and evaporated again to about one third of the volume. The addition and evaporation of DMF was repeated until
2
atmosphere. After stirring for 5 min, the resulting suspension was decanted to the mesylate.
(
3
4
2
4
2
2
2
(
(
no more peroxide was detected in the receiver flask with KI/starch paper. The mixture was then concentrated completely, and water
1.0 L) and CH Cl (500 mL) were added, after which the aqueous layer was washed once with CH Cl (500 mL). Lyophilization of the
water layer afforded 3 as a white solid (47.6 g, 145 mmol, 97%).
Cbz-Pro-j[CH SO ]-F (5). To a suspension of Cbz-Pro-c[CH SO
XtalFluor-M (7, 13.2 g, 54.3 mmol), under N atmosphere. After addition of Et
7 h with the condenser fitted with a drying tube. Unreacted XtalFluor-M was quenched by addition of silica gel (approximately 20 g)
and stirring for 5 min. After filtration, the mixture was concentrated in vacuo and re-dissolved in EtOAc (400 mL) and water (100 mL). After
separation,theorganiclayerwasdirectlydried(Na SO ),concentratedtodryness,andloaded(dissolvedinCH Cl )onasilicagelcolumn.
After elution (eluent: CH Cl ) using slight pressure (balloon), sulfonyl fluoride 5 was obtained as a white solid (6.89 g, 22.9 mmol, 76%).
HCl.H-Pro-j[CH SO ]-F (6). To a solution of sulfonyl fluoride 5 (6.89 g, 22.9 mmol) in dichloromethane (230 mL) was added a
solution of HBr in acetic acid (33%, 138 mL). After stirring at r.t. for 30 min, the solvents were removed in vacuo. The residue was
dissolved in H O (230 mL) and washed with EtOAc (2 x 200 mL). Dowex 2x8 (13.8 g, Cl-form) was then added. After stirring for
min at r.t., the resin was filtered off. Lyophilization of the filtrate afforded HCl-salt 6 as an off-white solid (4.2 g, 20.6 mmol, 90%).
(
2
2
2
2
2
2
2
2 2 2
]-ONa (3, 9.64 g, 30 mmol) in CH Cl (600 mL) was added was
2
3
N.3HF (212 mL, 1.30 mmol), the mixture was refluxed for
1
2
4
2
2
2
2
2
2
2
5
Synthesis of the Final POP Inhibitors
General Procedure for the Synthesis of Intermediates 23-27. Benzylchloroformate (1.1 eq., 0.13 mmol) was added dropwise to a
stirred solution of the corresponding proline analog (1 eq., 0.12 mmol) and NaHCO
ꢂ
3
(2.5 eq., 0.3 mmol) in water/THF (1:1, 1 mL)
at 0 C. The reaction was stirred for 30 min at 0 C and for 5 h at r.t. The solution was acidified with 1M HCl aqueous solution and
ꢂ
extracted with AcOEt (3 x 10 mL). The combined organic layer was washed with brine (20 mL), dried over MgSO
orated. Crude products were used in the next step without further purification.
4
, filtered, and evap-
General Procedure for the Synthesis of Intermediates 28-36. 4-(benzyloxy)-3,5-difluorobenzoic acid was obtained as previously
described (Giralt et al., 2014). 4-phenylbutanoic acid or 4-(benzyloxy)-3,5-difluorobenzoic acid (1 eq., 0.12 mmol) were stirred at
Cell Chemical Biology 25, 1–7.e1–e4, July 19, 2018 e2

Please cite this article in press as: Guardiola et al., Targeted Covalent Inhibition of Prolyl Oligopeptidase (POP): Discovery of Sulfonylfluoride Peptido-
mimetics, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.04.013
ꢂ
r.t. in anhydrous toluene (1 mL) for 10 min. Oxalyl chloride (1.5 eq., 0.18 mmol) was added, and the reaction was stirred at 50 C for
1
and the corresponding proline (1 eq., 0.12 mmol) was added neat at 0 C. The reaction was stirred at 0 C for 1 h and at r.t. for 3 h. The
.5 hr. The solvent was removed in vacuo and the resulting crude product was redissolved in anhydrous THF. DIPEA (2 eq., 41.8 mL)
ꢂ
ꢂ
mixture was concentrated in vacuo, redissolved in water (3 mL), and acidified with 1M HCl aqueous solution. The aqueous solution
was extracted with AcOEt (3 x 10 mL); the combined organic layer was washed with brine (20 mL), dried over MgSO
evaporated. Crude products were used in the next step without further purification.
4
, filtered, and
General Procedure for L-prolylmethanesulfonyl Fluoride Coupling. HATU (1.1 eq., 0.13 mmol) and DIPEA (2 eq., 0.24 mmol) were
added to a solution of the corresponding intermediate 23-36 (1 eq., 0.12 mmol) in DCM (1 mL). After 5 min, neat (S)-pyrrolidin-2-ylme-
thanesulfonyl fluoride (HCl salt 6 or TFA salt 22) was added, and the mixture was stirred for 3 h at r.t. The reaction mixture was diluted
3 4
with DCM (10 mL), and 5% KHCO aqueous solution was added (10 mL). The organic layer was then dried over MgSO , filtered, and
evaporated in vacuo. The crude product was redissolved in 2:8 water/acetonitrile and purified by semi-preparative HPLC.
Covalent Docking of the POP Inhibitors
In order to assess the potential binding of our compounds to POP and to study the key interactions within the catalytic site, we applied
the docking software CovDock to predict the position of covalently bound 18 in the POP active site. As no crystal structure of human
POP has been elucidated to date, porcine POP, which shares 97% identity with the human form, was used for the docking experiment
(
PDB code: 2XDW). After docking, a short minimization of the complex was performed, and the top 20 results were ranked. In its lowest
energy conformation, 18 adopted a very similar conformation to that reported for ligands ZPP and KYP-2047 co-crystalized with POP.
POP Activity Assays
POP was obtained by expression in E. coli and affinity purification as reported previously (Tarrag o´ et al., 2006). POP activity was
determined following the method described by Toide and coworkers (Toide et al., 1995). The reactions were performed in 96-well
microplates, thereby allowing the simultaneous monitoring of multiple reactions. For each reaction, the activity buffer (137 mL,
ꢂ
1
00 mM of Na/K phosphate buffer, pH 8.0) was pre-incubated for 15 min at 37 C with POP (5 nM) and with the corresponding inhibitor
solution (3 mL). Stock solutions at a range of inhibitor concentrations (typically 1 nM to 5 mM) were prepared in DMSO. After pre-
incubation, ZGP-AMC (N-benzyloxycarbonyl-Gly-Pro-methylcoumarinyl-7-amide, 10 mL, 450 mM in 40% of 1,4-dioxane) was added,
ꢂ
and the reaction was monitored at the fluorimeter at 37 C . The formation of AMC was measured fluorimetrically (excitation and
emission wavelengths were 360/40 and 485/20 nm, respectively). For each sample, a negative control containing 3 mL of the sample
in DMSO, 10 mL of ZGP-AMC solution and 137 mL of activity buffer were added into the well plate. The positive control of the reaction
consisted of 3 mL of DMSO, 10 mL of ZGP-AMC solution and 137 mL of 5 nM POP in activity buffer.
For the kinetic determination of compound 18, the same methodology was applied using a range of inhibitor concentrations (1 to
00 nM) and several preincubation times (0 to 60 min). Then, fluorogenic substrate ZGP-AMC was added and fluorescence was read
1
after 30 minutes. The fitting of the data and calculation of the kinetic parameters was performed as shown in Figure 2.
Mass Spectrometry Assays
Exact mass determination was performed on a LTQ-FT Ultra instrument (Thermo Scientific) by direct infusion of sample (1 mM
concentration in 1:1 water/acetonitrile, 0.1% formic acid) through a nanoESI chip; detection method scan (150-2000 a.m.u. range).
For the detection of inhibitor 18 bound to the catalytic site, a POP sample (130 mL, 20 mM in 100 mM of Na/K phosphate buffer, pH
ꢂ
8
4
.0) was incubated with 3 equivalents of 18 at 25 C during 1 h. The sample was lyophilized, resuspended in 100 mL of water containing
ꢂ
0% formic acid and 0.25 M cyanogen bromide and digested in darkness for 16 h at 25 C. Samples were desalted on a PolyLC C4
column and eluted with 95% acetonitrile in 1% formic acid. The samples were dried in a vacuum centrifuge and reconstituted in 50 mL
of water/acetonitrile (1:1) containing 1% formic acid, and were directly infused on an LTQ-FT Ultra instrument (Thermo Scientific).
Cell Activity Assays
Human SH-SY5Ycells(on passage15)wereseeded(5000cellsperwell) inafluorescence96-well platesuitableforcellculture. After72h
ꢂ
ofincubationat37 C, cells reached 90-100%confluence. Cells werewashedwithPBS and treatedwithdifferent concentrationsofcom-
pound 18 (0.2 to 200 nM) in 140 mL of PBS. After 10 min of pre-incubation time, ZGP-AMC (10 mL, 150 mM in 40% of 1,4-dioxane) was
ꢂ
added, and the reaction was monitored at the fluorimeter at 37 C. The formation of AMC was measured fluorimetrically (excitation and
emission wavelengths were 360/40 and 485/20 nm, respectively). Triplicates for each concentration, as well as blanks, were included.
Fluorescence values during the linear time of the inhibition (typically, 30 min) were determined and the IC50 values were then calculated.
Selectivity Versus FAP
To determine the activity of FAP in the presence of the inhibitors, reactions were performed in 96-well microplates. For each reaction,
ꢂ
the activity buffer (129.5 mL, 100 mM of Na/K phosphate buffer, pH 8.0) was pre-incubated for 15 min at 37 C with recombinant human
FAP (7.5 mL, 0.57 nM) and with the corresponding inhibitor solution (3 mL). Stock solutions at a range of inhibitor concentrations were
prepared in DMSO. After pre-incubation, ZGP-AMC (10 mL, 1.5 mM in 40% of 1,4-dioxane) was added, and the reaction was incubated
ꢂ
for 60 min at 37 C in a fluorimeter. The formation of AMC was monitored fluorimetrically (excitation and emission wavelengths were
360/40 and 485/20 nm, respectively). For each sample, a negative control containing 3 mL of the sample in DMSO, 10 mL of ZGP-AMC
solution, and 129.5 mL of activity buffer were added to the well plate. The positive control of the reaction consisted of 3 mL of DMSO,
e3 Cell Chemical Biology 25, 1–7.e1–e4, July 19, 2018

Please cite this article in press as: Guardiola et al., Targeted Covalent Inhibition of Prolyl Oligopeptidase (POP): Discovery of Sulfonylfluoride Peptido-
mimetics, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.04.013
10 mL of ZGP-AMC solution, and 137 mL of 0.57 nM FAP in activity buffer. Triplicates for each condition were included. Fluorescence
values versus time were represented and the lineal time was determined. Afterwards, the percentage of FAP inhibition was calculated
at point concentrations of inhibitors (50, 10, 5 and 1 mM) by comparing the fluorescence intensity of the sample with that of the blank.
Selectivity Versus DPPIV
The ectodomain (residues 39-766) of dipeptidyl peptidase-IV (DPPIV) was obtained by expression in Sf9 insect cells using the
Baculovirus technique. DPPIV activity was determined following the method previously described (Checler et al., 1985). The reactions
were performed in 96-well microplates. For each reaction, the activity buffer (132 mL, 50 mM Tris, 1 M NaCl, 1 mg/mL BSA, pH 7.5) was
ꢂ
pre-incubated for 15 min at 37 C with recombinant human DPPIV (5 mL, 12.3 mg/mL) and with the corresponding inhibitor solution
(3 mL). Stock solutions at a range of inhibitor concentrations were prepared in DMSO. After pre-incubation, GP-AMC (H-Gly-Pro-
methylcoumarinyl-7-amide, 10 mL, 0.75 mM in 40% of 1,4-dioxane) was added, and the reaction was incubated for 20 min at
ꢂ
3
4
7 C. The formation of AMC was monitored fluorimetrically every 5 min (excitation and emission wavelengths were 360/40 and
85/20 nm, respectively). For each sample, a negative control containing 3 mL of the sample in DMSO, 10 mL of GP-AMC solution
and 132 mL of activity buffer were added to the well plate. The positive control of the reaction consisted of 3 mL of DMSO, 10 mL of
GP-AMC solution and 132 mL of 12.3 mg/mL DPPIV in activity buffer. Triplicates for each condition were included. Fluorescence values
versus time were represented and the lineal time was determined. Afterwards, the percentage of DPPIV inhibition was calculated at
point concentrations of inhibitors (50, 10, 5 and 1 mM) by comparing the fluorescence intensity of the sample with that of the blank.
Parallel Artificial Membrane Permeability Assay (PAMPA)
The PAMPA was used to determine the capacity of the compounds to cross the BBB by passive diffusion. The effective permeability
of the compounds was measured at an initial concentration of 200 mM. The PAMPA buffer solution was prepared from a concentrated
one and the pH was adjusted to 6.8. 1-propanol (20%) was added as a cosolvent to the buffer solution. Stock solutions of all com-
pounds at 20 mM concentration were prepared in DMSO, and 1:100 dilution with PAMPA buffer was performed in each case. For the
assays, the PAMPA sandwich was disenabled, and the donor wells were filled with 200 mL of the compound solution of interest.
Before this, the polycarbonate filter of the membrane was coated with 4 mL of the commercial mixture of phospholipids. The acceptor
compartments were filled with PAMPA buffer containing 1% DMSO (200 mL). After this, both plates were reassembled. The PAMPA
sandwich was incubated in humidity-saturated chamber for 4 h with orbital agitation at 100 rpm. After this time, the donor and
acceptor wells were analyzed and quantified by HPLC at l=220 nm. Propranolol was used as a positive control. Experiments
were performed in triplicate. The phospholipid mixture used was a porcine polar brain lipid extract. Composition: phosphatidylcho-
line (PC) 12.6%, phosphatidylethanolamine (PE) 33.1%, phosphatidylserine (PS) 18.5%, phosphatidylinositol (PI) 4.1%, phosphatidic
acid 0.8%, and 30.9% of other compounds. After 4 h, the effective permeability (P
e
), the standard parameter that quantifies transport
independently of time and concentrations, was calculated:
ꢀ
ꢁ
ꢀ
218:3
2C
A
ðtÞ
ꢀ
6
P
e
=
log 1 ꢀ C
x 10
t
D
ðt Þ
o
where t is the running time (4 h), C
compound in the donor well at t=0 h. Transport (%) values were obtained by dividing the amount in the acceptor well at time t, C
by the amount in the donor well at time zero, C (t ), multiplied by 100. Permeability was considered excellent when values were
A
(t) is the concentration of compound in the acceptor well at time t and C
D
(t
0
) is the concentration of
A
(t),
D
0
–
6
–6
–6
–6
>
4.0 3 10 cm/s, uncertain between 2.0 3 10 and 4.0 3 10 cm/s, and poor below 2.0 3 10 cm/s (Di et al., 2003).
MDCK Permeability Assay
A monolayer of MDCK cells, already prepared in a 96-well plate, was purchased from ReadyCell, S.L. The Trans-Epithelial Electrical
Resistance (TEER) was measured prior to the assay in order to verify the integrity of the monolayer. Stock solutions of all compounds
at 4 mM concentration were prepared in DMSO, and 1:100 dilutions with sterile HBSS buffer (supplemented with 5 mM glucose, pH 7)
were performed in each case. Apical (donor) wells were filled with 100 mL of the compound solution (40 mM). Basal (acceptor) compart-
ments were filled with PAMPA buffer containing 1% DMSO (250 mL). Then, both plates were reassembled and the multi-well plate was
ꢂ
incubated for 2h at 37 C. After this time, the donor and acceptor wells were disassembled and each analyzed by UPLC-MS. Metoprolol
was used as a positive control. A Lucifer Yellow paracellular permeability assay was performed after the assay for quality control. All
experiments were performed in triplicate. The apparent permeability for each compound was calculated using the following equation:
dQ=dt
P
app
=
o
C x A
QUANTIFICATION AND STATISTICAL ANALYSIS
In vitro experiments (Figures 2, 3, and S2) were performed in triplicate. For the enzyme activity and cellular assays, fluorescence
values versus time were plotted and the lineal time was identified. Afterwards, the percentage of inhibition at each concentration
was calculated and the results were fitted using a four parameter log[inhibitor] versus response and reported directly from the
GraphPad Prism output.
Cell Chemical Biology 25, 1–7.e1–e4, July 19, 2018 e4