Full Paper
later confirmed by co-crystal structures analyses, as shown
below.
protein–ligand complexes, both active sites were found popu-
lated by a ligand (Supporting Information, Section 6.1.2, Fig-
ure S20).
The two enantiomers of the benzyl sulfonamide (ꢂ)-17 were
separated by chiral-phase HPLC to afford pure (+)-17 and
(ꢁ)-17. As already observed with other pyrazolopyran-based li-
gands,[25,26] there is a high level of chiral recognition at the
active site of PfSHMT and large discrepancies in biological ac-
tivity were measured for (+)-17 and (ꢁ)-17. Ligand (+)-17 in-
hibited PfSHMT (IC50 =150 nm) much more efficiently than
(ꢁ)-17 (only 38% inhibition at 250 mm) (Supporting Informa-
tion, Section S2, Table S1). The preference for the (+)-enantio-
mer is not surprising when considering the binding mode of
pyrazolopyran-based ligands. Indeed, for all analogues co-crys-
tallized with PvSHMT in this program, exclusively the (+)-enan-
tiomers were found to be bound to the enzyme. Cell-based ef-
ficacy confirmed this trend, as (+)-17 was again much more
potent than (ꢁ)-17 (EC50 =56 and 1584 nm, respectively)
(Table 2).
Binding mode of (+)-11, (+)-17, and (+)-20
Gratifyingly, the three co-crystal structures provided an explan-
ation for the lacking additional gain in affinity by the lipophilic
residues at the termini of the aryl sulfonamides and aryl sul-
fones.
The binding mode of the pyrazolopyran core was found to
be the same as in all other co-crystals with wild-type PvSHMT
solved so far, with a strong hydrogen-bonding network an-
choring this scaffold into the pteridine binding pocket (Fig-
ure 7a and Supporting Information, Section S6.1.2, Figures
S21a and S22a). The sulfonamide moieties in (+)-11 and
(+)-17 and the sulfone in (+)-20 are nicely found in their
favored conformation with Csp2-Csp2-S-N (or Csp2-Csp2-S-Csp3) tor-
sion angles of 748, 868, and 988, respectively. However, the ori-
entation of the sulfonamide moiety in (+)-11 (Supporting In-
formation, Section S6.1.2, Figure S21b) and (+)-17 (Figure 7b),
and the sulfone moiety in (+)-20 (Supporting Information, Sec-
tion S6.1.2, Figure S22b) was found to be opposite to that pre-
dicted by modeling (Supporting Information, Section S5, Fig-
ures S13 and S14). The SO2 moiety is pointing towards the hy-
drophobic residues at the exit of the pABA channel, while the
hydrophobic terminal substituents of the ligands are all point-
ing towards solvent instead of interacting with the protein.
Indeed, the sulfonamide moiety of (+)-17 is at close distances
to Val141 (d(O···HꢁCVal141)=2.8 and 3.3 ꢁ) and Pro267 (d(O···Hꢁ
Three ligands containing an aryl sulfone moiety ((ꢂ)-18–20)
and two reverse sulfonamides ((ꢂ)-21 and (ꢂ)-22) were also
prepared, yielding target- and cell-based activities in a similar
range to the aryl sulfonamide ligands (Table 2 and Supporting
Information, Section S2, Table S1). Crystals of the reverse sulfo-
namide (ꢂ)-21 also featured a staggered conformation with a
Csp3-N-S-Csp3 torsion angle of ꢁ71.88 (Supporting Information,
Section 6.2.5, Figure S32). This moiety is twisted almost orthog-
onally with respect to the phenyl ring with a Csp2-Csp2-N-S tor-
sion angle of 77.68 (Supporting Information, Section 6.2.5, Fig-
ure S31).
Although several ligands of both biphenyl and aryl sulfona-
mide/aryl sulfone series proved to be highly potent in our in
vitro assays, they were not studied further due to their limited
metabolic stability in human liver microsomes (t1/2 <10 min)
(Supporting Information, Section S2, Table S2). The terminal
fragments on the phenyl ring departing from the core are pre-
sumably responsible for this intrinsic instability and not the
pyrazolopyran core. Indeed, we recently reported a series of
pyrazolopyran-based ligands with half-lives up to 4 h.[26]
CPro267)=3.2 ꢁ), establishing several van der Waals interactions
with apolar atoms and forming weak hydrogen-bond-type
contacts to aliphatic CꢁH moieties (Figure 7b).[41] Very similar
interactions were found for (+)-11 and (+)-20 with contacts
ranging from 2.8 to 3.4 ꢁ (Supporting Information, Sec-
tion S6.1.2, Figures S21b and S22b). Those three C364A-
PvSHMT–ligand complexes highlight the low hydrophilicity of
the SO2 group, which resembles the low hydrophilicity of the
nitro group[66] that also prefers pointing into hydrophobic
pockets rather than into solvent. We performed a search in
both CSD and PDB for hydrogen bonding from strong hydro-
gen-bond donors (OꢁH and NꢁH, but excluding CꢁH) to the
SO2 group in aryl sulfonamides and aryl sulfones (Supporting
Information, Section S3, Table S3). Of 7856 hits in the CSD,
6617 sulfonamides formed no hydrogen bond and 1239 one
hydrogen bond. A higher frequency of single hydrogen bonds
was retrieved from the PDB, as 847 out of 1410 hits formed
one hydrogen bond. Cases of two hydrogen bonds were rarer
in both the CSD and PDB, with 37 and 496 hits, respectively.
Regarding structures containing a sulfone moiety, only 344 hits
out of the 2352 in the CSD established one hydrogen bond,
and approximately half of the hits in the PDB. Only a limited
number of sulfone formed two hydrogen bonds (Supporting
Information, Section S3, Table S3). The distances of the hydro-
gen bonds to the SO2 group are in a range of 2.8 to 3.5 ꢁ (Sup-
porting Information, Section S3, Figure S4) and no specific
directionality was observed as the hydrogen bonds cover the
Crystal structure determination of C364A-PvSHMT in com-
plex with (ꢂ)-11, (ꢂ)-17, and (ꢂ)-20
Several attempts made to co-crystallize either (ꢂ)-11, (ꢂ)-17,
or (ꢂ)-20 with wild-type PvSHMT showed no electron density
of the bound ligands. Instead, a partial or full formation of the
disulfide bridge between Cys125 and Cys364 was observed. To
circumvent the cysteine oxidation and prevent the disulfide
bridge formation, which might be linked to the binding of the
ligands, Cys364 was mutated to Ala364. Subsequently, co-crys-
tal structures of (+)-11, (+)-17, or (+)-20 with the C364A-
PvSHMT mutant were obtained. The co-crystals diffracted to
2.4, 2.2, and 2.6 ꢁ resolution, respectively, and belong to the
C2 space group. The structures were solved by molecular re-
placement using the coordinates of a chain A protomer of
PvSHMT (PDB ID code: 4OYT) as the template.[24] Despite using
a racemic mixture of ligands for co-crystallization, only the
(+)-(S) enantiomer was present in all structures. In the three
Chem. Eur. J. 2017, 23, 1 – 14
9
ꢂ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ