Full Paper
fluorescence-based assays.[16] Previous attempts to develop
ALDH1A1-specific radiotracers have so far failed due to the
poor cellular retention of the carboxylate product, presumed
to be a consequence of its high hydrophobicity.[17] Here, we
report the synthesis and biological evaluation of 18F-fluorinated
aldehyde-based probes for the non-invasive detection of
ALDH1A1 activity in tumor cell models.
using recombinant human ALDH1A1, ALDH2 and ALDH3A1 en-
zymes, which are commonly expressed in human cancer.
ALDH2 is expressed in the mitochondrial matrix and plays a
critical role in alcohol metabolism,[19] whereas ALDH3A1 is lo-
calized in both the nucleus and cytosol and functions to detox-
ify aldehydes formed during UV-induced lipid peroxidation.[20]
Furthermore, these three isoforms exhibit three different rate-
limiting steps: ALDH1A1’s being the cofactor dissociation,[21]
the deacylation-step for ALDH2,[22] and hydride transfer for
ALDH3A1.[23]
Results and Discussion
We made a single-point modification to 2, with the addition
of a fluorine to the aromatic ring—an essential requirement
for 18F-radiofluorine-based radiotracers—to give compounds
3a and 3b. The lithium/bromine exchange on 11 followed by
quenching with DMF afforded aldehyde 12. Stirring 12 with
NaBH(OAc)3 and diethylamine in DCE afforded the crude re-
ductive amination product which was directly treated with
aqueous hydrochloric acid in THF to furnish 3a (Scheme 1A).
Reduction of nitrile 13 with diisobutylaluminium hydride clean-
ly produced aldehyde 14 which was thereafter stirred with
excess diethylamine in THF to yield aldehyde 3b (Scheme 1B).
Amide 16 was prepared by reacting the acid chloride of 5-fluo-
ronicotinic acid 15 with amine B (Scheme 1C). The acid-cata-
lyzed cleavage of the acetal furnished compound 4a. LiAlH4 re-
duction of the amide bond in 16 followed by the acid-cata-
lyzed acetal cleavage afforded amine 5 (Scheme 1C). In a simi-
lar manner, the acid chloride of 6-fluronicotinic acid 16 was re-
acted with amine A to yield amide 18 which following an acid-
catalyzed acetal cleavage furnished aldehyde 4b. The sulfona-
mide 6 was accessed by reacting the commercially available 5-
fluoropyridine-3-sulfonyl chloride 19 with amine A following
an acid-catalyzed acetal cleavage (Scheme 1E).
ALDH1A1 chemical probes were designed to have a) an alde-
hyde that can serve as a substrate for ALDH1A1; b) contain a
(radio)fluorine atom that would allow for detection via gamma
counting/PET imaging; c) a suitable hydrophobic-hydrophilic
balance which would allow for passive diffusion in and out of
cells, and importantly; d) subsequent trapping of the in situ
generated carboxylic acid product within the cytosol as a
result of the acquired negative charge (Figure 1A). We took a
substrate-based approach for the imaging of ALDH1A1 to pro-
vide a functional readout of enzymatic activity. Substrate-
based radiotracers provide an advantage over radiolabeled in-
hibitors which only report on enzyme expression. Moreover,
multiple substrate molecules can be turned over by a single
enzyme, thereby increasing the sensitivity of detection when
compared to radiolabeled inhibitors.
The benzylic amine 2 showed a higher affinity (lower KM)
and catalytic efficiency (Vmax/KM) for ALDH1A1 over both
ALDH2 and ALDH3A1 (Table 1, entry 1 and Figure 2B, respec-
tively), indicating that the isozyme selectivity of DEAB was
maintained. Interestingly, 3a and 3b exhibited lower affinity
for ALDH1A1 than the non-fluorinated analogue 2, with a KM
of 0.28Æ0.12 mm, 0.26Æ0.08 mm and 0.16Æ0.03 mm, respec-
tively (Table 1, entries 1–3).
Figure 1. A) Schematic illustrating ALDH-mediated trapping of 18F-labeled al-
dehydes by conversion to their corresponding acid. B) Chemical structures
of DEAB (1), an ALDH1A1 inhibitor, and 2, an ALDH1A1 substrate.
Given that fluorination proximal to the aldehyde moiety de-
creased affinity for ALDH1A1, we next explored compounds
with fluorine atoms that were remote from the aldehyde. Com-
pounds 4a and 4b exhibited a five-fold increase in ALDH1A1
affinity over 2 (Table 1, entries 1, 4 and 5). Furthermore, the
enzyme efficiency for 4a was 7-fold higher for ALDH1A1 than
ALDH2, with the enzyme efficiency for ALDH1A1 over 20-fold
higher with respect to ALDH3A1; that is, the linking of the pyri-
dine via an amide bond resulted in improved selectivity for
ALDH1A1 (Figure 2B; Table 1, entry 4). The position of the fluo-
rine on the pyridine ring crucially did not play a key role in
substrate kinetics, as seen with amide 4b which exhibited
analogous behavior to 4a, albeit with marginally decreased
ALDH1A1 enzyme selectivity (Table 1, entry 5). In the absence
of the amide linkage, the tertiary amine 5, whilst exhibiting a
similar ALDH1A1 binding profile as compounds 4a and 4b,
was readily oxidized by ALDH2 when compared to the other
The starting point for our small molecule probe develop-
ment was N,N-diethylaminobenzaldehyde (DEAB) 1, which is a
well-known inhibitor of ALDH1A1 (Ki values of 10–40 nm).[18]
Given the advantages of substrate-based radiotracers, our ini-
tial goal was to convert 1 in to a substrate whilst maintaining
suitable ALDH1A1 selectivity over other commonly expressed
ALDH isozymes. 1 is thought to form a stalled acyl-enzyme in-
termediate as a result of the delocalization of the electron lone
pair of the para-substituted amine into the aromatic ring.[18a,c]
To convert 1 from an inhibitor to a substrate, we uncoupled
the amine nitrogen from the aromatic p-system through the
introduction of a methylene linker to give 2 (Figure 1B), which
was rapidly converted to the carboxylate product under the
action of ALDH1A1 (Figure S1). In order to assess the enzyme
kinetics of the compounds in this study, we examined the
effect of substrate concentration on the initial enzyme velocity
&
&
Chem. Eur. J. 2019, 25, 1 – 8
2
ꢁ 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!
Pereira, Raul
