Histidine triad nucleotide-binding proteins HINT1 and HINT2 share similar substrate specificities and little affinity for the signaling dinucleotide Ap4A

Article abstract of DOI:10.1002/1873-3468.13745

Human histidine triad nucleotide-binding protein 2 (hHINT2) is an important player in human mitochondrial bioenergetics, but little is known about its catalytic capabilities or its nucleotide phosphoramidate prodrug (proTide)-activating activity akin to the cytosolic isozyme hHINT1. Here, a similar substrate specificity profile (k_cat/K_m) for model phosphoramidate substrates was found for hHINT2 but with higher k_cat and K_m values when compared with hHINT1. A broader pH range for maximum catalytic activity was determined for hHINT2 (pK₁?=?6.76?±?0.16, pK₂?=?8.41?±?0.07). In addition, the known hHINT1-microphthalmia-inducing transcription factor-regulating molecule Ap₄A was found to have no detectable binding to HINT1 nor HINT2 by isothermal titration calorimetry. These results demonstrate that despite differences in their sequence and localization, HINT1 and HINT2 have similar nucleotide substrate specificities, which should be considered in future proTide design and in studies of their natural function.

Full text of DOI:10.1002/1873-3468.13745

Histidine triad nucleotide-binding proteins HINT1 and
HINT2 share similar substrate specificities and little
affinity for the signaling dinucleotide Ap4A
Alexander Strom1, Cher Ling Tong2 and Carston R. Wagner1
1
2
Department of Medicinal Chemistry, University of Minnesota, Minneapolis, MN, USA
Department of Biochemistry Molecular Biology and Biophysics, University of Minnesota, Minneapolis, MN, USA
Correspondence
Human histidine triad nucleotide-binding protein 2 (hHINT2) is an important
player in human mitochondrial bioenergetics, but little is known about its cat-
alytic capabilities or its nucleotide phosphoramidate prodrug (proTide)-acti-
vating activity akin to the cytosolic isozyme hHINT1. Here, a similar
substrate specificity profile (kcat/Km) for model phosphoramidate substrates
was found for hHINT2 but with higher kcat and Km values when compared
with hHINT1. A broader pH range for maximum catalytic activity was
determined for hHINT2 (pK1 = 6.76 ꢀ 0.16, pK2 = 8.41 ꢀ 0.07). In addi-
tion, the known hHINT1-microphthalmia-inducing transcription factor-regu-
lating molecule Ap4A was found to have no detectable binding to HINT1 nor
HINT2 by isothermal titration calorimetry. These results demonstrate that
despite differences in their sequence and localization, HINT1 and HINT2
have similar nucleotide substrate specificities, which should be considered in
future proTide design and in studies of their natural function.
C. R. Wagner, Department of Medicinal
Chemistry, University of Minnesota, 2231
6th S.E., Rm 2-141 CCRB, Minneapolis, MN
55455, USA
E-mail: wagne003@umn.edu
(Received 28 September 2019, revised 10
December 2019, accepted 18 December
2019, available online 17 February 2020)
doi:10.1002/1873-3468.13745
Edited by Peter Brzezinski
Keywords: Ap4A; hHINT1; hHINT2; pH profile; phosphoramidate;
proTide; substrate specificity
The human histidine triad nucleotide-binding protein 2
(hHINT2) is a mitochondrially localized member of
the histidine triad superfamily, containing the charac-
teristic His-X-His-X-His-X-X active site motif, where
X is a hydrophobic residue with the second histidine
acting as a catalytic nucleophile [1,2]. The closely
related enzyme HINT1 is localized in the cytosol and
hydrolyzes nucleotide phosphoramidates, as well as
nucleotide acyl phosphates [3]. hHINT1 has been
shown to be essential for the activation of nucleotide
phosphoramidate prodrugs (ProTides) including the
first curative antiviral, sofosbuvir [4,5].
nucleotide, thus helping mask the nucleotide’s anionic
charge and in turn easing the usually difficult pharma-
cokinetic task of crossing the cell membrane [6,7].
Once inside the cell, the removal of the masking amino
acid group necessitates hHINT1’s phosphoramidase
activity to yield the bare nucleotide which can be con-
verted to the active therapeutic triphosphorylated form
by kinases. In addition to its usefulness as a tool in
the ProTide drug delivery approach, hHINT1 has also
been shown to be involved in a number of diverse bio-
logical processes including regulation of the transcrip-
tion factor microphthalmia-inducing transcription
factor (MITF) in mast cells and the interaction
between the l-opioid receptor and N-methyl-D-aspar-
tate receptor [8–12].
In the ProTide approach for nucleotide drug deliv-
ery, an amino acid among other requisite functional
groups is coupled to the phosphate of a therapeutic
Abbreviations
hHINT2, human histidine triad nucleotide-binding protein 2; ITC, Isothermal titration calorimetry; MITF, microphthalmia-inducing transcription
factor; SPR, surface plasmon resonance.
FEBS Letters 594 (2020) 1497–1505 ª 2020 Federation of European Biochemical Societies
1497

HINT1 and HINT2 share similar substrate specificities with little Ap4A affinity
A. Strom et al.
Studies of hHINT2 employing hHINT2ꢁ/ꢁ mice
have demonstrated the enzyme’s widespread impor-
tance in bioenergetic processes such as glucose home-
ostasis, lipid accumulation, and respiratory chain
control [13–15]. Looking at the molecular level in
HINT2ꢁ/ꢁ mice, metabolic phenotypes have been
observed that are consistent with an overall hyper-
acetylation of the mitochondrial proteome and a meta-
bolic shift to steatosis [15,16]. The mitochondria of
HINT2ꢁ/ꢁ mice exhibit decreased deacetylase activity
by sirtuins 3 and 5, diminished glutamate dehydroge-
nase activity, and decreased functional respiratory
complexes I, II, and IV [13,15]. Investigations focused
on mitochondrial membrane channel proteins have
also shown that HINT2 expression levels impact mito-
chondrial influx of ions and small molecules via the
mitochondrial Ca2+ uniporter and mitochondrial per-
meability transition pore with implications for the
onset and treatment of hepatocellular and pancreatic
carcinomas [17–19].
study performed with tissue prepared sheep liver
HINT2 with adenosine and cytosine phosphoramidate
[22]. Here, we have synthesized eleven nucleotide phos-
phoramidates to explore the substrate specificity of
HINT2 in which the nucleobase, ribose sugar, and
coupled side-chain amino acid were varied. A key fea-
ture of our analysis is the incorporation of a fluores-
cent indole moiety that upon phosphoramidate
hydrolysis becomes unquenched, allowing the reaction
to be continuously monitored (Fig. 1) [4]. Previously,
using steady-state kinetic and X-ray crystallographic
structural studies, we have established that HINT1
prefers phosphoramidate and acyl nucleotidylate sub-
strates composed of purine nucleobases, ribose sugars,
alkyl amines, and D-amino acids [4,23]. Structural
analysis of the catalytically dead HINT1 mutant,
H112N, bound with these substrates later established
that the specificity of the active site results from a
large hydrophobic nucleobase binding site, pivotal
hydrogen bonds with the ribose 20 and 30 hydroxyl
groups, and the ability for bulky D-amino acids to
avoid major steric clashes with the boundaries of the
active site [23].
Despite the myriad of mitochondrial processes asso-
ciated with HINT2, a molecular explanation for the
role of HINT2 on these bioenergetic pathways has yet
to be realized. In particular, little is known about the
catalytic mechanism or substrate specificity of HINT2
and how it compares to HINT1. Consequently, as a
first step, a substrate specificity profile of hHINT2
would begin to shed light on differences between the
catalytic capabilities of hHINT1 and hHINT2 and
their relative phosphoramidate prodrug activation
potential. Additionally, while the native substrate
remains unknown, the important secondary signaling
dinucleotide Ap4A has been shown to play a role in
HINT1-mediated regulation of the MITF [9,20,21].
Thus, a comparison of Ap4A binding to hHINT1 and
hHINT2 could differentiate between their cellular
roles.
HINT1 and HINT2 share 61% sequence identity and
are even more similar within their active sites. When
overlapping both enzymes’ active site residues by align-
ꢀ
ing all atoms within 8 A of the enzyme-bound product,
AMP, the resulting root-mean-squared deviation in
ꢀ
alignment is minimal (RMSD = 0.21 A) (Fig. 2). Nearly
every residue is identical between these isozymes’ active
sites, with the exception of minor hydrophobic residue
substitutions such as Ile-27 and Ile-44 in HINT1 and
Leu-64 and Val-81 in HINT2, respectively. Given the
significant similarity between the HINT1 and HINT2
active sites, we hypothesized that the substrate specifici-
ties of the enzymes were likely to be similar. As stated
before, hHINT1 is crucial in activating nucleotide pro-
drugs (ProTides) in the cytosol, and thus, characterizing
the substrate specificity of hHINT2 with a wide array of
To our knowledge, studies of the catalytic activity
of human HINT2 have not been reported beyond a
Fig. 1. Scheme of nucleotide phosphoramidate hydrolysis under HINT catalysis. In solution, the substrate is quenched, but after hydrolysis
the indole-containing tryptamine is freed allowing it to fluoresce.
1498
FEBS Letters 594 (2020) 1497–1505 ª 2020 Federation of European Biochemical Societies

A. Strom et al.
HINT1 and HINT2 share similar substrate specificities with little Ap4A affinity
crystallized and kinetically analyzed [24]. The hHINT2
cDNA construct was inserted into the pMCSG9 vector
which adjoins a TEV-cleavable His-6-tagged maltose-bind-
ing protein for improved protein yields and purification.
Rosetta2 (DE3) pLysS competent bacteria were trans-
formed, and protein expression was induced with 1 mM
IPTG at OD600 = 0.6 under a 37 °C incubation for 4 h.
Cells were lysed, purified by Ni-NTA chromatography, the
N-terminal His6-MBP was cleaved with 2% by weight
TEV protease, and the final dimeric hHINT2 protein pro-
duct was purified via SEC-FPLC. hHINT1 has been shown
to readily form homodimers, and hHINT2 was observed to
do so as well [27]. The hHINT1 used in Ap4A binding
studies was expressed and purified identically, but the
pMCSG7 vector was used instead. The expected molecular
weight of both hHINT1 and hHINT2 proteins was verified
Fig. 2. Alignment of the hHINT1 (magenta) and hHINT2 (green)
by LC/MS using reverse-phase chromatography on
a
active sites with the enzymatic product AMP bound (blue circle)
ꢀ
with all atoms within 8 A used in the active site alignment. The
Thermo Scientific (Waltham, MA, USA) LTQ XL Mass
Spectrometer (Fig. S1).
catalytic H112 for hHINT1 and H149 for hHINT2 are circled in
ꢀ
black. PDB IDs 3TW2 and 4INI. Alignment RMSD = 0.21 A.
Substrate synthesis
Substrate compounds were synthesized as previously
described through EDC coupling of either tryptamine or
the amine of tryptophan methyl ester with a nucleotide
monophosphate in water at pH 7.0 for 4 h [4,28]. Com-
pounds were purified by cation exchange and flash chro-
nucleotide phosphoramidates would be valuable infor-
mation regarding the feasibility of a nucleotide prodrug
approach for drug targets localized in the mitochondria.
An additional analysis of HINT2’s catalytic rate as
a function of pH was also conducted to compare
with HINT1’s previously reported pKa values [3].
Despite active site homogeneity, differences in distant
residues could alter the rate-limiting step of HINT
catalysis. The kinetic mechanism for HINT1 has been
shown to proceed through rapid nucleotidylation of a
catalytic histidine (H112) followed by partially rate-
limiting hydrolysis of the essential nucleotidylated his-
tidine and by partially rate-limiting product dissocia-
tion [3]. Further, since the secondary signaling
molecule Ap4A has been postulated to regulate the
intracellular behavior of HINT1, we evaluated and
compared the affinity of Ap4A for HINT1 and
HINT2 [9,21].
1
matography and characterized for purity by H-NMR, 31P-
NMR, and MS.
Continuous fluorescence phosphoramidase
activity kinetic assay
Described in detail previously [3,4], the kinetic fluorescence
assay tracks the hydrolysis and release of the fluorescent
indole bearing side-chain product as it is freed from the
intramolecular quenching of the substrate’s nucleobase, with
excitation and emission wavelengths of 280 and 360 nm
(Fig. 1). Measurements were taken in a quartz cuvette in a
temperature-controlled Varian/Cary Eclipse fluorimeter
(25 °C). Five nanomolar HINT enzyme was used in degassed
buffer (20 mM HEPES, 1 mM MgCl2, pH 7.2 buffer), sub-
strate concentrations were varied, and the linear fluorescence
slopes were measured and converted to velocities using fluo-
rescence standard curves of the reagents. Measurements were
taken in triplicate, and Michaelis–Menten curves were fitted
using GRAPHPAD PRISM (San Diego, CA, USA).
Materials and methods
Protein expression and purification
Human HINT2 was produced recombinantly with a 36-
residue N-terminal truncation for two reasons [24]. First,
hHINT2 contains an N-terminal mitochondrial localization
tag that upon entry into the organelle would be expected to
be cleaved at either residue position 30 or residue position
31, as predicted by localization predictive software Mito-
Prot II and MitoFates, respectively [25,26]. Second, we
found that loss of the 36-residue N-terminal peptide greatly
reduced protein aggregation allowing the protein to be
pH rate profile determination
The maximum velocity of hHINT2 was measured under 19
pH conditions in 100 mM phosphate buffer from pH 5.8–
8.0 or 100 mM Tris from pH 7.5–9.1 using a saturating
concentration of substrate 1 (30 µM) and 5 nM of enzyme.
Within the overlapping pH range of 7.5–8.0 in which both
FEBS Letters 594 (2020) 1497–1505 ª 2020 Federation of European Biochemical Societies
1499

HINT1 and HINT2 share similar substrate specificities with little Ap4A affinity
A. Strom et al.
Results and Discussion
hHINT1 and hHINT2 substrate specificity profile
comparison
As can be seen in Table 1, both hHINT1 and hHINT2
are highly efficient enzymes with substrate specificities
ꢁ1
up to 107
M
ꢂsꢁ1. The substrate specificity results
Scheme 1. The three-state two-pKa model used to fit the measured
demonstrate that HINT2 prefers purine nucleobases
over pyrimidines, a trend that is also a defining char-
acteristic of HINT1 (Table 1). The majority of this
preference is contributed by an increase in the Km for
pyrimidines (comparing compounds 1-5) indicating
that the smaller hydrophobic nucleobase has difficulty
binding the active site, but at higher substrate concen-
trations is still able to undergo rapid rates of enzyme-
induced hydrolysis. While hHINT1 and hHINT2 both
pH profile curve of HINT2 for comparison with HINT1.
phosphate and Tris were used, three measurements were
made in Tris while two were made in phosphate in order to
help account for any effects the necessary change between
buffering agents might impart upon the fitted pKas. Mea-
surements were taken in triplicate and fit to the following
equation which has been previously determined to describe
the three-state two-pKa model of the dual histidine bearing
hHINT1 active site which hHINT2 also shares [3]. Fitting
was performed using the ORIGIN software package,
Northampton, MA, USA (Scheme 1).
show
a preference for purines, a key difference
between the isozymes is that while hHINT2’s kcat and
Km values are both higher than hHINT1’s kcat and Km
values, the enzymes exhibit similar substrate specificity
values (kcat/Km). This pattern not only holds true when
comparing purine and pyrimidines, but also holds true
when comparing with the other test substrates. While
the increase in kcat for hHINT2 over hHINT1 can be
almost fourfold (i.e., compound 5) and differences in
Km as high as sevenfold (i.e., compound 1), the sub-
strate specificities rarely differ by a factor greater than
3, due to a nearly balanced change in both the kcat
and Km values.
Isothermal titration calorimetry
Isothermal titration calorimetry (ITC) experiments were
conducted in
a MicroCal Auto-ITC200 System (GE
Healthcare Life Sciences, Marlborough, MA, USA). All
titration experiments were performed at 25 °C in ITC
buffer (10 mM Tris, 150 mM NaCl, pH 7.2). hHint1 and
hHINT2 were buffer exchanged into ITC buffer using
Micro Bio-Spin 6 columns (Bio-Rad, Hercules, CA, USA).
Ap4A was purchased from Sigma-Aldrich, St. Louis, MO,
USA. While a variety of concentrations were tested, all of
which yielded the same results, the final high reagent con-
centrations which we report were 1 mM HINT protein and
20 mM Ap4A ligand. Final protein concentrations were
determined via NanoDrop absorbance using an extinction
coeffect from the ExPASy ProtParam Web tool. Twenty
injections of ligand were injected (injection volume 2 lL)
into the protein cell. The resulting change in enthalpy was
measured, and the background heat of dilution was sub-
tracted by performing similar measurements in the absence
of protein. The background heat of dilution was subtracted
from the resulting data.
Compounds 6-9 exhibit the impact of altering the
ribose ring, namely at the 2’ position. The ribose-to-
arabinose substitution of compound 8 compared with
compound 1 reduced the substrate specificity by nearly
20-fold due to the b position of the 20-hydroxyl group,
however, not as dramatically as the 100-fold reduction
observed after removal of the 20-hydroxyl moiety. The
lack of a 20-hydroxyl group and incorporation of a
pyrimidine base reduced the substrate specificity of
compound
7 to such an extent that acceptable
Michaelis–Menten curves could not be achieved
because of the constraints of the fluorescence assay
(i.e., maximum substrate concentration 50 lM) due to
inner filter effects [4]. Compound 6, when compared to
compound 1, lacks a 20-hydroxyl resulting in an
increase in the Km value by approximately 24-fold,
while only modestly lowering the kcat by nearly three-
fold.
When compared to compound 1, the 20-methyl
group of compound 9, surprisingly, was found to
increase the Km values by approximately 18- and five-
fold for HINT1 and HINT2, respectively, and decrease
the kcat values by approximately four- and 16-fold for
The Ap4A used in the study, purchased from Sigma-
Aldrich, was verified to have a mass of 837.0 by positive
mode mass spectrometry. 1H as well as 31P-NMR also
confirmed the quality and identity of the compound (Figs
S2 and S3). Positive control ITC experiments were also
conducted, showing that the hHINT1 and hHINT2 used
in the study could bind known and previously published
hHINT binding ligands (AMP and an acyl-sulfamate
inhibitor) despite not being capable of binding Ap4A
(Fig. S4).
1500
FEBS Letters 594 (2020) 1497–1505 ª 2020 Federation of European Biochemical Societies

A. Strom et al.
HINT1 and HINT2 share similar substrate specificities with little Ap4A affinity
Table 1. Substrate specificity of hHINT1 and hHINT2 measured over seven substrate concentrations in triplicate with standard error
reported of the nonlinear regression Michaelis–Menten fit. Substrates with <L.O.D reported were relatively very poor substrates, and a
suitable Michaelis–Menten curve could not be generated due to the limits of the assay.
ꢁ1
kcat (sꢁ1
)
Km (µM)
kcat/Km (M ꢂsꢁ1) 9 103
Compound R1
R2
R3
R4
hHINT1a
hHINT2
hHINT1a
hHINT2
hHINT1a
hHINT2
1
2
Adenine
OH
OH
OH
OH
H
H
2.1 ꢀ 0.1
6.02 ꢀ 0.22
5.66 ꢀ 0.20
2.05 ꢀ 0.05
4.46 ꢀ 0.14
8.00 ꢀ 0.21
0.13 ꢀ 0.02 0.91 ꢀ 0.14 15000 ꢀ 3000 6630 ꢀ 1030
0.21 ꢀ 0.02 0.82 ꢀ 0.11 11000 ꢀ 1000 6930 ꢀ 998
Guanine
Cytosine
Uracil
H
H
2.3 ꢀ 0.1
1.2 ꢀ 0.1
2.5 ꢀ 0.3
2.6 ꢀ 0.04
3
H
H
2.3 ꢀ 0.4
2.2 ꢀ 0.4
5.59 ꢀ 0.42
4.86 ꢀ 0.46
600 ꢀ 200
1200 ꢀ 500
3700 ꢀ 300
195 ꢀ 16
367 ꢀ 29
917 ꢀ 92
8600 ꢀ 899
64.2 ꢀ 7.0
< L.O.D.
4
H
H
5
Hypoxanthine OH
H
H
0.71 ꢀ 0.03 0.93 ꢀ 0.09
3.72 ꢀ 0.28 24.0 ꢀ 2.4
6
Adenine
Thymine
Adenine
Guanine
Guanine
Guanine
H
H
H
H
H
0.729 ꢀ 0.019 1.54 ꢀ 0.07
7
H
H
0.10 ꢀ 0.01
1.1 ꢀ 0.03
NA
32 ꢀ 5
> 50.0
3.1 ꢀ 1.0
8
OH
H
1.88 ꢀ 0.15
1.0 ꢀ 0.1
5.56 ꢀ 1.34
1100 ꢀ 100
226 ꢀ 50
338 ꢀ 86
90.4 ꢀ 20.5
21.5 ꢀ 4.9
< L.O.D.
9
OH Me
H
0.519 ꢀ 0.052 0.358 ꢀ 0.042 2.30 ꢀ 0.46 3.96 ꢀ 0.77
10
11
OH
OH
H
H
R-COOMe
0.26 ꢀ 0.02
0.111 ꢀ 0.007
3.3 ꢀ 0.6
31 ꢀ 1
5.17 ꢀ 1.15
> 50.0
80 ꢀ 30
S-COOMe 0.0071 ꢀ 0.002 NA
0.23 ꢀ 0.01
a
hHINT1 values are taken from our previous publication of the enzyme’s substrate specificity profile except for newly synthesized and mea-
sured compounds 6 and 9, hence their differences in significant figures [4].
HINT1 and HINT2, respectively. Consequently, the
substrate specificity of HINT1 for compound 9 is
nearly fourfold greater than that observed for HINT2.
When compared to compound 1, the 20-methyl group
reduced the substrate specificity of HINT1 and HINT2
by 138- and 322-fold, respectively. Thus, while the 20-
methyl group significantly affects the substrate speci-
ficity of HINT1 and HINT2, it is one of the few sub-
strates that impacted the catalytic activity of HINT2
more than HINT1. The root of this difference between
the enzymes is not clear given that the active sites are
nearly identical [23,24].
hHINT2 maintains catalytic activity in a broader
pH range than hHINT1
In addition to a substrate specificity profile, a pH pro-
file with substrate 1 from pH 5.8–9.1 was conducted to
assess the effect of pH on hHINT2’s catalytic activity
since its localization in the mitochondria may have
evolved it to accommodate different pH conditions
than hHINT1. Using a saturating concentration of
substrate 1 in 100 mM phosphate buffer from pH 5.8–
8.0 or 100 mM Tris from pH 7.5–9.1, the maximum
velocity of hHINT2 was observed (Fig. 3). The result-
ing bell-shaped curve with an elevated leftward arm
tracks well with that of previously reported hHINT1,
and thus, the same three-state two-pKa model was
used for data fitting [3]. While operating most effec-
tively at a similar and nearly neutral physiological pH
as HINT1, hHINT2 appears to be slightly more toler-
ant to a range of varying pH conditions given that the
bell-shaped curve defined by the two pKas is modestly
expanded (hHINT1 pK1 = 7.52 ꢀ 0.18, hHINT2
pK1 = 6.76 ꢀ 0.16 and hHINT1 pK2 = 8.15 ꢀ 0.10,
hHINT2 pK2 = 8.41 ꢀ 0.07). The more acidic pK1
decrease and the basic pK2 increase result in a wider
range of catalytic capability at both ends of the pH
profile. Given that the pH of the cytosol is approxi-
mately 7.4, while the pH of the mitochondrial inner
Finally, compounds 10 and 11 are relatively poor
substrates due to the added steric bulk that the trypto-
phan-like side chain provides as opposed to the smal-
ler tryptamine utilized in the other substrates. These
types of substrates impact not only the Km, but also
the kcat, which has previously been shown to result in
alterations in the distance between the catalytic his-
tidine and the phosphoramidate with crystal structures
of an inactive hHINT1 mutant [24]. In both enzymes,
the L-tryptophan containing compound 11 is a notably
poor substrate, due to the especially sterically pro-
hibitive side-chain moiety. Similar to compound 7 with
hHINT2, the high Km of compound 11 prevented a
suitable Michaelis–Menten curve from being obtained
due to the limits of the assay.
FEBS Letters 594 (2020) 1497–1505 ª 2020 Federation of European Biochemical Societies
1501

HINT1 and HINT2 share similar substrate specificities with little Ap4A affinity
A. Strom et al.
associated with His151, which is important for binding
and proton transfer [4].
Ap4A binds neither hHINT1 nor hHINT2 by ITC
Beyond catalytic differences between HINT1 and
HINT2, the binding affinity of the important sec-
ondary signaling molecule Ap4A was examined by
ITC. HINT1 has been associated with tumor suppres-
sion and possesses an important transcriptional regula-
tory role through its association with the MITF
[20,21]. The dinucleotide Ap4A has been shown to dis-
rupt the association of HINT1 and MITF in cells, and
a crystal structure of an Ap4A analog bridging two
hHINT1 active sites has been solved. Nevertheless, the
thermodynamics of HINT1 binding to Ap4A has not
been clearly defined [31]. We hypothesized that Ap4A
binding affinity would differ between HINT1 and
HINT2, as the latter enzyme is localized in the mito-
chondria and no evidence suggests it is involved in the
regulation of MITF.
pK1 = 6.76
pK2 = 8.41
Fig. 3. Maximum catalytic rate of hHINT2 with substrate 1 as a
function of pH measured in triplicate and fit to a 3-state 2-pKa
model described in the methods. Confidence interval = 0.987.
membrane space and mitochondrial matrix is 6.8 and
7.7, respectively, we postulate that hHINT2 is able to
maintain effective catalytic activity both in the cytosol
and in the mitochondria despite the potential pH dif-
ferences [29,30]. Similar to hHINT1, pKa1 of hHINT2
is likely associated with the ionization state of the
nucleophilic residue, His149, while the second is likely
Surprisingly, ITC analysis of Ap4A protein binding
revealed that neither HINT1 nor HINT2 bound Ap4A
in vitro (Fig. 4). Steady-state kinetic competition
assays were also performed between Ap4A and sub-
strate 1, and again, inhibition of HINT1 activity was
not observed, thus indicating that Ap4A is not a com-
petitive inhibitor of HINT1 (data not shown). In
Fig. 4. (A) Representative ITC enthalpograms of 20 mM Ap4A injected into a 1 mM HINT1 containing reaction cell compared with Ap4A
injected into buffer alone. (B) Final subtracted enthalpogram of Ap4A-HINT1 binding (or lack thereof), accounting for the heat of dilution of
Ap4A. HINT2 data are not shown but also displayed a lack of binding.
1502
FEBS Letters 594 (2020) 1497–1505 ª 2020 Federation of European Biochemical Societies

A. Strom et al.
HINT1 and HINT2 share similar substrate specificities with little Ap4A affinity
contrast to these findings, crystal structure analysis
and a surface plasmon resonance (SPR) binding study
have suggested that Ap4A is able to bind directly to
HINT1 [21,31]. Addressing the crystallographic evi-
dence first, a possible explanation for this discrepancy
from our work is the fact that a much less polar glyc-
erol-containing Ap4A analog, not Ap4A itself, was
used in the reported crystal structure, thus removing
two of the four anionic phosphates from the true sys-
tem [31]. In another study that evaluated hHINT1
binding to the dinucleotide by SPR, the immobiliza-
tion of HINT1 as a glutathione S-transferase fusion
protein to an SPR flow chip was required [21]. In nei-
ther case was the catalytic viability and thus native
functionality of the enzyme determined. Thus, the dis-
crepancy observed between the ITC binding and cat-
alytic inhibition results and the previous results for
HINT1 and Ap4A binding may be dependent on mod-
ifications in the interaction between HINT1 and Ap4A
imposed by the unique methods used for those studies.
Recently, it was shown that HINT1 can be post-
translationally modified at two residues surrounding
its active site, whereas the protein used in the experi-
ments we have performed are not (Fig. S1) [9]. The
molecular interaction between Ap4A and the hHINT1-
MITF system in vivo may hinge on these modifications
or other nuances not yet captured in vitro. In addition,
since Ap4A is known to inhibit the ability of HINT1
to hydrolyze lysyl-AMP generated by lysyl-tRNA syn-
thetase (KRS) at low micromolar concentrations,
future studies on the interaction of Ap4A with HINT1,
KRS, and MITF may be better focused on the interac-
tions between Ap4A and KRS rather than HINT1
[20].
cytosolic hHINT1. Thus, although the native substrates
for both HINT1 and HINT2 are unknown, it is very
likely that they are similar. HINT1 has been shown to
efficiently hydrolyze lysyl-AMP produced by lysyl-
tRNA synthetase, whether this is also the case for
HINT2 and mitochondrial lysyl-tRNA synthetase
remains to be determined [32]. As for the secondary
signaling molecule, Ap4A, and its role in potentially
regulating MITF function, we have found that Ap4A
binds neither wild-type non-post-translationally modi-
fied hHINT1 nor hHINT2. Ongoing studies to eluci-
date the role of Ap4A on the regulation of this and
other cellular functions will hopefully shed light on the
molecular mechanism governing the variety of cellular
functions associated with HINT proteins.
Acknowledgements
This work was supported by the American Foundation
for Pharmaceutical Education’s Predoctoral Fellowship
awarded to Alexander Strom and the University of
Minnesota Foundation.
Author contributions
AS and CRW conceived the work, analyzed data, and
revised the manuscript. AS wrote the manuscript. AS
and CLT conducted the experiments and analyzed the
data. CLT also revised the manuscript.
References
1 Brenner C, Bieganowski P, Pace HC and Huebner K
(1999) The histidine triad superfamily of nucleotide-
binding proteins. J Cell Physiol 181, 179–187.
2 Mootha VK, Bunkenborg J, Olsen JV, Hjerrild M,
Wisniewski JR, Stahl E, Bolouri MS, Ray HN, Sihag S,
Kamal M et al. (2003) Integrated analysis of protein
composition, tissue diversity, and gene regulation in
mouse mitochondria. Cell 115, 629–640.
3 Zhou X, Chou TF, Aubol BE, Park CJ, Wolfenden R,
Adams J and Wagner CR (2013) Kinetic mechanism of
human histidine triad nucleotide binding protein 1.
Biochemistry 52, 3588–3600.
4 Chou TF, Baraniak J, Kaczmarek R, Zhou X, Cheng J,
Ghosh B and Wagner CR (2007) Phosphoramidate
pronucleotides: a comparison of the phosphoramidase
substrate specificity of human and Escherichia coli
histidine triad nucleotide binding proteins. Mol Pharm
4, 208–217.
5 Cha A and Budovich A (2014) Sofosbuvir: a new oral
once-daily agent for the treatment of hepatitis C virus
infection. P T. 39, 345–352.
Conclusions
As stated previously, neither the specific mitochondrial
destination nor the natural substrate of hHINT2 is
known, but parallels with hHINT1 can be drawn in
light of the fact that they exhibit similar substrate
specificities. Both enzymes prefer purines over pyrim-
idines, ribose sugars, and D over L-amino acid side
chains with the first two components having the great-
est effect on the Km and the third impacting both kcat
and Km. Under varying pH conditions, hHINT2 is able
to operate maximally under physiological pH, similar
to hHINT1, while tolerating both acidic and basic con-
ditions modestly better. Given hHINT2’s ability to
hydrolyze a relatively large spectrum of nucleotide
phosphoramidates, we believe that nucleotide prodrugs
(ProTides) would likely be metabolized by hHINT2 in
the mitochondria to a similar extent as observed with
FEBS Letters 594 (2020) 1497–1505 ª 2020 Federation of European Biochemical Societies
1503

HINT1 and HINT2 share similar substrate specificities with little Ap4A affinity
A. Strom et al.
6 Abraham TW, Kalman TI, McIntee EJ and Wagner
CR (1996) Synthesis and biological activity of aromatic
amino acid phosphoramidates of 5-fluoro-2’-
deoxyuridine and 1-beta-arabinofuranosylcytosine:
evidence of phosphoramidase activity. J Med Chem 39,
4569–4575.
7 Mehellou Y, Rattan HS and Balzarini J (2018) The
ProTide Prodrug technology: from the concept to the
clinic. J Med Chem 61, 2211–2226.
8 Razin E, Zhang ZC, Nechushtan H, Frenkel S, Lee Y-
N, Arudchandran R and Rivera J (1999) Suppression of
microphthalmia transcriptional activity by its
association with protein kinase C-interacting protein 1
in mast cells. J Biol Chem 274, 34272–34276.
9 Motzik A, Amir E, Erlich T, Wang J, Kim B-G, Han
JM, Kim JH, Nechushtan H, Guo M, Razin E et al.
(2017) Post-translational modification of HINT1
mediates activation of MITF transcriptional activity in
human melanoma cells. Oncogene 36,
17 Ndiaye D, Collado-Hilly M, Martin J, Prigent S,
Dufour JF, Combettes L and Dupont G (2013)
Characterization of the effect of the mitochondrial
protein Hint2 on intracellular Ca(2+) dynamics.
Biophys J 105, 1268–1275.
18 Chen L, Sun Q, Zhou D, Song W, Yang Q, Ju B,
Zhang L, Xie H, Zhou L, Hu Z et al. (2017) HINT2
triggers mitochondrial Ca2+ influx by regulating the
mitochondrial Ca2+ uniporter (MCU) complex and
enhances gemcitabine apoptotic effect in pancreatic
cancer. Cancer Lett 411, 106–116.
19 Li W, Cai S, Wang L, Yang C, Zhou B and Wang H
(2017) HINT2 downregulation promotes colorectal
carcinoma migration and metastasis. Oncotarget 8,
13521–13531.
20 Chou TF and Wagner CR (2007) Lysyl-tRNA
synthetase-generated lysyl-adenylate is a substrate for
histidine triad nucleotide binding proteins. J Biol Chem
282, 4719–4727.
4732–4738.
21 Lee YN, Nechushtan H, Figov N and Razin E
(2004) The function of lysyl-tRNA synthetase and
Ap4A as signaling regulators of MITF activity in
FcepsilonRI-activated mast cells. Immunity 20,
145–151.
ꢁ
ꢁ
~
ꢁ
ꢁ
10 Garzon J, Rodrıguez-Munoz M and Sanchez-Blazquez
P (2012) Direct association of Mu-opioid and NMDA
glutamate receptors supports their cross-regulation:
molecular implications for opioid tolerance. Curr Drug
Abuse Rev 5, 199–226.
ꢁ
22 Bretes E, Wojdyła-Mamon AM, Kowalska J, Jemielity
J, Kaczmarek R, Baraniak J and Guranowski A (2013)
ꢁ
ꢁ
~
11 Garzon J, Herrero-Labrador R, Rodrıguez-Munoz M,
ꢁ
Shah R, Vicente-Sanchez A, Wagner CR and Sanchez-
ꢁ
Hint2, the mitochondrial nucleoside 5’-
ꢁ
Blazquez P (2015) HINT1 protein: a new therapeutic
phosphoramidate hydrolase; properties of the
homogeneous protein from sheep (Ovis aries) liver. Acta
Biochim Pol 60, 249–254.
target to enhance opioid antinociception and block
mechanical allodynia. Neuropharmacology 89, 412–423.
12 Shah RM, Peterson C, Strom A, Dillenburg M, Finzel
B, Kitto KF, Fairbanks C, Wilcox G and Wagner CR
(2019) Inhibition of HINT1 modulates spinal
nociception and NMDA evoked behavior in mice. ACS
Chem Neurosci 10, 4385–4393.
23 Maize KM, Shah R, Strom A, Kumarapperuma S,
Zhou A, Wagner CR and Finzel BC (2017) A crystal
structure based guide to the design of human histidine
triad nucleotide binding protein 1 (hHint1) activated
ProTides. Mol Pharm 14, 3987–3997.
13 Martin J, Maurhofer O, Bellance N, Benard G, Graber
F, Hahn D, Galinier A, Hora C, Gupta A, Ferrand G
et al. (2013) Disruption of the histidine triad nucleotide-
binding hint2 gene in mice affects glycemic control and
mitochondrial function. Hepatology 57, 2037–2048.
14 Martin J, Balmer ML, Rajendran S, Maurhofer O,
Dufour JF and St-Pierre MV (2016) Nutritional stress
exacerbates hepatic steatosis induced by deletion of the
histidine nucleotide-binding (Hint2) mitochondrial
protein. Am J Physiol Gastrointest Liver Physiol 310,
G497–G509.
15 Rajasekaran R, Felser A, Nuoffer JM, Dufour JF and
St-Pierre MV (2018) The histidine triad nucleotide-
binding protein 2 (HINT-2) positively regulates
hepatocellular energy metabolism. FASEB J 32, 5143–
5161.
24 Maize KM, Wagner CR and Finzel BC (2013)
Structural characterization of human histidine triad
nucleotide-binding protein 2, a member of the histidine
triad superfamily. FEBS J 280, 3389–3398.
25 Claros MG and Vincens P (1996) Computational
method to predict mitochondrially imported proteins
and their targeting sequences. Eur J Biochem 241, 779–
786.
26 Fukasawa Y, Tsuji J, Fu SC, Tomii K, Horton P and
Imai K (2015) MitoFates: improved prediction of
mitochondrial targeting sequences and their cleavage
sites. Mol Cell Proteomics 14, 1113–1126.
27 Chou TF, Tikh IB, Horta BA, Ghosh B, De Alencastro
RB and Wagner CR (2007) Engineered monomeric
human histidine triad nucleotide-binding protein 1
hydrolyzes fluorogenic acyl-adenylate and lysyl-tRNA
synthetase-generated lysyl-adenylate. J Biol Chem 282,
15137–15147.
16 Anderson KA, Wang D and Hirschey MD (2013)
HINT2 and fatty liver disease: mitochondrial protein
hyperacetylation gives a hint? Hepatology 57, 1681–
1683.
28 Okon A, Matos de Souza MR, Shah R, Amorim R, da
Costa LJ and Wagner CR (2017) Anchimerically
1504
FEBS Letters 594 (2020) 1497–1505 ª 2020 Federation of European Biochemical Societies

A. Strom et al.
HINT1 and HINT2 share similar substrate specificities with little Ap4A affinity
activatable antiviral ProTides. ACS Med Chem Lett 8,
Supporting information
958–962.
Additional supporting information may be found
online in the Supporting Information section at the end
of the article.
29 Santo-Domingo J and Demaurex N (2012) Perspectives
on: SGP symposium on mitochondrial physiology and
medicine: the renaissance of mitochondrial pH. J Gen
Physiol 139, 415–423.
30 Porcelli AM, Ghelli A, Zanna C, Pinton P, Rizzuto R
and Rugolo M (2005) pH difference across the outer
mitochondrial membrane measured with a green
fluorescent protein mutant. Biochem Biophys Res
Commun 326, 799–804.
Fig. S1. Full protein mass spectra (left) and deconvo-
luted mass (right) of purified hHINT1 (top) and
hHINT2 (bottom) used in all studies presented.
1
Fig. S2. H-NMR spectra of the Ap4A purchased from
Sigma-Aldrich and utilized in Ap4A binding studies.
Fig. S3. 31P-NMR spectra of the Ap4A purchased
from Sigma-Aldrich and utilized in Ap4A binding
studies.
31 Dolot R, Kaczmarek R, Seda A, Krakowiak A,
z
Baraniak J and Nawrot B (2016) Crystallographic
studies of the complex of human HINT1 protein with a
non-hydrolyzable analog of Ap4A. Int J Biol Macromol
87, 62–69.
Fig. S4. Isothermal titration calorimetry positive con-
trols, verifying the ability of the hHINT1 and hHINT2
protein used in the study to bind known, previously
published hHINT binding ligands despite not binding
Ap4A.
ꢁ
32 Sissler M, Gonzalez-Serrano LE and Westhof E (2017)
Recent advances in mitochondrial Aminoacyl-tRNA
Synthetases and disease. Trends Mol Med 23, 693–708.
FEBS Letters 594 (2020) 1497–1505 ª 2020 Federation of European Biochemical Societies
1505