Development of bisubstrate analog inhibitors of aspartate N-acetyltransferase, a critical brain enzyme

Article abstract of DOI:10.1111/cbdd.13586

Canavan disease (CD) is a fatal leukodystrophy caused by mutations in the aspA gene coding for the enzyme aspartoacylase. Insufficient catalytic activity by this enzyme leads to the accumulation of its substrate, N-acetyl-l-aspartate (NAA), and diminished production of acetate in brain oligodendrocytes of patients with CD. There is growing evidence that this accumulation of NAA is the cause of many of the developmental defects observed in these patients. NAA is produced in the brain by a transacetylation reaction catalyzed by aspartate N-acetyltransferase (ANAT), and this membrane-associated enzyme has recently been purified as a soluble maltose binding protein fusion. Designing selective inhibitors against ANAT has the potential to slow the accumulation of NAA and moderate these developmental defects, and this is the goal of this project. Several bisubstrate analog inhibitors of ANAT have been synthesized that have achieved nanomolar level binding affinities against this enzyme. Truncated versions and fragments of these bisubstrate analog inhibitors have identified the essential structural elements needed for high binding affinity. More drug-like versions of these inhibitors can now be built, based on these essential core structures.

Full text of DOI:10.1111/cbdd.13586

PROFESSOR RONALD E VIOLA (Orcid ID : 0000-0002-3117-9168)
Article type : Research Article
Development of bisubstrate analog inhibitors of
aspartate N-acetyltransferase, a critical brain enzyme
Vinay Mutthamsetty1, Gopal P. Dahal2, Qinzhe Wang3 and Ronald E. Viola*
Department of Chemistry and Biochemistry, University of Toledo, Toledo, Ohio 43606, USA
Correspondence
Ronald E. Viola, Department of Chemistry and Biochemistry, University of Toledo, Toledo,
Ohio, USA. E-mail: ron.viola@utoledo.edu
1 Present address: Olon Ricerca Bioscience, Painesville, OH 44077
2 Present address: Merck & Company, Inc., Elkton, VA 22827
3
Present address: Department of Biochemistry, University of Utah School of Medicine, Salt
Lake City, UT 84112
Running title: Bisubstrate analog inhibitors of ANAT
Keywords: drug design; enzyme mechanism; Canavan disease; bisubstrate analogs; aspartate
N-acetyltransferase
This article has been accepted for publication and undergone full peer review but has not
been through the copyediting, typesetting, pagination and proofreading process, which may
lead to differences between this version and the Version of Record. Please cite this article as
doi: 10.1111/cbdd.13586
This article is protected by copyright. All rights reserved.

Abbreviations
ANAT, aspartate N-acetyltransferase; ASPA, aspartoacylase; boc, t-butyloxy carbonyl; CD,
Canavan disease; CNS, central nervous system; CoA, coenzyme A; DCM, dichloromethane;
DTNB, dithionitrobenzoate; EDCl, 1-ethyl-3-(3-(dimethylamino)propyl)carbodiimide; NAA,
N-acetyl-L-aspartate; PDB, protein data bank; TLC, thin layer chromatography; THF,
tetrahydrofuran.
Abstract
Canavan disease (CD) is a fatal leukodystrophy caused by mutations in the aspA gene coding
for the enzyme aspartoacylase. Insufficient catalytic activity by this enzyme leads to the
accumulation of its substrate, N-acetyl-L-aspartate (NAA), and diminished production of
acetate in brain oligodendrocytes of CD patients. There is growing evidence that this
accumulation of NAA is the cause of many of the developmental defects observed in these
patients. NAA is produced in the brain by a transacetylation reaction catalyzed by aspartate
N-acetyltransferase (ANAT), and this membrane-associated enzyme has recently been
purified as a soluble maltose binding protein fusion. Designing selective inhibitors against
ANAT has the potential to slow the accumulation of NAA and moderate these developmental
defects, and this is the goal of this project. Several bisubstrate analog inhibitors of ANAT
have been synthesized that have achieved nanomolar level binding affinities against this
enzyme. Truncated versions and fragments of these bisubstrate analog inhibitors have
identified the essential structural elements needed for high binding affinity. More drug-like
versions of these inhibitors can now be built, based on these essential core structures.
This article is protected by copyright. All rights reserved.

1. INTRODUCTION
Canavan disease (CD) is a neurological disorder caused by inherited genetic defects
(Canavan, 1931), leading to brain atrophy and degeneration into spongy tissues with
microscopic fluid-filled cystic cavities. CD is classified as a leukodystrophy, since it causes
disruptions in the growth and maintenance of the myelin sheath that serves to insulate
neuronal connections. Children born with the most severe forms of this disorder lack motor
skills and speech, and they also develop conditions such as hypotonia and macrocephaly
(Adachi et al., 1973, Adornato et al., 1972). The average life expectancy of CD patients is
only about 10 years. The underlying molecular basis for CD is due to multiple mutations in
the aspA gene (Matalon et al., 1988), which leads to deficiencies in the catalytic activity of
the enzyme aspartoacylase (ASPA) (Zano et al., 2013) that is localized in oligodendrocytes in
the brain. ASPA is the enzyme responsible for the production of the important CNS
metabolite acetate by hydrolysis of N-acetyl-L-aspartate (NAA). Acetate is the key starting
point for fatty acid and steroid synthesis, compounds which are further used as building
blocks in making myelin sheaths (Moffett et al., 2007). Different strategies have been
employed in an attempt to identify treatment therapies for CD, including brain-targeted gene
therapy against defective aspA gene (Janson et al., 2002), metabolic therapy by providing
acetate from external sources (Arun et al., 2010), lithium therapy (Janson et al., 2005), and
enzyme replacement therapy by intraperitoneal injection of protein surface modified ASPA
(Zano et al., 2011). So far, while each of these approaches have shown some promising
results in lowering NAA levels, none of these different strategies have identified a clear path
to a treatment therapy.
This article is protected by copyright. All rights reserved.

Some recent findings have supported the need for a completely different approach, with
the potential to produce a novel treatment for CD. It has been found that a knockout of the
Nat8l (N-acetyltransferase 8-like) gene that codes for aspartate N-acetyltransferase (ANAT),
the enzyme responsible for the synthesis of NAA, led to normal CNS myelination during the
early stages of brain development (Guo et al., 2015, Sohn et al., 2017), thereby eliminating
the symptoms of this leukodystrophy in a mouse model of CD. ANAT belongs to the
acetyltransferase enzyme superfamily, and catalyzes the acetylation of L-aspartate for
conversion into the nervous system specific metabolite NAA. This compound is an important
constituent in CNS metabolism, in fact NAA is one of the most abundant amino acids in the
brain, constituting about 3-4% of the total brain osmolarity (Baslow, 2000). NAA is
synthesized in neurons and is subsequently transported into the cytoplasm of
oligodendrocytes (Baslow, 2003), where it is hydrolyzed by the enzyme aspartoacylase to
release the acetate moiety which is then utilized as a building block in myelin biosynthesis
(Figure 1).
Based on these studies showing the important role that elevated NAA plays in the
etiology of CD, our planned approach involves designing and synthesizing selective
inhibitors against ANAT. The advantage of this approach is the capability to directly control
the elevated levels of NAA found in the brain of CD patients by rebalancing the rates of
NAA synthesis and utilization. In previously published work, we have produced and
characterized a soluble form of this membrane-associated enzyme (Wang et al., 2016), and
have synthesized several initial ANAT inhibitors based on two different core structures
(Thangavelu et al., 2017). These compounds are the first synthesized inhibitors against
ANAT that have been produced to date, with the best compounds having inhibition constants
(Ki values) in the low to sub-micromolar range.
This article is protected by copyright. All rights reserved.

To expand this search for improved inhibitors we have incorporated the structures of the
ANAT substrates into the design strategy. Coenzyme A (CoA) functions primarily through
the activation of acyl groups via formation of a thioester for their subsequent transfer to a
variety of different acceptor molecules (Jencks, 1973). There is general agreement that the
purpose of the extended coenzyme A structure, composed of adenosine-3′-monophosphate
linked by a pyrophosphate to a pantothenate coupled in a peptide linkage to β-
mercaptoethylamine, is to provide the specificity and binding affinity to its enzyme partners
(Engel & Wierenga, 1996). However, these individual moieties typically have little or no
affinity to these enzymes, suggesting a requirement for a coordinated ensemble of
interactions to achieve tight binding of this coenzyme. This need for binding cooperativity
between multiple functional groups makes the design and synthesis of potent and selective
inhibitors against CoA-utilizing enzymes particularly challenging.
As part of a systematic effort to develop inhibitors against an important brain enzyme
that catalyzes an acetyl transfer reaction, the affinities of bisubstrate analog inhibitors and
various truncated versions have been explored. We now report the synthesis and kinetic
evaluation of a series of CoA-based bisubstrate analogs with very high affinity towards
ANAT. In addition, this study includes the development and evaluation of a series of
truncated derivatives of this bisubstrate analog, compounds that are now available to be
further elaborated as a new family of potent inhibitors against ANAT.
This article is protected by copyright. All rights reserved.

2. METHODS AND MATERIALS
2.1 Materials
Solvents and reagents were used as purchased without further purification. The reaction time
courses for the newly synthesized enzyme inhibitors were monitored by TLC using an ethyl
acetate/hexane solvent system on Silica gel HLF plates (Analtech, Inc., Newark, DE). The
final products were purified by silica gel (230-400 mesh) flash column chromatography.
2.2 Spectroscopy
1H and 13C NMR spectra were recorded on either a Varian VXRS 400 MHz or a Varian
INOVA 600 MHz spectrometer. Electrospray–Atmospheric Pressure ionization (ES-API)
mass spectrometry of the final inhibitors were carried out on Agilent 1260 infinityII
spectrometer.
2.3 Enzyme production and inhibitor measurements
The recombinant maltose-binding protein-human aspartate N-acetyltransferase (MBP-ANAT)
fusion enzyme, produced in order to solubilize the membrane-associated enzyme, was
expressed and purified as reported previously (Wang et al., 2016). ANAT activity was
measured by using a previously described DTNB-based assay (Wright & Viola, 1998) using a
SpectraMax 190 spectrophotometer plate reader (Molecular Devices, CA). For inhibition
constant (Ki) determinations a two-fold serial dilution of each inhibitor was performed
starting from a stock inhibitor solution, with the starting concentration determined according
to the potency of the inhibitors. A negative control, in which no inhibitor was added, and a
background signal control, in which neither L-aspartate nor inhibitor were added, were
included in each set of assays.
This article is protected by copyright. All rights reserved.

3 RESULTS
3.1 Design of bisubstrate analog inhibitors
The transfer of the acetyl unit from acetyl-CoA to L-aspartate catalyzed by the enzyme
ANAT involves the nucleophilic attack of the α-amino group of aspartate on the carbonyl
carbon of the acetyl group. This reaction is proposed to go through a transition state in which
the newly forming C-N bond will alter the geometry around the carbonyl carbon of the acetyl
group as the C-S bond becomes elongated during the departure of coenzyme A (Figure 2). To
mimic this transition state a pair of bisubstrate analogs has been produced in which aspartate
is covalently coupled to acetyl-CoA, with the distance between the carbonyl carbon and the
departing thiol group varied from 1 to 2 bonds (Figure 2).
3.2 Synthesis of full length bisubstrate analogs
These full-length bisubstrate analogs have been synthesized by coupling coenzyme A to
activated N-acetyl or N-propionyl aspartate (Scheme 1). The bisubstrate analog with a single
methylene carbon linker (BA1) is a potent inhibitor of ANAT, with a Ki value of 275 nM
(Table 1). The importance of the aspartate carboxyl groups in substrate binding is
demonstrated by a greater than 60-fold loss of affinity when those functional groups are
converted to methyl esters (BA1Me). Increasing the distance between the thiol group and the
carbonyl carbon through the incorporation of an additional methylene carbon leads to a
further 6-fold improvement in binding affinity, with this bisubstrate analog (BA2) possessing
a Ki value of 48 nM as an inhibitor of ANAT (Table 1).
This article is protected by copyright. All rights reserved.

O
O
O
O
O
Br
O
O
Br
O
n
OH
O
HO
O
NH
O
a
b
NH2
1
O
NH2
HCl
O
n
Br
3
2
HO
OH
HO
O
P
O
O
N
HO
N
c
O
SH
P
P
N
H
N
H2N
O
O
H
O
N
OH
O
N
OH
O
4
O
O
O
HO
OH
HO
O
O
HN
O
P
O
N
HO
N
O
S
O
P
N
N
H
O
n
H2N
O
O
H
O
N
OH
O
N
P
OH
O
5
d
LiO
O
O
HO
OH
HO
O
O
HN
OLi
P
O
N
HO
N
O
S
O
P
P
N
N
H
O
n
H2N
O
O
H
O
N
OH
n = 1, BA1
2, BA2
O
N
OH
O
SCHEME 1 Synthesis of full length bisubstrate analogs BA1 and BA2. Reagents and
o
conditions: (a) SOCl2 in MeOH at 0 C to RT 14 h; (b) Bromoacetyl bromide or
bromopropionyl bromide, Et3N in THF at RT; c) MeOH/Et3N, H2CO3 (pH = 8) in (1:1) ratio,
8 h; (d) LiOH in THF/H2O, RT, 5 h.
3.3 Analysis of bisubstrate analog fragments
While binding potency is an important parameter in the development of inhibitors against a
target, the presence of multiple charged functional groups would likely preclude the transport
of these bisubstrate analogs to the brain of CD patients. Removal of the adenosyl phosphate
moiety from the intact bisubstrate analog (Figure 3) would eliminate several negatively-
charged phosphate groups that contribute to the nondrug-like properties of this enzyme
inhibitor. However, it is not clear how much the presence of this ADP moiety contributes to
This article is protected by copyright. All rights reserved.

the overall binding affinity of CoA. To examine the relative contributions of the different
structural elements to enzyme binding, several cofactor fragments were tested as potential
inhibitors. Neither adenosine nor pantothenate shows any measureable inhibition of ANAT
when tested at 2 mM concentrations. Adenosine 5′-monophosphate is a very weak inhibitor
of ANAT, but increasing the number of phosphate groups on the adenosine structure does
confer a slight improvement in binding affinity (Table 3). Adenosine-3′,5′-diphosphate,
adenosine-5′-diphosphate and adenosine 5′-triphosphate were each observed to be weak, but
slightly improved inhibitors of ANAT.
3.4 Synthesis of truncated bisubstrate analog derivatives
The overall affinity of the transition state analog mimics of the ANAT reaction are quite
strong, but the affinity contributed from the adenosyl phosphate moiety by itself is fairly
weak. It would be helpful to learn the importance of each of the other structural elements in
contributing to the overall high affinity of these bisubstrate analogs. To accomplish this aim a
series of truncated bisubstrate analogs were synthesized starting from the amino acyl end,
with the structures of these truncated analogs shown in Figure 3. To compare the individual
interactions/affinities of each truncated bisubstrate analog with ANAT, each fragment was
systematically assembled by using linear synthetic approaches (Schemes 2 and 3).
This article is protected by copyright. All rights reserved.

NHBoc
HS
O
O
O
O
R
O
R
R
O
(R=H,COOCH3)
b
Br
S
N
H
BocHN
BocHN
N
H
n
n
O
O
R
O
O
O
OLi
a
LiO
O
n=1, (3)
R=H, (6), 87%
NH
n
O
R=COOMe, (7), 82%
a
Br
O
OH
O
OLi
O
O
n=1 (TA1Br)
c
S
S
CF3COO H3N
N
H
N
H
LiO
n
n
HO
O
O
R=H & n=1 (TA2*)
R=COONa & n=1 (TA2a*)
R=H & n=1 (TA2)
R=COONa & n=1 (TA2a)
SCHEME 2 Synthesis of truncated bisubstrate analogs TA1Br and TA2. Reagents and
conditions: (a) LiOH in THF/H2O, RT, 5 h; (b) MeOH/Et3N, H2CO3 (pH = 8) in (1:1) ratio, 8
h; (c) CF3COOH in DCM at 0 oC, overnight.
SCHEME 3 Attempted synthesis of the truncated bisubstrate analog TA3. Reagents and
conditions: (a) CF3COOH in DCM, overnight; (b) N-boc-β-alanine, EDCl, Et3N in DCM for
5 h; (c) LiOH in THF/H2O, RT, 5 h. (d) CF3COOH in DCM at 0 oC, overnight.
This article is protected by copyright. All rights reserved.

3.5 Analysis of truncated bisubstrate analog inhibitors
N-acetylaspartate (TA1) is also a weak inhibitor of ANAT (Ki = 1.6 mM), with a binding
affinity that is comparable to that of the ADP moiety. The N-chloroacetyl derivative of
aspartate (TA1Cl) shows an 8-fold improvement in binding, while the N-bromoacetyl
derivative (TA1Br) does not show any measurable affinity to ANAT (Table 2). Coupling N-
bromoacetyl aspartate with cysteamine adds the thiol end of CoA, but this fragment (TA2)
does not show any measurable inhibitor of ANAT. However, the t-butyloxycarbonyl (boc)-
protected precursor of this fragment (TA2*) was found to be a weak inhibitor (Ki = 2.1 mM).
Replacing cysteamine with cysteine introduced an additional carboxyl group (TA2a),
resulting in weak binding to ANAT. As with the cysteamine adduct, the boc-protected
precursor of the cysteine adduct (TA2a*) is a slightly better inhibitor (Ki = 0.87 mM)
suggesting the presence of a hydrophobic pocket in the cofactor binding site. Addition of the
next moiety, β-alanine, was attempted to extend the structure of the CoA analog fragment
(Scheme 3, TA3) but, unfortunately, multiple attempts at deprotection of the protected
intermediate (compound 9) each resulted in substantial degradation of the final product.
Pantothenic acid, the middle component of the coenzyme A structure that connects ADP to β-
cysteamine (Figure 3), does not bind to ANAT with any measurable affinity. So it is unlikely
that extending the CoA structure by coupling pantothenate to produce an even longer
fragment would lead to a substantial increase in the affinity of this truncated fragment series.
Since the expanded bisubstrate analog (BA2) is the most potent ANAT inhibitor (Table
1), a related set of truncated analogs were synthesized starting with N-propionyl aspartate.
However, none of these compounds, including N-bromopropionyl aspartate, the cysteamine
adduct, or the boc-protected cysteamine adduct, showed any appreciable inhibition of ANAT
(data not shown).
This article is protected by copyright. All rights reserved.

3.6 Cooperative binding of bisubstrate analog fragments
The failure to observe improved binding affinity through the inclusion of the additional
structural components of the CoA structure (TA1 and TA2) suggests that the high affinity of
the bisubstrate analogs structures (BA1 and BA2) comes not from interactions with the
central structural components of CoA, but from the combined interactions of the functional
groups at each end of this cofactor structure. To test this idea the binding of the truncated
fragment series was re-examined in the presence of high levels of several different adenosyl
phosphates.
For each of the truncated analogs tested the presence of 3′,5′-ADP, 5′-ADP or ATP was
examined at a minimum of five-times excess over their respective Ki values to measure any
possible effects on analog binding to ANAT. The addition of 12 mM 3′,5′-ADP had no
measurable effect on the binding of either TA2 or the boc-protected version (TA2*).
However, for the extended analogs that included a cysteamine or cysteine moiety (TA2 and
analogs), the presence of 5′-ADP or ATP led to an average of ~1.4-fold improvement in
binding affinity (Table 4). The greatest cooperative effect for these analogs was seen between
the binding of 5′-ADP and that of N-(acetylcysteinyl)aspartate (TA2a), with a 2.5-fold
increase in analog affinity. For the shorter truncation analogs (TA1) the presence of ATP
caused the greatest increase in binding, with a 2.8-fold and 4.7-fold improvement in the Ki
values for TA1 and TA1Cl, respectively (Table 4).
The reverse experiments were also conducted to measure any possible enhancements
induced in ADP or ATP binding in the presence of high levels of several TA inhibitors. Here
the effects were much less pronounced, with only slight enhancements in adenine nucleotide
binding in most cases. The largest increases in affinity were observed for the combinations
that also gave the best improvements in analog affinity, such as ATP binding in the presence
of TA1 and TA1Cl. But even here the enhancements in ATP affinity was less than 2-fold.
This article is protected by copyright. All rights reserved.

4. DISCUSSION
4.1 General design of bisubstrate analogs
Bisubstrate analogs, sometimes referred to as transition state analogs, were originally
developed to examine the chemical mechanism of an enzyme (Jin et al., 1999, Parmentier et
al., 1992), to measure the effects of mutations on substrate recognition (Zeringo & Bellizzi,
2014, Hengge et al., 1996), and to mimic the transition state in enzyme structural studies
(Lolis & Petsko, 1990, Betts et al., 1994, Lowther et al., 1999, Focia et al., 1998). Bisubstrate
analogs are designed to incorporate the binding motifs of both substrates for the target
enzyme within the same molecule. As a consequence, bisubstrate analogs tend to have
affinities that are significantly greater than the sum of binding energies of the individual
substrates (Parang & Cole, 2002, Wolfenden, 1972). While this high affinity feature is part
of the design goals, because of the difficulty in mimicking the transient structure of a reaction
transition state, this enhanced affinity is not always achieved. For example, the bisubstrate
inhibitors P1-(5-adenosyl)-P3-(6-glucosyl) triphosphate and P1-(5-adenosyl)-P4-(6-glucosyl)
tetraphosphate, that were synthesized to mimic the transition state of the reaction catalyzed
by hexokinase, were found to be only weak inhibitors with their binding affinities lower than
the either of the substrates ATP and glucose (Danenberg & Danenberg, 1977). It is clearly
important to understand the mechanism of the reaction that is being catalyzed, as well as the
nature of the proposed transition state structure in order to design bisubstrate analogs with the
optimal potency.
4.2 Bisubstrate analog inhibitors of ANAT
Identifying the active site structure of an enzyme target is an important criteria to guide the
development of new inhibitors. Because the structure of ANAT is still unknown, developing
selective active site-specific inhibitors against this enzyme is more challenging. There is,
This article is protected by copyright. All rights reserved.

however, detailed structural information about the substrates for this enzyme-catalyzed
reaction that can be used to help guide inhibitor design. ANAT is an acetyltransferase,
catalyzing an acylation reaction where the α-amino group of L-aspartate acts as a nucleophile
attacking the carbonyl carbon of acetyl CoA (Figure 2). To achieve optimal potency a
bisubstrate analog must closely mimic the transition state of the reaction. Because the exact
distance between the two substrate molecules in the transition state of this reaction is not
precisely known, bisubstrate analogs were designed with two different linkers between the
substrates to investigate the effect of this change on the binding affinity to the enzyme. The
direct coupling of the acetyl group of N-acetylaspartate to the thiol group of coenzyme A
produces a bisubstrate analog (BA1) with mid-nanomolar level binding affinity to ANAT.
The incorporation of an additional methylene carbon between the transferred acetyl group
and coenzyme A (BA2) leads to a further increase in binding affinity beyond that of the
already potent inhibitor with the shorter linker (Table 1). This improved affinity suggests that
this enzyme-catalyzed reaction has a late transition state in which the acetyl group has been
substantially transferred to L-aspartate, with the bond to the departing thiol of coenzyme A
significantly elongated in the transition state.
4.3 Role of structural moieties in bisubstrate analog binding
Now that an optimized bisubstrate analog structure has been synthesized, it would be useful
to identify the essential binding determinants, and also to eliminate non-essential
components, to arrive at a core structure that can be used to guide the development of more
drug-like ANAT inhibitors. By synthesizing truncated derivatives of this bisubstrate analog,
the correlation between the affinities of these truncated structures and the functional groups
that are present in each derivative can be examined. Given the importance of precisely
orienting the two substrates in the active site of ANAT for efficient acetyl transfer, it was
This article is protected by copyright. All rights reserved.

expected that the functional groups at each end of the bisubstrate analog would make the
majority of the important contributions to the overall affinity. However, neither the adenosine
phosphate moiety nor the product NAA were found to have greater than millimolar affinities
towards ANAT, values that are about four orders-of-magnitude weaker than the best
bisubstrate analog inhibitor (Table 2). Coupling several additional structural moieties of
coenzyme A, including cysteamine and β-alanine, to NAA did not confer any additional
affinity to the target enzyme.
There are several possible reasons for the failure of these extended substrate analogs to
show increased binding affinity to ANAT. These linear molecules are quite flexible, with
multiple rotatable bonds that lead to a large number of possible conformers. This lack of
rigidity decreases the population of the optimal binding conformation, leading to lower
overall affinity to a given target. A similar phenomenon was observed when examining the
transport and uptake properties of small molecule drugs. Drugs that are more flexible tend to
accumulate to a much lower level in Gram-negative bacteria when compared to more rigid
drugs with similar overall pharmacological properties (Richter et al., 2017).
In addition to being quite flexible, the central pantothenate and cysteamine moieties have
relatively few functional groups that can make productive binding interactions in the
coenzyme A binding pocket. To identify the importance of different CoA functional groups
in binding affinity, a series of structures of N-acetyltransferase enzymes were examined. The
target sequence of ANAT was provided and a closest homolog template search was
performed using the SWISS-MODEL server (Arnold et al., 2006). The results were filtered by
selecting structures with either acetyl CoA or CoA bound and any redundant structures were
discarded. The top 20 structures were then downloaded from the Protein Data Bank (PDB)
(Berman et al., 2000), and further analyzed using Ligplot (Wallace et al., 1995) to identify
the key functional groups involved in either hydrogen bonding or hydrophobic interactions
This article is protected by copyright. All rights reserved.

with their receptor residues. As expected, in all of the structures studied the adenosine moiety
makes multiple hydrogen bonding interactions, and the negatively charged phosphate groups
make numerous electrostatic interactions with positively charged residues in the binding
pocket. To eliminate these dominant interactions, subsequent analysis was limited to only the
pantothenate and cysteamine moieties in order to identify the potential effect of the functional
groups in each truncated analog on the binding affinity. There are five functional groups in
the pantothenate and cysteamine region of the structure that can potentially participate in
hydrogen bonding interactions (Figure 3). However, in half of the structures that were
examined there are two or fewer hydrogen bonds made, and only two structures show more
than three hydrogen bonds with these functional groups. The hydrogen bonding interactions
identified from this region of the CoA structure occur primarily through the carbonyl group
from the pantothenate moiety and the amide group from the cysteamine. The weak binding
affinity of these truncated analogs with ANAT is likely due to a combination of the limited
number of bonding interactions to this region of the cofactor structure, as well as to the
higher flexibility of the truncated analogs.
4.4 Binding cooperativity to improve truncated analog binding
The presence of millimolar levels of adenosine nucleotides as metabolites should lead to
some occupancy of the adenosine sub-site in the CoA binding pocket of ANAT under cellular
conditions. To test the possible effects of nucleotide binding on the affinity of the set of
truncated bisubstrate analogs several adenosine phosphates were included in ANAT inhibitor
binding studies. While the addition of 3′,5′-ADP had no effect on truncated analog binding,
both 5′-ADP and ATP did show cooperative effects on analog binding. The addition of
millimolar levels of ADP had a greater effect on the binding of the longer truncated analogs
(TA2), with up to a 60% improvement in the Ki value. For the shorter truncated analogs
This article is protected by copyright. All rights reserved.

(TA1) the presence of ATP had the greater effect, leading to as much as an 80%
improvement in binding affinity. This 40-fold increase in binding affinity relative to NAA
binding brings this truncated analog inhibitor (TA1Cl) to within a factor of 200 of the binding
affinity of the full length bisubstrate analog (BA1). These results suggest that as improved
inhibitors are developed that optimize the potential interactions in the substrate and thiol end
of the CoA binding site of ANAT, the normal cellular levels of ADP and ATP could serve to
enhance inhibitor binding to this target enzyme. These studies also highlight the importance
of occupying both the nucleotide and the aspartate binding pockets to achieve maximum
inhibition.
5 CONCLUSIONS
Bisubstrate analogs have been quite useful in a wide variety of applications, and many of
these compounds have been found to be quite potent enzyme inhibitors. However, because
these analogs typically incorporate charged and polar functional groups to make additional
interactions at the active site, these compounds generally do not have the optimal structural
properties to be considered as viable drug candidates. The general inability of acetyl-CoA and
other CoA analogs to penetrate into cell plasma membranes also makes them poor starting
points for drug candidate development (Robishaw & Neely, 1985). In particular, the highly
charged ATP region of acetyl CoA makes any bisubstrate analogs that incorporate this
moiety much less likely to cross a membrane barrier. These bisubstrate analog inhibitors of
ANAT have provided information on the transition state of this reaction and established the
drugability of this important enzyme. In order to develop drug-like candidates that mimic the
substrate structures of ANAT it will be necessary to design inhibitors based on this core
bisubstrate analog structure, but with neutral isosteric replacements of the charged functional
groups.
This article is protected by copyright. All rights reserved.

CONFLICT OF INTEREST
The authors declare that they have no competing interests.
References
Adachi, M., Schneck, L., Cara, J. & Volk, B. W. (1973). Human Pathol. 4, 331-347.
Adornato, B. T., O'Brien, J. S., Lampert, P. W., Roe, T. F. & Neustein, H. B. (1972).
Neurology 22, 202-210.
Arnold, K., Bordoli, L., Kopp, J. & Schwede, T. (2006). Bioinformatics 22, 195-201.
Arun, P., Madhavarao, C. N., Moffett, J. R., Hamilton, K., Grunberg, N. E., Ariyannur, P. S.,
Gahl, W. A., Anikster, Y., Mog, S., Hallows, W. C., Denu, J. M. & Namboodiri, A. M. A.
(2010). J. Inherit. Metab. Dis. 33, 195-210.
Baslow, M. H. (2000). J. Molec. Neurosci. 15, 61-69.
Baslow, M. H. (2003). Neurochem. Res. 28 941-953.
Berman, H. M., Westbrook, J., Feng, Z., Gilliland, G., Bhat, T. N., Weissig, H., Shindyalov,
I. N. & Bourne, P. E. (2000). Nucl. Acids Res. 28, 235-242.
Betts, L., Xiang, S., Short, S. A., Wolfenden, R. & Carter, C. W. (1994). J. Molec. Biol. 235,
635-656.
Canavan, M. M. (1931). Arch. Neurol. Psychiat. 25, 299-308.
Danenberg, P. V. & Danenberg, K. D. (1977). Biochim. Biophys. Acta 480, 351-356.
Engel, C. & Wierenga, R. (1996). Curr. Opin. Struct. Biol. 6, 790-797.
Focia, P. J., Craig, S. P. & Eakin, A. E. (1998). Biochemistry 37, 17120-17127.
Guo, F., Bannerman, P., Ko, E. M., Miers, L., Xu, J., Burns, T., Li, S., Freeman, E.,
McDonough, J. A. & Pleasure, D. (2015). Ann. Neurol. 77, 884-888.
Hengge, A. C., Denu, J. M. & Dixon, J. E. (1996). Biochemistry 35, 7084-7092.
Janson, C., McPhee, S., Bilaniuk, L., Haselgrove, J., Testaiuti, M., Freese, A., Wang, D. J.,
Shera, D., Hurh, P., Rupin, J., Saslow, E., Goldfarb, O., Goldberg, M., Larijani, G., Sharrar,
This article is protected by copyright. All rights reserved.

W., Liouterman, L., Camp, A., Kolodny, E. H., Samulski, J. & Leone, P. (2002). Human
Gene Therapy 13, 1391-1412.
Janson, C. G., Assadi, M., Francis, J., Bilaniuk, L., Shera, D. & Leone, P. (2005). Pediat.
Neurol. 33, 235-243.
Jencks, W. P. (1973). The Enzymes 9, 483-496.
Jin, L., Stec, B., Lipscomb, W. N. & Kantrowitz, E. R. (1999). Proteins 37, 729-742.
Lolis, E. & Petsko, G. A. (1990). Ann. Rev. Biochem. 59, 597-630.
Lowther, W. T., Zhang, Y., Sampson, P. B., Honek, J. F. & Matthews, B. W. (1999).
Biochemistry 38, 14810-14819.
Matalon, R., Michals-Matalon, K., Sebesta, M., Deanching, M., Gashkoff, P. & Casanova, J.
(1988). Am. J. Med. Genet. 29, 463-471.
Moffett, J. R., Ross, B., Arun, P., Madhavarao, C. N. & Namboodiri, M. A. A. (2007). Prog.
Neurobiol. 81, 89-131.
Parang, K. & Cole, P. A. (2002). Pharm. Therap. 93, 145-157.
Parmentier, L. E., O'Leary, M. H., Schachman, H. K. & Cleland, W. W. (1992). Biochemistry
31, 6598-6602.
Richter, M. F., Drown, B. S., Riley, A. P., Garcia, A., Shirai, T., Svec, R. L. & Hergenrother,
P. J. (2017). Nature 545, 299-304.
Robishaw, J. D. & Neely, J. R. (1985). Am. J. Physiol. 248, E1-E9.
Sohn, J., Bannerman, P., Guo, F., Burns, T., Miers, L., Croteau, C., Singhal, N. K.,
McDonough, J. A. & Pleasure, D. (2017). J. Neurosci. 37, 413-421.
Thangavelu, B., Mutthamsetty, V., Wang, Q. & Viola, R. E. (2017). Bioorg. Med. Chem. 25,
870-885.
Wallace, A. C., Laskowski, R. A. & Thornton, J. M. (1995). Prot. Engr. 8, 127-134.
Wang, Q., Zhao, M., Parungao, G. G. & Viola, R. E. (2016). Prot. Expres. Purif. 119, 11-18.
Wolfenden, R. (1972). Acc. Chem. Res. 5, 10-18.
Wright, S. K. & Viola, R. E. (1998). Anal. Biochem. 265, 8-14.
This article is protected by copyright. All rights reserved.

Zano, S., Malik, R., Szucs, S., Matalon, R. & Viola, R. E. (2011). Molec. Genet. Metab. 102,
176-180.
Zano, S., Wijayasinghe, Y. S., Malik, R., Smith, J. & Viola, R. E. (2013). J. Inherit. Metab.
Dis. 36, 1-6.
Zeringo, N. A. & Bellizzi, J. J. (2014). Chem. Biol. Drug Des. 84, 697-703.
TABLE 1
Kinetic evaluation of ANAT bisubstrate analogs
Analog
number
bisubstrate structure
Ki (nM)
BA1
BA1Me
BA2
275 10
18,000 2,000
48
8
This article is protected by copyright. All rights reserved.

TABLE 2
Kinetic evaluation of truncated analog inhibitors of ANAT
Analog
truncated analog name
Ki (mM)
numbera
TA1
TA1Cl
TA1Br
TA2
N-acetyl aspartate
N-chloroacetyl aspartate
1.60 0.12
0.20 0.04
>2
N-bromoacetyl aspartate
N-(acetylcysteaminyl) aspartate
N-(acetylcysteaminyl boc ester) aspartate
N-(acetylcysteinyl) aspartate
N-(acetylcysteinyl boc ester) aspartate
>2
TA2*
TA2a
TA2a*
2.10 0.37
1.03 0.20
0.87 0.16
a asterisk (*) indicates the boc-protected form of the inhibitor
TABLE 3 Kinetic Evaluation of Coenzyme A Fragments as Possible Inhibitors
substrate analog
Ki (mM)
adenosine
n.i.a
n.i.a
pantothenate
adenosine 5’-monophosphate
adenosine 5’-diphosphate
adenosine 5’-triphosphate
adenosine 3’,5’-diphosphate
4.0 0.8
1.10 0.15
0.96 0.12
2.30 0.24
a no inhibition of ANAT observed when tested at 2 mM
This article is protected by copyright. All rights reserved.

TABLE 4
Cooperative effect of simultaneous binding of ANAT inhibitors
Ki (mM) (plus adenosine inhibitor)
Analog
number
none
+ 5 mM ATP
0.58 0.08
0.043 0.005
----
+ 5.5 mM ADP
0.85 0.13
0.170 0.025
----
TA1
TA1Cl
TA1Br
TA2*
TA2a
1.61 0.12
0.20 0.04
>2
2.10 0.37
1.03 0.20
0.87 0.16
1.52 0.20
0.66 0.10
0.84 0.09
1.46 0.19
0.41 0.05
0.73 0.10
TA2a*
This article is protected by copyright. All rights reserved.

Figure Legends
FIGURE 1 Metabolism of N-acetyl-L-aspartate in the brain. Aspartate N-acetyltransferase
(ANAT) catalyzes the transfer of the acetyl group from acetyl-coenzyme A to produce N-
acetyl-L-aspartate (NAA). NAA is transported to oligodendrocytes by a specific
dicarboxylate transporter (NaDC3) where it is hydrolyzed by aspartoacylase to release acetate
and regenerate L-aspartate. Activation of acetate by coenzyme A provides the precursor for
fatty acid biosynthesis.
FIGURE 2 ANAT-catalyzed reaction, showing the transfer of the acetyl group from
coenzyme A to L-aspartate, and the proposed transition state for this reaction. The structure
of a pair of bisubstrate analogs designed to mimic this transition state are shown, where n is
either one (BA1) or two (BA2).
FIGURE 3 Truncation series of ANAT bisubstrate analog inhibitors. The sub-structures of
acetyl CoA-aspartate bisubstrate analogs are indicated, from N-acetylaspartate (TA1), to the
cysteamine adduct (TA2), the β-alanine adduct (TA3) and the bridging pantothenate.
This article is protected by copyright. All rights reserved.

This article is protected by copyright. All rights reserved.

This article is protected by copyright. All rights reserved.