Supercharging Prions via Amyloid-Selective Lysine Acetylation

Article abstract of DOI:10.1002/anie.202103548

Repulsive electrostatic forces between prion-like proteins are a barrier against aggregation. In neuropharmacology, however, a prion's net charge (Z) is not a targeted parameter. Compounds that selectively boost prion Z remain unreported. Here, we synthes

Full text of DOI:10.1002/anie.202103548

A Journal of the Gesellschaft Deutscher Chemiker
Angewandte
Chemie
International Edition
www.angewandte.org
Accepted Article
Title: Supercharging prions via amyloid-selective lysine acetylation
Authors: Katelyn M. Baumer, Christopher D. Cook, Collin T. Zahler,
Alexandra A. Beard, Zhijuan Chen, Jordan C. Koone, Chad
M. Dashnaw, Raul A. Villacob, Touradj Solouki, John L.
Wood, David R. Borchelt, and Bryan F. Shaw
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Angew. Chem. Int. Ed. 10.1002/anie.202103548
Link to VoR: https://doi.org/10.1002/anie.202103548

10.1002/anie.202103548
Angewandte Chemie International Edition
Supercharging prions via amyloid-selective lysine acetylation
Running title: Supercharging amyloid
Katelyn M. Baumera, Christopher D. Cooka, Collin T. Zahlera, Alexandra A. Bearda, Zhijuan
Chenb, Jordan C. Koonea, Chad M. Dashnawa, Raul A. Villacoba, Touradj Soloukia, John L.
Wooda, David R. Borcheltb, Bryan F. Shawa*
aDepartment of Chemistry and Biochemistry, Baylor University, Waco, TX
bDepartment of Neuroscience, University of Florida, Gainesville, FL
*To whom correspondence should be addressed: bryan_shaw@baylor.edu
Keywords: superoxide dismutase, neurodegeneration, amyotrophic lateral sclerosis, protein
aggregation, acetylation.
This article is protected by copyright. All rights reserved.

10.1002/anie.202103548
Angewandte Chemie International Edition
Abstract
Repulsive electrostatic forces between prion-like proteins are a barrier against aggregation. In
neuropharmacology, however, a prion’s net charge (Z) is not a targeted parameter. Compounds
that selectively boost prion Z remain unreported. Here, we synthesized compounds that amplified
+
the negative charge of misfolded superoxide dismutase-1 (SOD1) by acetylating lysine-NH3 in
amyloid-SOD1, without acetylating native-SOD1. Compounds resembled a “ball and chain” mace:
a rigid amyloid-binding “handle” (benzothiazole, stilbene, or styrylpyridine); an aryl ester “ball”;
and a triethylene glycol chain connecting ball to handle. At stoichiometric excess, compounds
acetylated up to 9 of 11 lysine per misfolded subunit (∆Zfibril = -8,100 per 103 subunits). Acetylated
amyloid-SOD1 seeded aggregation more slowly than unacetylated amyloid-SOD1 in vitro and
organotypic spinal cord (these effects were partially due to compound binding). Compounds
exhibited reactivity with other amyloid and non-amyloid proteins (e.g., fibrillar α-synuclein was
peracetylated; serum albumin was partially acetylated; carbonic anhydrase was largely
unacetylated).
1
This article is protected by copyright. All rights reserved.

10.1002/anie.202103548
Angewandte Chemie International Edition
Introduction
Because most proteins carry an intrinsic net electrostatic charge (Z) between Z = +10 and
-10 (at cytosolic pH), they exert Coulombic forces upon each other.[1] Debye-Hückel theory
predicts that these intermolecular forces—attractive or repulsive per each protein’s sign of net
charge—reach 10 Å through solvent.[2] In the case of a partially desolvated protein undergoing
self-assembly, the Debye length could reach ~ 20-50 Å through its hydrophobic interior, assuming
an interior ionic strength of I = 0.001 and a dielectric constant (ε) between 4 and 20.
Despite the central importance of Z in molecular recognition, protein aggregation[3] and
catalysis,[4] the net charge of a protein is not typically targeted as a druggable parameter in
pharmacology—at least not intentionally. Typically, pharmacons are not designed to function
Coulombically, by adjusting long-range electrostatic forces.[5] Rather, most pharmacons are
thought to function sterically by inserting hydrophobic or dipole-dipole interactions that alter other
classical parameters, including the Km or Vmax of an enzyme; ΔGfold; or the Kd of a protein-protein
interaction. There are, however, some drugs that have high values of net charge and unclear
mechanisms of action (e.g., the century-old drug, suramin[6]). These drugs might function by
altering electrostatic interactions.
The central hypothesis of this study is that small molecules can be designed and
synthesized to selectively increase the net charge of prion-like amyloid assemblies[7] by
magnitudes that inhibit their propagation. The rationale for this hypothesis is based upon: (i) the
general negative correlation between the rate of aggregation of proteins and their magnitude of net
charge;[8-9] (ii) the occurrence of pathogenic missense mutations that accelerate aggregation of the
gene product by diminishing its net charge;[3, 8, 10] and (iii) the use of anhydrides or aryl esters to
post-translationally “supercharge”[9] anionic proteins by non-selective hyperacylation of lysine-
+ [11-12]
NH3 .
2
This article is protected by copyright. All rights reserved.

10.1002/anie.202103548
Angewandte Chemie International Edition
To this end, we designed and synthesized compounds to selectively amplify the net
negative charge of amyloid fibrils of Cu, Zn superoxide dismutase-1 (SOD1)[13] by amyloid-
+
selective acetylation of lysine-ε-NH3 . Misfolded forms of SOD1 (a negatively charged protein[14]
)
cause amyotrophic lateral sclerosis (ALS) by a prion-like mechanism.[15] Previous studies using
“protein charge ladders” and capillary electrophoresis—one of the few methods for measuring the
net charge of folded proteins at pH ≠ pI—found that native wild-type (WT) SOD1 retains a small
negative charge across subcellular pH and Cu/Zn stoichiometry.[14] For example, the net charge of
SOD1 varies from Z = -3.69 to -6.06 per subunit (pH 7.4), depending upon the number of bound
Cu2+ and Zn2+.[14] The holo-enzyme (with 2 Cu2+ and 2 Zn2+ per dimer) remains anionic from pH
5 to 8, bearing a minimal net charge of Z = -1.9 per subunit at lysosomal pH.[14] Several ALS-
linked mutations in SOD1 located at the protein’s surface (e.g., D90A, E100K, N139K, D101N)
are thought to promote SOD1 aggregation by reducing the net charge of SOD1.[10, 16-17]
Compounds were designed to accomplish amyloid-selective acylation by linking an
aspirin-like aryl ester (that is weakly reactive with lysine) to one of three different hetero- or
homocyclic scaffolds that bind amyloid[18-20] (Figure 1). We hypothesized that these
amyloidophilic, aspirin-like compounds would bind amyloid fibrils of SOD1 and selectively
acetylate its lysine, while being too unreactive to acylate residues in native SOD1 (Figure 1). We
chose lysine acylation to supercharge amyloid SOD1 because lysine acylation with non-selective
anhydrides or free aspirin inhibits the prion-like propagation of aggregated SOD1 in vitro and in
+
organotypic spinal cord.[21-22] Lysine-ε-NH3 groups are also involved in specific short-range
interactions between misfolded proteins.[23]
3
This article is protected by copyright. All rights reserved.

10.1002/anie.202103548
Angewandte Chemie International Edition
Results & Discussion
Design and synthesis of “ball and chain” molecular maces
Three different compounds were synthesized to contain a nitrated aryl ester tethered to a
benzothiazole (BT), stilbene (SB), or styrylpyridine (Py) group via a triethylene glycol linker
(Scheme 1). Benzothiazole, stilbene, and styrylpyridine motifs were used as “handles” because
derivatives of these species exhibit general binding affinity for amyloid-like fibrils.[18-20] The
chemical anatomy and putative function of these amyloidophilic aryl esters is analogous to a
medieval “ball and chain” mace: an amyloidophilic “handle”; a reactive aryl ester “ball” that
+
acetylates lysine-ε-NH3 ; and a flexible triethylene glycol “chain” that permits acetylation of
+
lysine-ε-NH3 within ~12 Å of the handle’s binding site (Figure 1). Synthesis of “ball and chain”
maces is outlined in Scheme 1. Synthesis of the BT, SB, or Py handles and chain utilized previously
published procedures.[19-20, 24] The full synthesis is described in Supporting Information.
The “handles” for all three compounds (and “chains” for stilbene and styrylpyridine
derivatives) are similar to three FDA-approved 18F Positron Emission Tomography (PET) agents
for imaging amyloid plaques in Alzheimer’s disease.[18-20] The length of the glycol chain was
chosen to be identical to the glycol chains present in two of these PET agents. A weakly reactive
nitrated aryl ester—a derivative of aspirin[22] —was used as the reactive “ball.” The mild reactivity
of the aryl ester was empirically tuned by conjugation to the glycol and addition of the nitro group
(which we found to be required for acetylation of lysine in water). The weak reactivity of the ester
is intended to minimize off-target acylation of non-amyloid proteins that do not bind the compound
+
(i.e., increased collision frequency between aryl ester and Lys-ε-NH3 in the amyloid-mace
complex will promote acylation, while the free compound will be too unreactive to acylate non-
amyloid proteins). The long half-life of amyloid-like complexes in vivo will allow, we hypothesize,
4
This article is protected by copyright. All rights reserved.

10.1002/anie.202103548
Angewandte Chemie International Edition
for successive binding and acetylation, leading to cumulative acetylation and supercharging of
amyloid-like assemblies (Figure 1B).
In this study, metal-free (apo) wild-type SOD1 was generated by demetallation in EDTA,
with a final metal content of ≤ 0.08 molar equivalents of Cu2+ or Zn2+ per dimer (according to ICP-
MS). This study targeted wild-type SOD1 because: (i) WT SOD1 promotes the toxicity of ALS-
linked mutant SOD1 and is required for pathogenesis in some ALS models (and most ALS-SOD1
mutations are heterozygous);[25-26] and (ii) WT apo-SOD1 forms amyloid fibrils in vitro at rates
identical to several ALS mutant apo-SOD1 proteins.[27]
Ball and chain molecular maces acetylate fibrillar apo-SOD1 but not native holo-SOD1
Reactivity of the molecular mace compounds with fibrillar apo-SOD1 or native holo-SOD1
was tested by incubating 30 µM SOD1 (60 µM SOD1 monomer) with 3 mM compound in a
gyrating microplate for 5 days (37 °C, 3 % DMSO). The effective molar excess of mace per SOD1
monomer is likely lower than 50:1, as compounds are minimally soluble in pure water (similar to
their PET analogs) with precipitation observed upon dilution into protein solution. The predicted
solubility of mace compounds in pure water (25 ℃) ranged from 0.20-0.93 µg/mL; 0.24 to 1.69
µM. Experimentally measured solubility in 97 % H2O/ 3 % DMSO (37 ℃) ranged from 0.18 to
4.6 µM (Figure S1).
The extent of acetylation after incubation was quantified with mass spectrometry (Figure
2) and capillary electrophoresis using UV absorbance at 214 nm (Figures 3, S2). See Supporting
Information for a discussion of the use of mass spectrometry and capillary electrophoresis to
quantify lysine acetylation. After a five-day incubation, fibrillar SOD1 and native holo-SOD1 were
both treated with 1,1,1,3,3,3-hexafluoro-isopropanol (HFIP) (Figure S3). Hexafluoroisopropanol
5
This article is protected by copyright. All rights reserved.

10.1002/anie.202103548
Angewandte Chemie International Edition
was used to defibrillize SOD1 (Figure S3) prior to analysis via mass spectrometry or capillary
electrophoresis because it effectively quenches acetylation by lowering solvent pH to ~ 3.
Native holo-SOD1 remained unacetylated after incubation with all three molecular maces
(Figure 2), however, fibrillar SOD1 was acetylated by each mace. The weighted average number
of acetylated lysine residues in fibrillar apo-SOD1 (per subunit) following incubation with BT,
SB, or Py is as follows: BT, 2.27 acetylated lysine (20.17 % of SOD1 proteins unacetylated); SB,
1.66 (10.52 % unacetylated); Py, 3.98 (0.00 % unacetylated) (Figure 2). These differences in
reactivity between compounds might be caused in part by differences in solubility and/or unique
binding sites on fibrils. No clear correlation was found between solubility and reactivity (Figures
S1, S4). Note the concentration of acetylated lysine in SOD1 (e.g., [3.98 Lys-Ac] × [60 μM SOD1
monomer], in the Py-treated sample) is much larger than the maximum solubility of the Py
compound. Thus, the compound must be partitioning from the colloidal phase to fibrillar SOD1
over the reaction period.
Capillary electrophoresis of acetylated monomers recovered from fibrils demonstrated that
lysine acetylation increased net negative charge by ~ -0.9 units, rather than -1 because of charge
regulation (Figure 3).[2] Charge regulation refers to the adjustment in pKa of ionizable residues, in
response to the new electrostatic environment caused by acetylation.
Note that the BT-treated fibrils have, relative to SB-treated fibrils, both a higher weighted
average number of acetyl modifications and a higher percentage of unmodified protein (Figure 2).
This paradox is caused by a greater degree of positive skewness in the distribution of acetyl peaks
for the BT sample (reaching 8 acetyl modifications) than the SB sample (reaching only 4 acetyl
modifications). This positive skewness might suggest cooperativity in acetylation of fibrils by BT
(e.g., perhaps acetylation promotes partitioning of compound to fibril).
6
This article is protected by copyright. All rights reserved.

10.1002/anie.202103548
Angewandte Chemie International Edition
To determine if maces exhibited high reactivity with other fibrillar proteins, we incubated
each mace with fibrillar α-synuclein. Alpha-synuclein has 14 lysine and a formal net charge of Z
= -9.0 (the form we used lacked N-terminal acetylation). After a five day incubation with each
compound at a ratio of ~ 43:1 (compound:protein), the weighted average number of acetylations
is as follows; BT, 10.22; SB, 7.83; Py, 7.11 with no unacetylated protein remaining (Figure S5).
Thus, as far as the percent of lysine that were acetylated, maces were most reactive with fibrillar
α-synuclein compared to fibrillar SOD1 (or any native protein in this study), doubling the formal
net charge up to Z = -18. We did not perform a five day control experiment with native α-synuclein
(an intrinsically disordered protein[28]) as it readily fibrillizes.
The exact residues acetylated by the molecular mace compounds were identified via
sequential trypsin and pepsin proteolysis and liquid chromatography tandem mass spectrometry
(LC-MSE) (Figure 2, S6, S7). The sequence coverage for each solution of mace-acetylated SOD1
fibrils was 100 % and the identity of acetylated residues was based upon manual inspection of
mass spectra and the following threshold criteria: a peptide mass error < 6 ppm; fragment root
mean square mass error < 30 ppm; signal to noise ratio > 3.
Tandem mass spectrometry suggested that lysine residues were selectively acetylated over
other nucleophilic residues in amyloid-SOD1 (Figure 2, S6). The N-terminus of human SOD1
protein in this study is properly acetylated by the eukaryotic host expression system. A comparison
of the MS/MS and MS spectra suggests that each molecular mace does not exhibit selectivity to a
single lysine residue in fibrillar SOD1 (Figure 2, S6, S7). All lysine residues were acetylated by
each molecular mace with the exception of: (i) Lys 128 when reacted with BT, and (ii) Lys 3 when
reacted with SB (Figure 2, S6). Acetylation of multiple lysine in fibrillar SOD1 is presumably
permitted by the flexible glycol linker in combination with the multiple binding sites on each
7
This article is protected by copyright. All rights reserved.

10.1002/anie.202103548
Angewandte Chemie International Edition
fibril.[29-31] Although we present mass reconstructs of acetylated and unacetylated fragments
including each lysine (Figure S6), we reiterate that the intensity of mass spectral signals cannot be
used to quantify acetylation because: (i) acetylation of lysine can lower ionization efficiency of
small peptides, and (ii) alter proteolysis sites of trypsin or pepsin.
Capillary electrophoresis was used to estimate the electrostatic effects of each compound
on fibrillar SOD1 (Figure 3). The capillary electropherograms of defibrillized SOD1 are generally
superimposable with the mass spectra in Figure 2. According to the increased mobility of
acetylated SOD1, acetylation increased the magnitude of Z of each SOD1 polypeptide from Z = -
6.92 for unacetylated apo-SOD1 (previously determined)[14] to between Z = -7.80 and -13.96 for
BT (weighted average: Z = -8.92); between Z = -7.86 and -10.68 for SB (weighted average: Z = -
8.48); and between Z = -7.87 and -15.47 for Py (weighted average: Z = -10.70) (Figure 3). In a
fibril consisting of 103 peptides, this amplification of Z is equivalent to ∆Zave = - 2,400 units/fibril
and ∆Zmax = -8,100 units/fibril.
Molecular maces were tested for reactivity against other abundant, natively folded proteins
including holo-myoglobin (equine heart, 19 lysine), holo-α-hemoglobin (bovine blood, 13 lysine),
holo-cytochrome c (equine heart, 19 lysine), α-lactalbumin (bovine, 12 lysine), carbonic anhydrase
II (bovine, 18 lysine), ubiquitin (bovine, 6 lysine), and serum albumin (human, 60 lysine). These
proteins were chosen because: (i) each does not form amyloid or aggregate under conditions used
+
for fibrillization of apo-SOD1 (Figure S8); (ii) each contains an unacetylated N-terminal α-NH3
+
(which is generally more reactive with aryl esters than lysine-ε-NH3 ); (iii) each (except for
ubiquitin) contains more lysine than SOD1; and (iv) the net charge of some of these proteins are
either net positive or less negative than SOD1, suggesting that their lysine residues have lower pKa
8
This article is protected by copyright. All rights reserved.

10.1002/anie.202103548
Angewandte Chemie International Edition
(and are thus more reactive with aryl esters). Thus, these proteins are good controls for off-target
acetylation.
Following incubation in a gyrating microplate (360 rpm; 37 °C) for five days, acetylation
of each protein was determined via LC-MS (Figure S9) and verified with capillary electrophoresis
(see Supporting Information). The weighted average acetylation of each protein varied between
0.10 acetyl groups for carbonic anhydrase II and 0.14 acetyl groups for cytochrome c and
myoglobin, up to 2.69 acetyl groups for hemoglobin (Figure S9).
Serum albumin was the most acetylated native protein that we studied and is a special case.
We could not integrate all ions to determine accurate weighted averages of numbers of acetylated
lysines because the mass spectra of HSA are reportedly difficult to deconvolute due to persistent
impurities.[32] We could, however, calculate the number of acetylated lysine on the most abundant
ions. Of the 60 lysine in HSA, we estimate that BT acetylated between 3-34 Lys (median ~ 15
Lys), SB acetylated 1-43 Lys (median ~ 15 Lys), and Py acetylated 12-45 Lys (median ~25 Lys).
Unacetylated HSA was not detected. This high degree of acetylation is not particularly surprising
considering its high affinity for drug-like molecules including stilbenes and benzthiazoles.[33-34]
This result offers more evidence to support the hypothesis that binding of the mace promotes
acylation of lysine. The identity of acetylated residues was determined by tandem mass
spectrometry (Figure S10, Tables S1-S7), with sequence coverage ranging from 86 to 100 %.
+
Tandem mass spectrometry suggested that maces react with N-terminal α-NH3 in Mb,
CAII, Ubq, and α-Lac however, we did not detect N-terminal acetylation in Cyt c or Hb (detecting
N-terminal fragments with LC-MS and MS/MS can be challenging due to their small size).
Acetylated lysine residues were detected in all control proteins (Figure S10, Tables S1-S7). We
cannot entirely rule out that maces are binding these heme proteins specifically, however, we
9
This article is protected by copyright. All rights reserved.

10.1002/anie.202103548
Angewandte Chemie International Edition
suspect that these proteins are more reactive than native holo-SOD1 because each contains an N-
+
terminal α-NH3 , more lysine residues than SOD1, and are less anionic/more cationic than SOD1.
In vitro seeding activity of SOD1 parent fibrils acetylated by molecular mace compounds
The ability of supercharged and normally charged apo-SOD1 (formed in the presence of
BT, SB, Py, or a DMSO control) to recruit soluble apo-SOD1 into the fibrillar state (i.e., to seed
aggregation) was measured with Thioflavin-T (ThT) fluorescence (Figure 4). We examined how
acetylation by molecular maces altered the seeded fibrillization of SOD1 (rather than of unseeded
fibrillization) for two reasons. First, seeded assays using “parent” fibrils permit us to test how
molecular maces affect the prion-like propagation of fibrils in vitro, which can be compared to
seeding assays in organotypic mouse spinal cord. Second, the unseeded fibrillization of SOD1 is
highly stochastic, with lag times ranging from ~ 5 to > 20 hours for identical solutions of SOD1.[35]
Seeded fibrillization of native apo-SOD1 with pre-formed “parent” fibrils (consisting of apo-
SOD1) is less stochastic and statistically significant kinetic values can be easily obtained.[36]
The seeded fibrillization of SOD1 with untreated (unacetylated) SOD1 parent fibrils
(denoted DMSO control) exhibited exponential increases in ThT fluorescence without a calculable
lag time (as expected when by-passing primary nucleation by addition of an amyloid seed[37]).
Fitting of data to exponential functions was performed because sigmoidal fits resulted in illogical
kinetic parameters, i.e., infinite error for any parameter or initial or maximum fluorescence values
that were orders of magnitude above (i.e., 104-fold larger) what is observed. Parent fibrils treated
(acetylated) with the Py molecular mace compound exhibited a diminished ability to seed the
fibrillization of SOD1, as illustrated by an increase in lag time of more than four hours (p ≤ 0.0001)
and two-fold decrease in propagation rate (p ≤ 0.0001) compared to fibrils grown in the presence
10
This article is protected by copyright. All rights reserved.

10.1002/anie.202103548
Angewandte Chemie International Edition
of DMSO treated seeds (Figure 4). Treating fibrils with BT and SB molecular mace compounds
resulted in weaker inhibitory effects on seeding compared to Py treated fibrils. Note that while the
BT and SB compound did increase lag time compared to the DMSO control (p ≤ 0.0001) the SB
compound did not affect propagation rate compared to the DMSO control (p = 0.5563, Figure 4E).
These data might be rationalized by the fewer numbers of lysine residues acetylated by BT and SB
(Figure 4).
To determine how maces might affect seeding activity of fibrillar SOD1, independent of
lysine acetylation, “whip” compounds were synthesized that contained the handle and glycol chain
but lacked the reactive nitrated aryl ester (Figure 4). Fibrils that were treated with unreactive
“whips” (denoted BTOH, SBOH, PyOH) did not produce increased lag times of fibrillization in
assays of seeded fibrillization (Figure 4D). The BTOH and SBOH whips were slightly more
effective at inhibiting fibril elongation than their corresponding maces. This effect might be due
to the intrinsic inhibitory effects of the handle on elongation, combined with the higher solubility
of whips compared to maces (whips are predicted to be 100-fold more soluble in water than
maces). It has been shown that high concentrations of amyloid-binding compounds, such as
Thioflavin-T, can slow the fibrillization of certain proteins, however, the effects on SOD1 are
unknown.[38]
Is acetylation—or more generally, the molecular mace—inhibiting fibrillization by a
purely electrostatic mechanism? According to our data, probably not. Previous studies examined
how SOD1 fibrillization is affected by acylation with different non-specific anhydrides that attach
groups of variable charge, flexibility, and hydrophobicity.[21-22] These studies suggest that the
inhibitory effect of lysine acylation on SOD1 fibrillization is largely electrostatic in nature.[21-22]
Previous reports analyzed how the acetylation of soluble SOD1 with aspirin affected the nucleation
11
This article is protected by copyright. All rights reserved.

10.1002/anie.202103548
Angewandte Chemie International Edition
and elongation of amyloid-like fibrils.[22] These prior experiments involved the free aryl methyl
ester (aspirin), unattached to an amyloid-binding moiety, and found that acetylation of ~ 3 lysine
per SOD1 subunit increased lag time by 2.6 hours (in unseeded assays); acetylation of ~ 6 lysine
increased lag time by 11 hours.[22] Despite these results, we cannot rule out inhibitory effects from
increased polydispersity, caused by myriad regioisomeric forms of acetylated SOD1 and from the
binding of the compound itself (as demonstrated by control studies with whips).
Control experiments were performed to determine how acetylation of lysine by the free
aryl ester “ball” affected the seeding properties of fibrillar SOD1. The free aryl ester acetylated
lysine in holo-SOD1, fibrillar apo-SOD1, and in native control proteins (Figure S11). Addition of
the free aryl ester to SOD1 produced a weighted average of 6.51 acetyl groups per subunit for
fibrillar apo-SOD1 and 7.28 acetyl groups per subunit of native holo-SOD1 (unacetylated protein
= 0 % for native and fibrillar SOD1; Figure S11). The increased reactivity of the free “ball”
compared to the mace is attributed to the greater intrinsic reactivity of the untethered aryl ester
(fewer steric constraints during nucleophilic attack by lysine side chain). The SOD1 fibrils that
were acetylated by the free aryl ester exhibited weak seeding activity, with a lag time of 21.9 ± 1.4
-1
hours and a propagation rate of 3.7 ± 0.1 hr (Figure 4). These rates are slower than those of
seeded fibrillization with unacetylated or mace-acetylated amyloid seeds (Figure 4) and are
comparable to rates of unseeded fibrillization.[36] This long lag time suggests that the extensive
acetylation by the free ester inhibited the formation of a competent amyloid seed.
Seeding of ALS-linked G85R SOD1 by supercharged SOD1 fibrils in organotypic spinal cord
We compared the prion-like activity of supercharged SOD1 fibrils and unacetylated SOD1
fibrils in organotypic murine spinal cord. The spinal cord cultures were derived from transgenic
12
This article is protected by copyright. All rights reserved.

10.1002/anie.202103548
Angewandte Chemie International Edition
mice expressing human ALS-variant SOD1 fused to yellow fluorescent protein (G85R-
SOD1:YFP). This transgenic line has been shown to be a valid ex vivo assay for prion-like seeding
activity of SOD1 amyloid fibrils that are produced exogenously.[39] The expression level of mutant
SOD1:YFP in these transgenic animals is too low to produce disease or SOD1:YFP inclusion
pathology.[40] Rather, large inclusions of endogenously expressed ALS-mutant SOD1:YFP
(micron-scale) are only observed when exogenous SOD1 fibrils are injected into culture media (or
when spinal cord extract from a symptomatic ALS-SOD1 Tg mouse is added).[40] Currently, we
do not know the mechanism(s) by which exogenously added SOD1 fibrils interact with (and seed
aggregation of) endogenously expressed G85R SOD1:YFP in organotypic spinal cord. A prion-
like mechanism involving endocytosis of misfolded WT SOD1 is possible.[15]
The prion-like seeding activity of acetylated SOD1 fibrils (produced in vitro and treated
with molecular mace compounds) was monitored by adding fibrils directly to cultured cross-
sectional slabs of G85R-SOD1:YFP spinal cord (Figure 5A-F). As a positive control, seeds from
spinal cord homogenate (of an ALS- symptomatic G93A SOD1 mouse) were added to G85R-
SOD1:YFP slabs, resulting in extensive inclusion pathology (Figure 5A). Experiments involving
mice were approved by the University of Florida Institutional Animal Care and Use Committee
(IACUC).
The addition of unacetylated SOD1 fibrils from in vitro preparations also produced
multiple inclusions (Figure 5C), as previously observed[21] for unacetylated fibrils of WT SOD1.
The addition of SOD1 fibrils treated with BT, SB, or Py (identical to those fibrils used to measure
in vitro seeding activity) had variable effects on ex vivo seeding activity in spinal cord (Figures
5D-F, S12A-C). Consistent with in vitro seeding data (i.e., effect of compounds on lag time or
fibril propagation rate), the most pronounced inhibitory effects on ex vivo seeding activity were
13
This article is protected by copyright. All rights reserved.

10.1002/anie.202103548
Angewandte Chemie International Edition
observed for fibrils treated with the Py molecular mace, and with BTOH and SBOH whips (Figures
5G-H). Additional replicates and images of spinal cord slices incubated with SOD1 seeds treated
with whip compounds can be found in the supporting information (Figure S12).
The addition of spinal cord homogenate from ALS G93A mice (i.e., “control” seeds)
produced 285 ± 172 inclusions per cultured spinal cord slab (Figure 5G). Addition of unacetylated
fibrillar SOD1 to 12 slabs (labelled “DMSO” in Figure 5) produced 90 ± 48 inclusions. The
addition of mace-treated acetylated fibrils resulted in < 50 inclusions per slab (12 slabs), with the
Py-treated fibrils producing the lowest value of 23 ± 14 inclusions per slab. The effect of mace-
treated fibrils on the formation of inclusions in spinal cord slabs correlated with seeding activity
of in vitro fibrillization assays (Figure 5H). In both experiments, the Py molecular mace was the
most effective at inhibiting aggregation. Note that the Py mace was more effective than the Py
whip at inhibiting inclusion formation, whereas the inhibitory effects of other maces were
statistically similar to their corresponding whip (Figure 5H, Figure S12). Moreover, the BTOH
and SBOH whips were as effective as the Py mace at inhibiting inclusion formation (Figure 5H).
These results mirror the comparative effects of whips and maces on fibril elongation rates in vitro
(Figure 4E, Figure 5H). We hypothesize that BTOH and SBOH whips inhibited inclusion
formation (as equally well as the Py compound) due to their possible inhibitory effects on fibril
elongation[38] and 100-fold greater solubility (predicted) compared to mace compounds.
Cytotoxicity of Molecular Maces
The intrinsic cytotoxicity of each molecular mace and the free aryl ester “ball” were
estimated via an MTT (3-[4,5-dimethylthiazole-2-yl]-2,5-diphenyltetrazolium bromide) viability
assay using CHO-K1 cells (ATCC CCL-61). Cells were incubated with each compound at
14
This article is protected by copyright. All rights reserved.

10.1002/anie.202103548
Angewandte Chemie International Edition
concentrations ranging from 0.1 to 100 µM in 5% DMSO. No decrease in cell viability was
observed. (Figure S13). A caveat of this result is that compounds are only immediately soluble up
to ~ 4 µM (Figure S13). The free aryl ester “ball” is, however, highly soluble and did not lead to
a decrease in cell viability up to 100 µM.
“Drugging Z”
The “ball and chain” molecular maces described in this study represent, to our knowledge,
the first example of compounds that selectively modify a protein (covalently) in its amyloid-like
state, without reacting with lysine in the native state of the same protein. The potential utility of
molecular maces as lead compounds, and their shortcomings, are discussed in Supporting
Information. While this proof of concept study designed molecular maces to transfer one of the
simplest and least potent (in terms of ∆Z) types of acyl groups (R-C=OCH3), this strategy is not
limited to acetylation. Future ball and chain maces can include non-biological acyl groups that
impart more than a single unit of charge (and are not substrates for de-acylation[41]). This study
suggests that any useful “supercharging” compound would require a more electrostatically potent
group than acetyl, as acetylation did not greatly reduce aggregation in all cases, compared to the
corresponding whip. The molecular mace compounds described in this study also provide a means
13
of covalently attaching probes (including C and 18O) to lysine residues in amyloid-like species
for in vivo tracking, imaging[42] or proteomic analysis. The structural similarity between “ball and
chain” molecular maces and FDA approved PET agents for amyloid plaques of Aβ1-42 also
suggests that molecular maces will be capable of supercharging other misfolded prion-like proteins
besides SOD1 and α-synuclein.[43]
15
This article is protected by copyright. All rights reserved.

10.1002/anie.202103548
Angewandte Chemie International Edition
Acknowledgement
We would like to thank the Mass Spectrometry Center, Molecular Bioscience Center, and the
Center for Microscopy and Imaging at Baylor University for use of facilities and instrumentation.
Funding
This research was supported by grants from the National Science Foundation (CHE: 1856449,
CHE: 1352122), the Welch Foundation (AA-1854), and the National Institutes of Health
(1R01NS092788).
16
This article is protected by copyright. All rights reserved.

10.1002/anie.202103548
Angewandte Chemie International Edition
Figure 1. Amplifying the net charge (Z) of misfolded SOD1 by amyloid-selective lysine
acetylation. (A) Structures of the three “ball and chain” molecular mace compounds synthesized
+
in this paper. (B) Acetylation (neutralization) of Lys-ε-NH3 will increase the magnitude of net
negative charge of amyloid-like assemblies. In this illustration, the rate of acylation (kAc) is slower
than the rate of binding (kb).
17
This article is protected by copyright. All rights reserved.

10.1002/anie.202103548
Angewandte Chemie International Edition
Scheme 1. Synthesis of BT, SB, and Py molecular maces. Reagents and conditions: (i) K2CO3,
DMF, 180 °C, 16 hours (ii) 2-nitroresorcinol, PPh3, DIAD, THF, 24 hours (iii) Et3N, AcCl, DCM,
~ 12 hours. See supplementary note for 1H and 13C NMR spectra.
18
This article is protected by copyright. All rights reserved.

10.1002/anie.202103548
Angewandte Chemie International Edition
Figure 2. Reactivity grid for acetylation of native or fibrillar WT SOD1 (vertical axis) by BT,
SB, or Py molecular maces (horizontal axis). Mass spectra of native WT SOD1 (top) and
(defibrillized) amyloid WT SOD1 (bottom) following 5-day incubation with BT, SB, or Py
molecular maces. Solutions of fibrillar and native SOD1 were solubilized with HFIP prior to mass
spectrometric analysis. The number of acetyl groups per monomer corresponding to each peak is
shown as Ac(n) (weighted average (WAVG) number of acetyl modifications per monomer). The
amount of protein remaining unacetylated (% Ac(0)) is noted in the bottom right of each panel. In
cases where acetylation was not observed (i.e., native WT SOD1 incubated with any of the three
molecular maces), the mass at which acetylated SOD1 would have appeared is indicated by “+42
Da”. The measured net charge of an SOD1 monomer with the respective number of acetyl
modifications is shown in red (determined via capillary electrophoresis). Transmission electron
micrographs of fibrillar or native WT SOD1 after reaction with each molecular mace are shown
below mass spectra. Scale bars = 100 nm. The SOD1 sequence illustrates residues that were
acetylated by molecular maces (•).
19
This article is protected by copyright. All rights reserved.

10.1002/anie.202103548
Angewandte Chemie International Edition
Figure 3. Quantifying effect of lysine acetylation on net charge (Z) of fibrillar SOD1 with
capillary zone electrophoresis. Capillary electropherograms of defibrillized apo-SOD1 following
incubation with (A) DMSO and (B-D) “ball and chain” molecular maces. Linear plots of
electrophoretic mobility of SOD1 chain vs. number of acetyl modifications (n) are shown to right
of each electropherogram (∆ZAc = 0.88-0.95 units per lysine acetylation). The value Zmono refers to
the net charge of each SOD1 polypeptide containing the weighted average number of acetylations
(inferred from MS). The value ∆Zfib refers to the corresponding increase in Z for an aggregate
composed of 103 SOD1 chains. Capillary electrophoresis was performed in triplicate. Note that the
satellite peak in A (at 7.5 mobility units) has been previously assigned as deamidation of N26 to
aspartate.[44]
20
This article is protected by copyright. All rights reserved.

10.1002/anie.202103548
Angewandte Chemie International Edition
Figure 4. Effect of “ball and chain” maces (BT, SB, Py) and corresponding “whips”
(BTOH, SBOH, PyOH) on seeding activity of fibrillar SOD1 in vitro. (A) Average,
normalized ThT fluorescence of apo WT-SOD1 aggregation seeded by fibrils grown in the
presence of “ball and chain” maces, “whips” (lacking aryl ester “ball”), “ball” (lacking handle
and chain) and a 3% DMSO control (n = 7 per trace). P values represent comparison of lag time
for ball and chain mace and corresponding whip. (B) Raw ThT fluorescence data used to produce
(A). Black trace represents the average ThT fluorescence of seven replicate measurements.
(C)Transmission electron micrographs of aggregates formed during ThT fluorescence assays;
scale bar = 100 nm. (D), Mean lag time, (E) propagation rate, and (F) maximum ThT
fluorescence of aggregation assays shown in (A-B). Lag times for DMSO, BTOH, SBOH, and
PyOH = 0 as aggregation curve is best fit to an exponential function. Error bars represent the
standard error of the mean; p values were calculated using an unpaired Student’s t-test.
21
This article is protected by copyright. All rights reserved.

10.1002/anie.202103548
Angewandte Chemie International Edition
Figure 5. Effect of BT, SB, and Py molecular maces on ability of SOD1 fibrils to induce SOD1
inclusion pathology in organotypic murine spinal cord. Confocal micrographs of cultured
G85R-SOD1:YFP spinal cord slab after: (A) addition of spinal cord homogenate of symptomatic
G93A-SOD1 mice, (B) no added sample (denoted Neg Ctrl) or (C-F) addition of SOD1 fibrils
(formed in vitro) that were treated with either DMSO (blank), or BT, SB, Py molecular mace
molecules. Scale bars represent 40 µm. Four of 15 biological replicates are shown for each
experiment. (G) Average number of inclusions per tissue slab. P values were calculated from an
unpaired Student’s t-test. (H) Comparison of seeding propensity as measured by in vitro
propagation rate and ex vivo inclusions count. All error bars represent the standard error of the
mean. See Supporting Information for additional replicates of each experiment and images of
tissue treated with whip analogues. Spinal cords were harvested from non-symptomatic G85R-
SOD1:YFP transgenic mice.
22
This article is protected by copyright. All rights reserved.

10.1002/anie.202103548
Angewandte Chemie International Edition
References
[1]
C. E. Felder, J. Prilusky, I. Silman, J. L. Sussman, Nucleic Acids Res. 2007, 35, W512-
521.
[2]
I. Gitlin, J. D. Carbeck, G. M. Whitesides, Angew. Chem. Int. Ed. Engl. 2006, 45, 3022-
3060.
[3]
[4]
[5]
F. Chiti, M. Stefani, N. Taddei, G. Ramponi, C. M. Dobson, Nature 2003, 424, 805-808.
S. D. Fried, S. Bagchi, S. G. Boxer, Science 2014, 346, 1510-1514.
M. Y. Tsai, W. Zheng, D. Balamurugan, N. P. Schafer, B. L. Kim, M. S. Cheung, P. G.
Wolynes, Protein Sci. 2016, 25, 255-269.
[6]
[7]
[8]
N. Wiedemar, D. A. Hauser, P. Maser, Antimicrob. Agents Chemother. 2020, 64.
G. Meisl, T. P. Knowles, D. Klenerman, Curr. Opin. Neurobiol. 2020, 61, 58-64.
F. Chiti, M. Calamai, N. Taddei, M. Stefani, G. Ramponi, C. M. Dobson, Proc. Natl.
Acad. Sci. U.S.A. 2002, 99 Suppl 4, 16419-16426.
[9]
M. S. Lawrence, K. J. Phillips, D. R. Liu, J. Am. Chem. Soc. 2007, 129, 10110-10112.
[10] E. Sandelin, A. Nordlund, P. M. Andersen, S. S. Marklund, M. Oliveberg, J. Biol. Chem.
2007, 282, 21230-21236.
[11] B. F. Shaw, H. Arthanari, M. Narovlyansky, A. Durazo, D. P. Frueh, M. P. Pollastri, A.
Lee, B. Bilgicer, S. P. Gygi, G. Wagner, G. M. Whitesides, J. Am. Chem. Soc. 2010, 132,
17411-17425.
[12] B. F. Shaw, G. F. Schneider, B. Bilgicer, G. K. Kaufman, J. M. Neveu, W. S. Lane, J. P.
Whitelegge, G. M. Whitesides, Protein Sci. 2008, 17, 1446-1455.
[13] M. Chattopadhyay, A. Durazo, S. H. Sohn, C. D. Strong, E. B. Gralla, J. P. Whitelegge, J.
S. Valentine, Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 18663-18668.
[14] Y. Shi, R. A. Mowery, B. F. Shaw, J. Mol. Biol. 2013, 425, 4388-4404.
[15] C. Munch, J. O'Brien, A. Bertolotti, Proc. Natl. Acad. Sci. U.S.A. 2011, 108, 3548-3553.
[16] J. A. Rodriguez, B. F. Shaw, A. Durazo, S. H. Sohn, P. A. Doucette, A. M. Nersissian, K.
F. Faull, D. K. Eggers, A. Tiwari, L. J. Hayward, J. S. Valentine, Proc. Natl. Acad. Sci.
U.S.A. 2005, 102, 10516-10521.
[17] B. G. Trist, J. B. Hilton, D. J. Hare, P. J. Crouch, K. L. Double, Angew. Chem. Int. Ed.
Engl. 2021, 60, 9215-9246.
[18] N. George, E. Gean, A. Nandi, J. R. Brašić, D. F. Wong, in Imaging of the Human Brain
in Health and Disease, Academic Press, Boston, 2014, pp. 407-416.
[19] W. Zhang, S. Oya, M. P. Kung, C. Hou, D. L. Maier, H. F. Kung, J. Med. Chem. 2005,
48, 5980-5988.
[20] W. Zhang, M. P. Kung, S. Oya, C. Hou, H. F. Kung, Nucl. Med. Biol. 2007, 34, 89-97.
[21] S. Rasouli, A. Abdolvahabi, C. M. Croom, D. L. Plewman, Y. Shi, J. I. Ayers, B. F.
Shaw, J. Biol. Chem. 2017, 292, 19366-19380.
[22] A. Abdolvahabi, Y. Shi, N. R. Rhodes, N. P. Cook, A. A. Marti, B. F. Shaw, Biophys. J.
2015, 108, 1199-1212.
[23] S. Sinha, D. H. Lopes, Z. Du, E. S. Pang, A. Shanmugam, A. Lomakin, P. Talbiersky, A.
Tennstaedt, K. McDaniel, R. Bakshi, P. Y. Kuo, M. Ehrmann, G. B. Benedek, J. A. Loo,
F. G. Klarner, T. Schrader, C. Wang, G. Bitan, J. Am. Chem. Soc. 2011, 133, 16958-
16969.
[24] J. Y. Kim, J. W. Lee, Y. S. Kim, Y. Lee, Y. B. Ryu, S. Kim, H. W. Ryu, M. J. Curtis-
Long, K. W. Lee, W. S. Lee, K. H. Park, Chembiochem 2010, 11, 2125-2131.
23
This article is protected by copyright. All rights reserved.

10.1002/anie.202103548
Angewandte Chemie International Edition
[25] A. Weichert, A. S. Besemer, M. Liebl, N. Hellmann, I. Koziollek-Drechsler, P. Ip, H.
Decker, J. Robertson, A. Chakrabartty, C. Behl, A. M. Clement, Neruobiol. Dis. 2014,
62, 479-488.
[26] G. Xu, J. I. Ayers, B. L. Roberts, H. Brown, S. Fromholt, C. Green, D. R. Borchelt, Hum.
Mol. Genet. 2015, 24, 1019-1035.
[27] A. Abdolvahabi, Y. Shi, S. Rasouli, C. M. Croom, A. Aliyan, A. A. Marti, B. F. Shaw,
ACS Chem. Neurosci. 2017, 8, 1378-1389.
[28] B. A. Killinger, R. Melki, P. Brundin, J. H. Kordower, NPJ Parkinson's Dis. 2019, 5.
[29] M. P. Kung, C. Hou, Z. P. Zhuang, D. Skovronsky, H. F. Kung, Brain Res. 2004, 1025,
98-105.
[30] H. F. Kung, C. W. Lee, Z. P. Zhuang, M. P. Kung, C. Hou, K. Plossl, J. Am. Chem. Soc.
2001, 123, 12740-12741.
[31] H. Watanabe, M. Ono, T. Ariyoshi, R. Katayanagi, H. Saji, ACS Chem. Neurosci. 2017,
8, 1656-1662.
[32] J. W. McCabe, C. S. Mallis, K. I. Kocurek, M. L. Poltash, M. Shirzadeh, M. J. Hebert, L.
Fan, T. E. Walker, X. Zheng, T. Jiang, S. Dong, C. W. Lin, A. Laganowsky, D. H.
Russell, Anal. Chem. 2020, 92, 11155-11163.
[33] P. Bourassa, C. D. Kanakis, P. Tarantilis, M. G. Pollissiou, H. A. Tajmir-Riahi, J. Phys.
Chem. B 2010, 114, 3348-3354.
[34] P. Sen, S. Fatima, B. Ahmad, R. H. Khan, Spectrochim. Acta A Mol. Biomol. Spectrosc.
2009, 74, 94-99.
[35] A. Abdolvahabi, Y. Shi, A. Chuprin, S. Rasouli, B. F. Shaw, ACS Chem. Neurosci. 2016,
7, 799-810.
[36] K. M. Baumer, J. C. Koone, B. F. Shaw, ACS Chem. Neurosci. 2020, 11, 304-313.
[37] L. Giehm, D. E. Otzen, Anal. Biochem. 2010, 400, 270-281.
[38] C. Xue, T. Y. Lin, D. Chang, Z. Guo, R. Soc. Open Sci. 2017, 4, 160696.
[39] J. I. Ayers, J. Diamond, A. Sari, S. Fromholt, A. Galaleldeen, L. W. Ostrow, J. D. Glass,
P. J. Hart, D. R. Borchelt, Acta Neuropathol. 2016, 132, 827-840.
[40] J. I. Ayers, S. Fromholt, M. Koch, A. DeBosier, B. McMahon, G. Xu, D. R. Borchelt,
Acta Neuropathol. 2014, 128, 791-803.
[41] C. M. Grozinger, S. L. Schreiber, Chem. Biol. 2002, 9, 3-16.
[42] L. M. Needham, J. Weber, J. W. B. Fyfe, O. M. Kabia, D. T. Do, E. Klimont, Y. Zhang,
M. Rodrigues, C. M. Dobson, S. Ghandi, S. E. Bohndiek, T. N. Snaddon, S. F. Lee, R.
Soc. Open. Sci. 2018, 5, 171399.
[43] A. W. Pilkington, J. Schupp, M. Nyman, S. J. Valentine, D. M. Smith, J. Legleiter, ACS
Chem. Neurosci. 2020, 11, 146-161.
[44] Y. Shi, N. R. Rhodes, A. Abdolvahabi, T. Kohn, N. P. Cook, A. A. Marti, B. F. Shaw, J.
Am. Chem. Soc. 2013, 135, 15897-15908.
24
This article is protected by copyright. All rights reserved.

10.1002/anie.202103548
Angewandte Chemie International Edition
Table of Contents
Lead compounds were designed to “supercharge” misfolded forms of proteins linked to
amyotrophic lateral sclerosis and Parkinson’s disease (i.e., superoxide dismutase-1; α-synuclein).
Compounds increased the net negative charge of amyloid fibrils via lysine peracetylation, with
minimal to no modification of various proteins in their native state. Supercharged fibrils exhibited
weaker seeding activity in vitro and in organotypic spinal cord.
25
This article is protected by copyright. All rights reserved.