The Hairpin Form of r(G4C2)exp in c9ALS/FTD Is Repeat-Associated Non-ATG Translated and a Target for Bioactive Small Molecules

Article abstract of DOI:10.1016/j.chembiol.2018.10.018

The most common genetic cause of amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD) is an expanded G₄C₂ repeat [(G₄C₂)^exp] in C9ORF72. ALS/FTD-associated toxicity has been traced to

Full text of DOI:10.1016/j.chembiol.2018.10.018

Article
The Hairpin Form of r(G4C2)exp in c9ALS/FTD Is
Repeat-Associated Non-ATG Translated and a
Target for Bioactive Small Molecules
Graphical Abstract
Authors
Zi-Fu Wang, Andrei Ursu,
Jessica L. Childs-Disney, ...,
Joseph E. Rice, Leonard Petrucelli,
Matthew D. Disney
Correspondence
disney@scripps.edu
In Brief
The most common cause of ALS is an
expanded RNA repeat [r(G4C2)exp] that
folds into two forms in vitro, a
G-quadruplex and a hairpin. Wang et al.
show that the hairpin form is present in
cells, undergoes aberrant translation that
causes toxicity, and thus is a target for
therapeutic development.
Highlights
d
d
d
An expanded RNA repeat that causes ALS folds into two
forms in cells
The hairpin, not quadruplex, form of the ALS RNA undergoes
RAN translation
Studies were enabled by development of a chemical probe
selective for the hairpin
Wang et al., 2019, Cell Chemical Biology 26, 1–12
February 21, 2019 ª 2018 Elsevier Ltd.
https://doi.org/10.1016/j.chembiol.2018.10.018

Please cite this article in press as: Wang et al., The Hairpin Form of r(G4C2)exp in c9ALS/FTD Is Repeat-Associated Non-ATG Translated and a Target
for Bioactive Small Molecules, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.10.018
Cell Chemical Biology
Article
The Hairpin Form of r(G4C2)exp in c9ALS/FTD
Is Repeat-Associated Non-ATG Translated
and a Target for Bioactive Small Molecules
Zi-Fu Wang,1 Andrei Ursu,1 Jessica L. Childs-Disney,1 Rea Guertler,1 Wang-Yong Yang,1 Viachaslau Bernat,1
Suzanne G. Rzuczek,1 Rita Fuerst,1 Yong-Jie Zhang,2 Tania F. Gendron,2 Ilyas Yildirim,3 Brendan G. Dwyer,1
Joseph E. Rice,4 Leonard Petrucelli,2 and Matthew D. Disney1,5,
*
1Departments of Chemistry and Neuroscience, The Scripps Research Institute, 130 Scripps Way, Jupiter, FL 33458, USA
2Department of Neuroscience, Mayo Clinic, 4500 San Pablo Road, Jacksonville, FL 32224, USA
3Department of Chemistry and Biochemistry, Florida Atlantic University, Jupiter, FL 33458, USA
4Department of Medicinal Chemistry, Ernest Mario School of Pharmacy, Rutgers, The State University of New Jersey,
160 Frelinghuysen Road, Piscataway, NJ 08854, USA
5Lead Contact
*Correspondence: disney@scripps.edu
https://doi.org/10.1016/j.chembiol.2018.10.018
SUMMARY
et al., 2011). The transcribed repeat RNA [r(G4C2)exp] is involved
in two putative pathological mechanisms: (1) sequestering pro-
The most common genetic cause of amyotrophic teins involved in RNA biogenesis to form toxic r(G4C2)exp-protein
foci (Prudencio et al., 2015); and (2) undergoing repeat-associ-
ated non-ATG (RAN) translation to produce toxic c9RAN dipep-
lateral sclerosis (ALS) and frontotemporal dementia
(FTD) is an expanded G4C2 repeat [(G4C2)exp] in
tide repeats (DPRs) in cellular (Almeida et al., 2013; Taylor
et al., 2016; Zu et al., 2013), mouse (Chew et al., 2015; O’Rourke
C9ORF72. ALS/FTD-associated toxicity has been
traced to the RNA transcribed from the repeat expan-
et al., 2015), and Drosophila (Mizielinska et al., 2014) models
sion [r(G4C2)exp], which sequesters RNA-binding
(Figure 1A).
proteins (RBPs) and undergoes repeat-associated
non-ATG (RAN) translation to generate toxic dipep-
Current efforts toward developing c9ALS/FTD therapeutics
include both antisense oligonucleotides (ASOs) and small
molecules that target r(G4C2)exp. Indeed, an ASO directed at
C9ORF72 mRNA improved ALS-associated defects in mouse
models (Gendron et al., 2017; Jiang et al., 2016). Small mole-
tide repeats. Using in vitro and cell-based assays,
we identified a small molecule (4) that selectively
bound r(G4C2)exp, prevented sequestration of an
RBP, and inhibited RAN translation. Indeed, biophys- cules, however, could be a viable alternative, owing to their
favorable pharmacological properties amenable to traditional
drug development (Su et al., 2014; Yang et al., 2015). As the
ical characterization showed that 4 selectively bound
the hairpin form of r(G4C2)exp, and nuclear magnetic
repeat lies in an intron (DeJesus-Hernandez et al., 2011; Renton
et al., 2011), small molecule binding could be advantageous in
resonance spectroscopy studies and molecular dy-
namics simulations defined this molecular recogni-
that it may not affect overall protein expression (Yang et al.,
tion event. Cellular imaging revealed that 4 localized
2015) of C9ORF72, which currently has an unknown function.
to r(G4C2)exp cytoplasmic foci, the putative sites of
Previously, we discovered that the small molecule 1a binds
RAN translation. Collectively, these studies highlight
that the hairpin structure of r(G4C2)exp is a therapeu-
tically relevant target and small molecules that bind it
can ameliorate c9ALS/FTD-associated toxicity.
r(G4C2)exp and inhibits RAN translation and foci formation in
both cultured cells and induced neurons, demonstrating the
power of small molecules to favorably modulate RNA toxicity
(Su et al., 2014).
Interestingly, r(G4C2)exp exists in an equilibrium between two
folded states, a hairpin and G-quadruplex (Figure 1A) (Su et al.,
2014; Taylor, 2014). This equilibrium may be perturbed by tem-
perature or by the presence of K+ required for G-quadruplex
INTRODUCTION
Amyotrophic lateral sclerosis (ALS), a devastating neurodegen- folding (Lane et al., 2008). Often the G-quadruplex is presumed
erative disorder caused by progressive loss of motor neurons, to be the therapeutically relevant structure (Conlon et al., 2016;
often results in paralysis and death within three to five years of Fay et al., 2017), as supported by a few studies that showed
onset (Taylor et al., 2016). Current treatment options for ALS favorable biological effects with known G-quadruplex-targeting
are scarce and only palliative (Taylor et al., 2016). With the dis- ligands (Simone et al., 2018; Zamiri et al., 2014). In contrast, 1a
covery that an expanded, transcribed G4C2 repeat in C9ORF72 binds the hairpin structure of r(G4C2)exp and improves c9ALS/
is the most common genetic cause of ALS and frontotemporal FTD-associated defects in iNeurons (Su et al., 2014). Thus, tar-
dementia (FTD) (hereafter ‘‘c9ALS/FTD’’), a new therapeutic geting the G-quadruplex or hairpin structure could both be
target has emerged (DeJesus-Hernandez et al., 2011; Renton important for the development of c9ALS/FTD therapeutics.
Cell Chemical Biology 26, 1–12, February 21, 2019 ª 2018 Elsevier Ltd.
1

Please cite this article in press as: Wang et al., The Hairpin Form of r(G4C2)exp in c9ALS/FTD Is Repeat-Associated Non-ATG Translated and a Target
for Bioactive Small Molecules, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.10.018
A
B
C
D
Figure 1. Screening Compounds for Binding to r(G4C2)8 by Displacement of 1a-TOQ and for Inhibiting RAN Translation
(A) Schematics of the equilibrium between the hairpin and the G-quadruplex forms of r(G4C2)exp (top) and RAN translation producing poly(GP) (bottom).
(B) Screening results as measured via displacement of 1a-TOQ (compound structures provided in Data S1) (top). Compounds with >90% similarity to 1a are
indicated with blue bars. Data represent mean values SEM of three independent measurements. Chemical structures of 1a and 1–4 (bottom).
(C) Schematic of a cell-based screen to monitor RAN translation of r(G4C2)exp
.
(D) Left, the cell-based assay can distinguish between selective and non-selective inhibitors of RAN translation by monitoring changes in fluorescence derived
from r(G4C2)66-No ATG-GFP (RAN) and mCherry (canonical translation). Right, inhibitory effect of compounds (10 mM) on RAN translation. Note: (1) concen-
trations of 2 and 3 were 25 mM and (2) concentration of CX-5461 was 5 mM due to cytotoxicity. BMVC and BMVC2 were studied with the NanoLuc reporter due to
interference with the GFP signal. Error bars are SD.
Thus, we sought to identify more potent inhibitors of r(G4C2)exp as demonstrated by the 3-fold enhancement of emission signal
toxicity than 1a, compare the activity of hairpin-binding small upon titration of r(G4C2)8. Fitting the change in emission of
molecules to known G-quadruplex ligands, and study which 1a-TOQ as a function of r(G4C2)8 concentration afforded a Kd
fold(s) is therapeutically relevant. Indeed, we found several of 110 14 nM (Data S1). To validate the assay, we competed
compounds that inhibited RAN translation better than off the increase in 1a-TOQ’s fluorescence observed upon bind-
known G-quadruplex ligands and the ASO Vivo-Morpholino, ing r(G4C2)8 with 1a, which afforded a dose response as ex-
mo(G2C4)4. The most potent compound, 4, inhibited two patho- pected. Using 1a, we calculated the Z factor of this assay,
mechanisms of c9ALS/FTD, sequestration of RNA-binding pro- which is equal to 0.88 and thus considered suitable for high
teins (RBPs), and generation of toxic DPRs via RAN translation. throughput screening (Zhang et al., 1999).
Nuclear magnetic resonance (NMR) spectroscopy studies and
Using the 1a-TOQ displacement assay, we identified ten 1a-
molecular dynamics simulations showed that 4 bound the like compounds that displaced >50% of 1a-TOQ from r(G4C2)8
1 3 1 nucleotide GG internal loop in the hairpin structure of (Figure 1B). Generally, compounds with >90% similarity to 1a
r(G4C2)exp. An in vitro translation assay and cellular imaging were the most potent (blue bars; Figure 1B). Several trends
further support that the hairpin structure of r(G4C2)exp is RAN were observed to elucidate chemical features that drive binding
translated and a biologically relevant structure for the develop- to r(G4C2)8. Compounds 2, 3, 5, and 12 have positive charges
ment of c9ALS/FTD therapeutics.
via N-alkylation of the pyridine ring and exhibited potent
binding, yet 1 and 4 also demonstrated similar binding affinities
despite being uncharged. This suggests that positive charges
contribute favorably but are not essential to molecular recogni-
tion between the small molecule probe and r(G4C2)exp. Dose-
dependent binding assays demonstrated that the best
RESULTS
A Dye Displacement Assay Identifies Small Molecules
that Bind More Tightly to r(G4C2)exp Than 1a
To identify more potent inhibitors of r(G4C2)exp than 1a (Su et al., performing compounds (1–4, Figure 1B) are more potent than
2014), we employed a chemical similarity search based on our 1a with half maximal inhibitory concentration (IC50) values
lead compound (Data S1). The search yielded a cohort of small ranging from 0.9 to 8.5 mM (Table 1).
molecules (n = 40) exhibiting at least 80% chemical similarity to
1a with varying charge, structure, and side chains (Data S1). 1a-like Small Molecules Inhibit RAN Translation in a
The ability of these 1a-like small molecules to bind r(G4C2)exp Cellular Reporter Assay
in vitro was measured via
a
dye displacement assay. To assess the bioactivity of r(G4C2)exp-binding compounds, we
Combining 1a with TO-PRO-1, a well-known fluorescent developed a cell-based screen to monitor selective inhibition
RNA intercalator (1a-TOQ [STAR Methods and Data S2]), of RAN translation of r(G4C2)exp (Figures 1C and 1D). In brief
yielded an excellent fluorescent reporter of r(G4C2)exp binding, (G4C2)exp was inserted into the 50 UTR of GFP lacking a canonical
2
Cell Chemical Biology 26, 1–12, February 21, 2019

Please cite this article in press as: Wang et al., The Hairpin Form of r(G4C2)exp in c9ALS/FTD Is Repeat-Associated Non-ATG Translated and a Target
for Bioactive Small Molecules, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.10.018
Compound 4 Exhibits Excellent Affinity and Selectivity
Table 1. IC50 Values for the Displacement of 1a-TOQ and for the
for Binding GG Internal Loops in the Hairpin Form of
r(G4C2)exp
Inhibition of r(G4C2)66-Mediated RAN Translation
Displacement of
We next investigated the affinity and selectivity of 4 binding to
1a-TOQ
Inhibition of RAN
various DNA and RNA targets by using biolayer interferometry
(BLI). Compound 4 bound the desired target, r(G4C2)8, with a
Kd of 0.26 mM, while weakly binding the antisense RNA
[r(G2C4)8] and a fully base paired RNA [r(G2C2)8] (Table 2). In
particular, 4 bound r(G4C2)8 ꢀ300-fold more tightly than to the
antisense r(G2C4)8 (Kd = 80 mM) and ꢀ540-fold more tightly
than to the r(G2C2)8 (Kd = 140 mM). Repeat length [r(G4C2)n
where n = 2, 4, and 6] did not affect 4’s binding affinity
(Table S2). The selectivity of 4 binding was studied further
with a well-known RNA G-quadruplex called TERRA, r(UAG3
(U2AG3)3). Although 4 did bind TERRA with a Kd of 8.6 mM, 4
Compound
IC50 (mM)
Translation IC50 (mM)
1a
2.6 1.0
1.3 0.5
1.4 0.4
0.9 0.2
8.5 2.0
–
11 1.0
0.27 0.05
>25
1
2
3
12 2.0
1.6 0.20
7.6 1.0
4
mo(G2C4)4
Error is reported as SD.
ATG start codon, akin to a construct previously reported to study bound 33-fold more tightly to r(G4C2)8 (Table 2). With respect
RAN translation of r(CGG)exp (Yang et al., 2015). Western blot to interactions with DNA, 4 binds to d(G4C2)8 with a Kd of
analysis confirmed that the product generated from this 0.31 mM; however, 4 does not affect (G4C2)66-No ATG-GFP
‘‘(G4C2)66-No ATG-GFP’’ construct is a RAN DPR-GFP fusion, RNA transcript levels as assessed by RT-qPCR (Figure S1D),
with no detectable amounts of the canonically translated GFP suggesting that binding interactions between 4 and G4C2 DNA
product (Figure S1A).
did not affect transcription and are therefore not biologically
To assess whether a small molecule selectively inhibits RAN relevant.
translation, cells were co-transfected with (G4C2)66-No ATG-
GFP and a plasmid encoding mCherry (ATG-mCherry). Thus, Compound 4 Is Selective for 1 3 1 Nucleotide GG Loops
two fluorescent signals could be produced, where GFP fluores- over Other Nucleic Acids In Vitro and in Cells
cence indicates RAN translation and mCherry fluorescence is a We previously found that r(UG3C2)exp produces toxic DPRs via
metric of canonical translation. By measuring the relative fluo- RAN translation, the genetic cause of spinocerebellar ataxia dis-
rescence of both proteins, the assay can distinguish whether ease type 36 (SCA36) (Liu et al., 2014; Matsuzono et al., 2017;
a small molecule globally affects protein translation non-selec- Obayashi et al., 2015). Similar to r(G4C2)exp, r(UG3C2)exp forms
tively (both GFP and mCherry are affected), selectively affects a hairpin structure with a GU internal loop stabilized by hydrogen
canonical translation (only mCherry is affected), or selectively bonding (Figure S2A). Therefore, r(UG3C2)exp serves as an excel-
inhibits RAN translation (only GFP is affected) (Figures 1C and lent comparison to determine whether 4 binds the 1 3 1 nucleo-
1D). The r(G4C2)exp RAN translation assay has an excellent Z tide GG internal loop in r(G4C2)exp selectively. Binding studies
factor of 0.81, as measured using compound 1a, and is there- revealed that 4 had an z1,000-fold lower affinity for binding
fore suitable for cell-based high throughput screening.
r(UG3C2)8 than r(G4C2)8 (Table 2); binding of 4 to r(UG3C2)8 was
The most promising compounds (1–4) from the 1a-TOQ also not detectable by NMR spectroscopy (Figure S2C). Not sur-
displacement screen inhibited RAN translation, with IC50 values prisingly, 4 is unable to inhibit RAN translation of r(UG3C2)62 in
ranging from 0.27 to 12 mM (blue bars in Figure 1D; Table 1). In transfected cells (Figure S2B).
fact, 1 and 4 are more potent than 1a, with IC50 values of 0.27,
1.6, and 11 mM, respectively (Table 1). Both 1 and 4 were better NMR Spectroscopy and Molecular Dynamics
inhibitors of r(G4C2)exp RAN translation than the oligonucleotide Simulations Reveal Features of the Molecular
Vivo-Morpholino, mo(G2C4)4 (Data S1), which had an IC50 of Recognition of r(G4C2)exp by 4
7.6 mM (p < 0.001) (Table 1). Thus, small molecules that target We next characterized the binding of 4 to r(G4C2) repeats using
structured RNAs can more potently modulate biological activity imino proton NMR and circular dichroism (CD) spectroscopies.
than oligonucleotides. This phenomenon is likely due to the ki- NMR characterization of 4 binding to r(G4C2)4 and to r(G4C2)8
netic issues of oligonucleotides binding to structured RNA tar- showed a new peak appearing at ꢀ10.5 ppm in the absence of
gets, which can be further confounded, in this case, by the oligo- K+ ions, resulting from 4 stabilizing the interactions between
nucleotide’s self-structure (Tran et al., 2014).
the 1 3 1 nucleotide GG internal loop (Figures 2A and S3A). As
To ensure that the relative decrease in the GFP as compared the imino proton signal for G-quadruplex is nearby that of
to the mCherry signal occurred as a result of inhibition of RAN hydrogen-bonded GG loops, we studied the binding of 4 to
translation, a counterscreen for 1–4 was completed with ATG- r(G4C2)2, which cannot form a G-quadruplex. Indeed, the same
GFP (canonical GFP translation) and ATG-mCherry (also canon- imino proton signal corresponding to the GG loops in r(G4C2)4
ical translation). As shown in Figures S1B and S1C, 2–4 did not and r(G4C2)8 was observed (Figure S4). Further, 4 prevents con-
significantly affect the GFP signal from the ATG-GFP construct version of the hairpin to the G-quadruplex (Figures 2A and S3A).
relative to the mCherry signal, while 1 induced a roughly 20% As the concentration of 4 increased, pronounced peak shifts of
decrease in the GFP signal at 2.5 mM. Therefore, 2–4 are selec- the GC base pair imino protons at 13.3 and 12.5 ppm were
tive RAN translation inhibitors, while 1 has some associated non- observed. By plotting the change in the NMR signals as a func-
specific effects. As 4 was the best selective inhibitor of RAN tion of [4]:[RNA], binding stoichiometry can be determined. As
translation in this cell-based assay, it was studied further.
shown in Figure S3A, 4 binds to r(G4C2)4 with a stoichiometric
Cell Chemical Biology 26, 1–12, February 21, 2019
3

Please cite this article in press as: Wang et al., The Hairpin Form of r(G4C2)exp in c9ALS/FTD Is Repeat-Associated Non-ATG Translated and a Target
for Bioactive Small Molecules, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.10.018
Table 2. Binding Parameters of 4 and r(G4C2)8, d(G4C2)8, r(G2C4)8 (Antisense Repeat), r((G2C2)4AAAA(G2C2)4) (GC Paired), r(UGGGCC)8,
and r(UAG3(U2AG3)3) (TERRA) by Biolayer Interferometry
Selectivity
for r(G4C2)8
RNA
K
d,1 (M)
kon,1 (1/M 3 s)
2.3 ( 0.3) 3 104
koff,1 (1/s)
6.0 ( 0.1) 3 10ꢂ3
(Fold)
–
Kd,2 (M)
kon,2 (1/M 3 s)
4.0 ( 0.2) 3 104
koff,2 (1/s)
r(G4C2)8
2.6 ( 0.3) 3 10ꢂ7
1.4 ( 0.1)
3 10ꢂ6
5.6 ( 0.3)
3 10ꢂ2
d(G4C2)8
3.1 ( 0.1) 3 10ꢂ7
2.6 ( 0.1) 3 104
8.0 ( 0.2) 3 10ꢂ3
1.2
4.0 ( 1.2)
3 10ꢂ6
2.0 ( 0.6) 3 104
8.0 ( 0.2)
3 10ꢂ2
r(G2C4)8 (antisense)
7.9 ( 5.9) 3 10ꢂ5
1.4 ( 0.9) 3 10ꢂ4
1.2 ( 0.9) 3 103
1.5 ( 0.9) 3 103
9.5 ( 0.3) 3 10ꢂ2
2.1 ( 0.8) 3 10ꢂ1
304
538
NA
NA
NA
NA
NA
NA
r((G2C2)4AAAA(G2C2)4)
(GC paired)
r(UG3C2)8
2.2 ( 0.9) 3 10ꢂ4
8.6 ( 0.2) 3 10ꢂ6
1.9 ( 0.8) 3 102
3.2 ( 0.1) 3 103
4.0 ( 0.2) 3 10ꢂ2
2.6 ( 0.1) 3 10ꢂ2
847
33
NA
NA
NA
NA
NA
NA
r(UAG3(U2AG3)3)
kon and koff values were generated by fitting the obtained data using ForteBio’s Data Analysis 7.1 software to 1:1 or 2:1 (heterogeneous) binding
equation using the entire time range. Values shown for kon and koff are representative of two independent measurements and the errors are the SD.
NA, not applicable.
ratio of 4:1 compound:RNA and to r(G4C2)8 with a stoichiometric G-Quadruplex Ligands Bind r(G4C2)exp but Do Not Inhibit
ratio of 6:1.
RAN Translation
Nuclear Overhauser effect spectroscopy (NOESY) analysis Collectively, the studies described above suggest that 4’s
revealed nuclear Overhauser effects (NOEs) between 4 and cellular mode of action is binding to r(G4C2)exp in its hairpin
the 1 3 1 nucleotide GG internal loops and adjacent GC pairs form and blocking RAN translation by stabilization of the
(Figure S3B). CD spectroscopy confirmed that 4 stabilizes the RNA’s structure. However, various reports have suggested
hairpin structure, as demonstrated by the increase in TM from that the therapeutically relevant fold of r(G4C2)exp is a G-quadru-
71ꢁC to 78ꢁC after the addition of 4 (Figure 2B). These studies plex (Figure 1A) (Biffi et al., 2014; Burger et al., 2005; Franceschin
also revealed that 4 has two binding modes, as demonstrated et al., 2006; Hershman et al., 2008; Huang et al., 2008; Rodriguez
by the two signals at 300 and 330 nm (Figure S3C). Fitting the et al., 2012; Rzuczek et al., 2010; Siddiqui-Jain et al., 2002;
resulting data from CD studies to a Hill equation indicates that Wang et al., 2014; Xu et al., 2017). We therefore compared the
binding of 4 is cooperative (Figure S3C). Collectively, these biological activity of 11 well-known G-quadruplex ligands
studies suggest that 4 binds between 1 3 1 nucleotide GG in- (Data S1) to 4 using our cell-based RAN translation assay (Fig-
ternal loops and GC pairs, stabilizing hydrogen bonding within ures 1C and 1D). (Note: BMVC and BMVC2 were tested using
the loop.
a (G4C2)66-No ATG-NanoLuc construct because they interfere
To confirm these results, a duplex model that contains three with the fluorescent signal of GFP; 4 was also tested using this
unique 1 3 1 nucleotide GG internal loops distinguishable by system, affording the same IC50 as it did for (G4C2)66-No ATG-
NMR spectroscopy (Figure S4A) was employed. By site-specif- GFP [Figure S1E].) Interestingly, the G-quadruplex ligands
ically methylating the 1 position of each G in a 1 3 1 nucleotide were poor inhibitors of RAN translation, significantly worse
GG internal loop, we were able to assign which G residues were than 4 (Figure 1D). The poor inhibitory effect was not due to
acting as hydrogen bond donors and acceptors in its interaction lack of affinity, as the G-quadruplex ligands demonstrated excel-
with 4 (Figures S4B–S4E). This analysis revealed that: (1) 4 stabi- lent binding to the quadruplex form of r(G4C2)4, induced by
lizes hydrogen bonding between the 1 3 1 nucleotide GG inter- addition of K+ (Table S1). TMPyP4 (Kd = 50 nM) and PDS
nal loop; (2) the 1 3 1 nucleotide GG internal loop hydrogen (Kd = 70 nM) bound more tightly to the quadruplex form than
bonds in a syn-anti orientation; and (3) several NOEs between 4 bound to the hairpin (Kd = 260 nM) (Tables 2 and S1). (Note:
4 and the imino protons of the guanine in the 1 3 1 nucleotide binding of G-quadruplex ligands to the G-quadruplex form was
GG internal loop were observed.
confirmed by NMR spectroscopy [Figure S5].) Thus, inhibiting
To provide additional insight into the molecular recognition of RAN translation is not simply a function of binding affinity, rather
4 and r(G4C2) repeats, we used molecular dynamics calculations the affinity for a particular structural form and its presence within
(see Supplemental Information). The lowest binding free energy a cell.
state from these studies showed 4 stacked between the 1 3 1
nucleotide GG internal loop and the closing GC base pairs, The Hairpin Is a Biologically Relevant Target for Small
stabilized by p-p interaction between 4 and loop residues (Fig- Molecule Modulation of r(G4C2)exp Biology
ures 2C–2E). Hydrogen bonding between 1 3 1 nucleotide GG The studies above suggest that the hairpin fold is therapeutically
internal loops in the absence of compound were observed in relevant for inhibiting r(G4C2)exp pathobiology, particularly RAN
the imino proton NMR spectrum at low temperature (283K) (Fig- translation. To gain further insight into the structural form adop-
ure S3D) but not at room temperature. The observation that ted by r(G4C2)exp in cells, we conducted both in vitro and cell-
hydrogen bonding within the GG loop in the presence of 4 was based experiments. Although both hairpin and G-quadruplex
observable at room temperature provides additional support forms are thermodynamically stable, we were interested in
that 4 stabilizes the RNA’s structure.
studying the kinetically favored form. Imino proton spectra
4
Cell Chemical Biology 26, 1–12, February 21, 2019

Please cite this article in press as: Wang et al., The Hairpin Form of r(G4C2)exp in c9ALS/FTD Is Repeat-Associated Non-ATG Translated and a Target
for Bioactive Small Molecules, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.10.018
Figure 2. CD and NMR Analysis of 4 Binding
to the 1 3 1 Nucleotide GG Internal Loops
(A) Imino proton NMR spectra of r(G4C2)4 with
increasing amounts of 4. A new peak appears
at ꢀ10.5 ppm upon addition of 4, consistent with 4
binding and stabilizing the GG pair’s hydrogen
bonds.
(B) CD melting of r(G4C2)4 in the presence and
absence of 4.
(C) Two molecules of 4 interacting with a model
r(G4C2)exp having two unique 1 3 1 GG internal loop
motifs (CGG and GGC) with syn-anti orientations.
(D) Top views of 4 interacting with 50-CGG/50-CGG.
(E) 50-GGC/50-GGC as displayed on the left. Blue,
red, and black colored residues represent the
first, second, and third base pairs from top,
respectively.
unfolding time (272–407 min) and the
population (0.4–0.6) (Figures S6D and
S6E) in the hairpin form increases from
4
to 12 repeats, indicating that the
hairpin structure has a higher folding pop-
ulation than the G-quadruplex and sug-
gesting that hairpin form is the kinetically
favored.
We next conducted an in vitro transla-
tion assay to determine whether the
hairpin or G-quadruplex form undergoes
RAN translation by using (G4C2)66-No
ATG-NanoLuc and luciferase signal as
a readout. The construct was used to
transcribe r(G4C2)66-No ATG-NanoLuc,
which was subsequently folded into a
G-quadruplex or hairpin form, as informed
by NMR spectroscopy studies. Signifi-
show that, upon addition of K+, hairpin signals decrease and the cant luciferase signal was only observed from r(G4C2)66-No
G-quadruplex signals increase (Figure 3A). A three-state kinetic ATG-NanoLuc folded into the hairpin form and not the G-quad-
model was employed to extrapolate the unfolding rate of the ruplex form (ꢀ3.5-fold difference compared with both the
hairpin into the G-quadruplex structure via an unfolded interme- G-quadruplex template and a reaction lacking an RNA template;
diate (Figure 3A and Equations 3.1– 3.4 in STAR Methods).
The unfolding rate of hairpin structure (k1) increased with tem-
p < 0.001) (Figure S2D).
Our in vitro studies suggest that the hairpin fold of r(G4C2)exp
perature (Kuo et al., 2015), while the folding rate of G-quadruplex predominates from a kinetic standpoint and is a main contributor
(k3) was temperature independent (Figures S6A and S6B), of pathobiology, at least in terms of RAN translation. To confirm
consistent with previous G-quadruplex folding studies (Gray these findings in cells, we used 4 to develop an r(G4C2)exp hairpin
and Chaires, 2008; Zhang and Balasubramanian, 2012). These imaging agent. In brief, 4 was conjugated with the well-known
data indicated that a transition state energy is required for hairpin RNA-binding dye TO-PRO-1, and this hybrid molecule is dubbed
structure unfolding, while there is no substantial barrier for 4-TO-PRO (Figure 3E). (G4C2)66-transfected cells were incu-
G-quadruplex folding. Using these temperature-dependent bated with 4-TO-PRO, revealing strong fluorescent spots in
measurements (Figure S6A), a sizable transition state energy, the cytoplasm as determined by fluorescence microscopy (Fig-
DGz = 20 kcal/mol, was calculated, suggesting that a substantial ure 3E). In contrast, these punctate spots were not visible using
kinetic barrier exists between the hairpin and unfolded states for TO-PRO-1 alone (Figure S7A). Pre-treating the cells with 4 or
r(G4C2)exp; the formation of the G-quadruplex is evidently limited RNase significantly reduced these fluorescent spots (Figures 3
by this kinetic barrier (Figure 3B) (Kuo et al., 2015).
and 3F), which suggests that 4 localized to these spots enriched
Computational analysis predicted that the stability of the with RNA. According to a previous study by Donnelly et al.
hairpin increases as repeat length increases (Figure S6C) (Zuker, (2013), these cytoplasmic spots are the putative sites of RAN
2003; Ash et al., 2013). A series of CD measurements confirmed translation. Therefore, the observed fluorescent signal from
this prediction, as increasing the number of repeats from 4 to 4-TO-PRO was most likely due to binding of cytoplasmic
8 to 12 lowered the observed free energy (Figure 3B). The r(G4C2)exp in the hairpin form.
Cell Chemical Biology 26, 1–12, February 21, 2019
5

Please cite this article in press as: Wang et al., The Hairpin Form of r(G4C2)exp in c9ALS/FTD Is Repeat-Associated Non-ATG Translated and a Target
for Bioactive Small Molecules, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.10.018
A
B
Hairpin
G-quadruplex
1.0
0.5
0.0
Anneal
Hairpin
12 h
8 h
4 h
1 h
G-quadruplex
0.5 h
No K+
0
400
800
Time (min)
1200
14
13
12
11
10
ppm
D
C
0
-4
-8
ΔG = 20kcal/mol
-5
-12
-16
-10
3.15
3.20
3.25
3.30
3.35
0
4
8
12
1/T (x103)
Repeat length
S
N
E
F
G
N
O
N
N
N
N
H
40
N
O
4-TO-PRO
N
H
30
20
10
0
DAPI
4-TO-PRO
*
*
Figure 3. Kinetic and Thermodynamic Study Reveals the Hairpin Structure Is the Major Population in rG4C2exp and Also Forms in Cells
(A) Time-resolved imino proton spectra of r(G4C2)4 were recorded in Li+-only buffer first, then at 0.5, 1, 4, 8, and 12 hr post addition of K+ (37ꢁC), and finally after
annealing at 95ꢁC in K+ (‘‘Anneal’’).
(legend continued on next page)
6
Cell Chemical Biology 26, 1–12, February 21, 2019

Please cite this article in press as: Wang et al., The Hairpin Form of r(G4C2)exp in c9ALS/FTD Is Repeat-Associated Non-ATG Translated and a Target
for Bioactive Small Molecules, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.10.018
To ascertain whether 4-TO-PRO bound the hairpin or G-quad-
The hnRNP H displacement assay was based on a previously
ruplex form of cytoplasmic r(G4C2)exp, the G-quadruplex-spe- developed time-resolved fluorescence resonance energy trans-
cific antibody BG4 was employed (Biffi et al., 2014) (STAR fer (TR-FRET) assay used to study whether small molecules in-
Methods). If 4 targets the hairpin form of r(G4C2)exp in cells as hibited the binding of muscleblind-like 1 protein to r(CUG)exp
expected, then BG4 should not co-localize with 4-TO-PRO; (Figure 5A) (Chen et al., 2012). In our assay, biotinylated
co-localization of BG4 and 4-TO-PRO would indicate that 4 r(G4C2)8 forms a complex with hnRNP H-His6. A FRET signal is
binds to the G-quadruplex form in cells. BG4 was added to the produced when streptavidin-XL665 binds the biotinylated RNA
cells first followed by addition of 4-TO-PRO to avoid disrupting and Tb-Anti-His6 binds to hnRNP H. Small molecules that bind
BG4 binding. Indeed, the BG4 signal did not co-localize with r(G4C2)8 and inhibit the formation of the r(G4C2)8-hnRNP H com-
4-TO-PRO (Figure 3G), providing evidence that 4 binds the plex reduce the FRET signal. In this assay, 4 inhibited hnRNP H
hairpin form of cellular r(G4C2)exp, and that the hairpin fold of binding with an IC50 of 19 mM (Figure 5B).
r(G4C2)exp occurs in cells.
We next measured reduction of foci upon compound treat-
ment in cells using RNA fluorescence in situ hybridization and im-
munostaining of hnRNP H. As shown in Figure 5C, 4 inhibits foci
formation in HEK293T cells expressing (G4C2)66-No ATG-GFP.
Investigating Compound Mode of Action: 4 Blocks
Polysome Assembly on r(G4C2)exp
We next studied 4’s mode of action, which could occur by: (1) Quantification confirmed that 4 treatment reduced the number
transcriptional silencing, (2) compound-induced cleavage of of foci-positive cells and decreased the number of foci in foci-
the mRNA, (3) blocking ribosomal binding, and/or (4) preventing positive cells by 2-fold (p = 0.0029; Figure 5B). Thus, 4 inhibits
ribosomal readthrough (stalling). No change in the steady-state two putative pathomechanisms of c9ALS/FTD—RAN translation
levels of r(G4C2)exp-containing transcript was observed, as and formation of r(G4C2)exp-protein complexes—while binding to
determined by qRT-PCR (Figure S1D); thus 4 does not appear the 1 3 1 nucleotide GG internal loop.
to silence transcription of the mRNA or induce its cleavage.
Polysome profiling studies were then employed to determine DISCUSSION
whether 4 inhibits ribosome binding or induces stalling. Treat-
ment of HEK293T cells expressing (G4C2)66-No ATG-GFP with Advancing the development of c9ALS/FTD therapeutics de-
4 reduced the amount of transcript loaded into high-molecular- pends on establishing whether the G-quadruplex and hairpin
weight polysomes by 28% (p < 0.001), low-molecular-weight forms are therapeutically relevant structures of r(G4C2)exp. Driven
polysomes by 18% (p < 0.01), and monosome-containing frac- by improving upon our previously reported small molecule 1a,
tions by 12% (p < 0.05) (Figures 4A and 4B). These results indi- which ameliorated r(G4C2)exp toxicity in iNeurons by binding to
cate that 4 acts as a steric block that hinders polysome assembly the hairpin structure (Su et al., 2014), we sought to identify better
on r(G4C2)exp, thus decreasing levels of toxic DPRs. Previous small molecule modulators of r(G4C2)exp toxicity and to use the
studies have found that small molecules targeting r(CGG)exp probes to study r(G4C2)exp structures adopted in cells. Indeed,
similarly block loading of the corresponding mRNA onto poly- we identified many compounds that bound to r(G4C2)exp and in-
somes and inhibit RAN translation (Yang et al., 2015).
hibited RAN translation better than our previously reported com-
pound 1a. The most promising compound, 4, exhibits excellent
Investigating Compound Mode of Action: 4 Blocks
Formation of Toxic r(G4C2)exp-RBP Complexes
Another pathomechanism of r(G4C2)exp is sequestration of RBPs. nucleotide GG internal loop in the hairpin structure in r(G4C2)exp
selectivity and potency toward r(G4C2)exp
.
Our biophysical studies indicated that 4 binds to the 1 3 1
,
As 4 effectively blocks ribosomal binding, we hypothesized that inducing 1 3 1 nucleotide GG hydrogen bonding by reducing the
4 could also block RBPs from binding r(G4C2)exp, which accumu- dynamics of the 1 3 1 nucleotide GG internal loop. The 1 3 1
late in nuclear foci (Mizielinska and Isaacs, 2014). A key protein nucleotide GG internal loop in the hairpin is critical for molecular
that binds r(G4C2)exp and colocalizes with r(G4C2)exp foci is het- recognition and binding of 4 to the r(G4C2)exp. The p-p stacking
erogeneous nuclear ribonucleoprotein H (hnRNP H) (Prudencio interaction of 4 with the individual guanine bases in the 1 3 1
et al., 2015, Lee et al., 2013). We measured the affinity of hnRNP nucleotide GG loop and adjacent GC closing pair stabilizes the
H for r(G4C2)8 using a gel shift assay, yielding a Kd of 65 26 nM complex’s structure.
(Figure S7B). Therefore, we developed an in vitro assay around
this complex to assess whether compound binding to r(G4C2)exp teractions, thereby inhibiting two putative c9ALS/FTD patho-
inhibits RNA-protein complex formation.
mechanisms of r(G4C2)exp: RAN translation of r(G4C2)exp and
The binding of 4 to r(G4C2)exp prevents key RNA-protein in-
(B) Kinetic trace analysis of r(G4C2)4 as monitored by the imino proton NMR signal of hairpin (solid squares) or G-quadruplex region (open squares) after addition of
K+ at 37ꢁC. The red line shows the fit to a three-state kinetic model.
(C) Activation free energy of r(G4C2)4 transition from hairpin structure to unfolded state by Eyring plot of k1. Data are represented as the mean SEM.
(D) Folding free energy of hairpin structure (DGꢁHairpin) as a function of r(G4C2)n repeat length (n = 4–12), as measured by CD melting temperature. Data are
represented as the mean SD.
(E) Chemical structure of 4-TO-PRO and representative images of 4-TO-PRO-treated (G4C2)66-transfected cells.
(F) Quantification of foci-positive cells treated with 4-TO-PRO only (n = 112 cells; 3 biological replicates), pre-treated with 4 and then stained with 4-TO-PRO
(n = 100 cells; 3 biological replicates), or treated with RNase and then stained with 4-TO-PRO (n = 142 cells; 3 biological replicates). Data are represented as the
mean SD.
(G) Representative images of cells treated first with BG4 and then 4-TO-PRO.
Scale bar, 5 mm (in both images). *p < 0.05 as determined by a two-tailed Student t test.
Cell Chemical Biology 26, 1–12, February 21, 2019
7

Please cite this article in press as: Wang et al., The Hairpin Form of r(G4C2)exp in c9ALS/FTD Is Repeat-Associated Non-ATG Translated and a Target
for Bioactive Small Molecules, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.10.018
with oligonucleotide-based modalities that target r(G4C2)exp
including: (1) the self-structure of oligonucleotides complemen-
tary to r(G4C2) repeats (Kovanda et al., 2015)]; (2) off-targets
due to sequence, rather than structure-based recognition;
and (3) effects on both RAN and canonical ATG translation
(Yang et al., 2015).
A previous report found that hnRNP H colocalized with the
G-quadruplex antibody BG4 in c9ALS astrocytes and that the
BG4 signal was higher in c9ALS cells than control cell lines (Con-
lon et al., 2016). A structural study, however, showed that the
conserved qRRM domain of hnRNP F/H binds to RNA G-rich se-
quences and prevents formation of G-quadruplexes (Dominguez
et al., 2010). Rather than preclude the formation of G-quadru-
plexes in r(G4C2)exp, our results indicate that the hairpin is a rele-
vant target for developing c9ALS/FTD chemical probes, as
demonstrated by the ability of 4 to inhibit the two most prominent
mechanisms of r(G4C2)exp pathology.
We compared the bioactivity of 4 to many G-quadruplex
ligands, as G-quadruplexes were previously reported to be the
target of small molecules (Fay et al., 2017; Haeusler et al.,
2014; Simone et al., 2018; Taylor, 2014; Zamiri et al., 2014).
These ligands did not effectively inhibit RAN translation despite
binding with low nM affinity to the G-quadruplex form of
r(G4C2)exp. An in vitro translation assay showed that the hairpin
form of r(G4C2)66 undergoes RAN translation, while the G-quad-
ruplex does not (Figure S2D). Other studies have supported this
finding: Rode et al. (2016) demonstrated that tRNA shifts the
equilibrium between hairpin and G-quadruplex in mRNAs toward
the hairpin and increases translational efficacy, while Guo and
Bartel (2016) showed that RNA G-quadruplexes are globally
unfolded in mammalian cells.
Perhaps the most well-studied G-quadruplex, in terms of
both structure (Lim et al., 2009; Luu et al., 2006; Phan et al.,
2007) and folding pathway (Ambrus et al., 2006; Gray and
Chaires, 2008; Gray et al., 2014; Mashimo et al., 2008; Ma-
shimo and Sugiyama, 2007), is the human telomere repeat
sequence d(TTAGGG)n. Although the final fold is dependent
upon cation concentration and the loop sequence, three major
species predominate. Two small molecules have been discov-
ered that bind a particular telomerase fold; N-methylmesopor-
phyrin IX, which stabilizes a parallel topology (Sabharwal et al.,
2014), and epiberberine, which binds the hybrid-2 quadurplex
(Lin et al., 2018). Interestingly, the epiberberine-hybrid-2 com-
plex is different than most other small molecule-G-quadruplex
complexes (Collie et al., 2015) and perturbs the equilibria be-
tween quadruplex forms, converting other structures into the
hybrid-2.
Figure 4. Polysome Profiling Experiments Elucidate 4’s Mode of
Inhibition of RAN Translation
(A) Polysome fractionation profiles of cell lysates obtained from HEK293T cells
transfected with (G4C2)66-No ATG-GFP plasmid upon treatment with 4 or
vehicle (DMSO). Polysome fractionation profiles are representative of two in-
dependent experiments.
(B) Percentage of (G4C2)66-No ATG-GFP mRNA transcript present within
monosome- and polysome-containing fractions with (black) and without
(white) treatment of 4. Fractions labeled as ‘‘Monosomes’’ contain 40S, 60S,
and 80S ribosomal subunits; ‘‘LMW’’ indicates low molecular weight poly-
somes; and ‘‘HMW’’ indicates high molecular weight polysomes. Data are
represented as the mean SD. *p < 0.05; **p < 0.01; ***p < 0.001, as deter-
mined by a two-tailed Student t test.
Collectively, our studies demonstrate that the hairpin is a ther-
apeutically relevant structure involved in the pathological mech-
formation of toxic RBP complexes. 4 inhibits RAN translation anisms of c9ALS/FTD (Cammas and Millevoi, 2017; Simone
by acting as a steric block to prevent polysome assembly. In et al., 2015). Although a recent study reported that a small
fact, 4 inhibited RAN translation more potently than the oligonu- molecule reduces r(G4C2)exp foci and inhibits RAN translation
cleotide mo(G2C4)4, demonstrating that small molecule probes in iNeurons by targeting r(G4C2)exp G-quadruplexes, the study
are a viable alternative to oligonucleotides. While an ASO did not investigate whether the hairpin structure could be a
complementary to the RAN translation initiation site inhibited ri- target for the compound (Simone et al., 2018). G-quadruplex
bosomal scanning and, hence, RAN translation (Tabet et al., structures are still a potential target to inhibit translation (Bugaut
2018), ASOs are most efficacious when targeting unstruc- et al., 2012; Bugaut and Balasubramanian, 2012; Endoh et al.,
tured regions; thus RNA repeat expansions are not ideal tar- 2013; Murat et al., 2014) based on their high thermostability
gets. Various factors compound the challenges associated and structural hierarchy.
8
Cell Chemical Biology 26, 1–12, February 21, 2019

Please cite this article in press as: Wang et al., The Hairpin Form of r(G4C2)exp in c9ALS/FTD Is Repeat-Associated Non-ATG Translated and a Target
for Bioactive Small Molecules, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.10.018
B
A
D
E
C
Figure 5. 4 Inhibits Binding of hnRNP H In Vitro and in HEK293T Cells Expressing (G4C2)66-No ATG-GFP
(A) Schematic of the in vitro TR-FRET assay used to monitor small-molecule inhibition of the r(G4C2)8-hnRNP H complex.
(B) Dose-response curve for 4 in the TR-FRET assay.
(C) Representative images of RNA fluorescence in situ hybridization (FISH) of r(G4C2)exp and immunostaining of hnRNP H in untreated cells.
(D) Representative images of RNA FISH of r(G4C2)exp and immunostaining of hnRNP H in 4-treated cells.
(E) Quantification of foci-positive cells in panels C and D (n = 152; 3 biological replicates for untreated cells and n = 124; 3 biological replicates for 4-treated cells).
**p < 0.01 as determined by a two-tailed Student t test.
SIGNIFICANCE
(1) 4 binds cellular r(G4C2)exp, and (2) r(G4C2)exp folds into the
hairpin structure in cells. Collectively, our studies demon-
The most common genetic cause of ALS and FTD is the strate that a small molecule targeting the hairpin structure
expanded G4C2 [(G4C2)exp] repeat in C9ORF72. The RNA tran- can inhibit r(G4C2)exp-associated toxicity. In summary, this
scribed from this repeat expansion directly causes ‘‘c9ALS/ work not only highlights the power of rationally designed
FTD’’ by sequestering RNA-binding proteins and producing chemical probes to identify and validate therapeutically rele-
toxic dipeptide repeats via RAN translation. Thus, r(G4C2)exp vant RNA structures in cells but also provides new insight into
is an important target for chemical probe and lead medicine the role of r(G4C2)exp structure in c9ALS/FTD biology.
development. The c9ALS/FTD RNA exists in equilibrium be-
tween two forms, a hairpin displaying 1 3 1 nucleotide GG in-
STAR+METHODS
ternal loops and a G-quadruplex. We demonstrated that our
small molecule, 4, binds the 1 3 1 nucleotide GG internal
Detailed methods are provided in the online version of this paper
and include the following:
loop in the hairpin structure of r(G4C2)exp and thereby inhibits
two putative pathomechanisms of c9ALS/FTD. Surprisingly,
known G-quadruplex ligands were poor inhibitors of RAN
translation. This finding prompted an investigation into the
energetics of r(G4C2)exp folding, which revealed that the
hairpin form is more kinetically favorable than the G-quadru-
plex. Further, in vitro translation assays showed that the
hairpin, not the G-quadruplex, form of r(G4C2)exp, undergoes
RAN translation. Direct imaging using a fluorogenic hybrid
molecule combining 4 with TO-PRO-1 demonstrated that:
d KEY RESOURCES TABLE
d CONTACT FOR REAGENT AND RESOURCE SHARING
d EXPERIMENTAL MODEL AND SUBJECT DETAILS
B Cell Culture
B Authentication of Cell Lines Used
d METHOD DETAILS
B Compounds
B Oligonucleotides
Cell Chemical Biology 26, 1–12, February 21, 2019
9

Please cite this article in press as: Wang et al., The Hairpin Form of r(G4C2)exp in c9ALS/FTD Is Repeat-Associated Non-ATG Translated and a Target
for Bioactive Small Molecules, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.10.018
B Plasmids
Received: March 21, 2018
Revised: July 16, 2018
B Affinity of 1a-TOQ for r(G4C2)8
B Dye Displacement Assay to Study Binding of Com-
pounds to r(G4C2)8
Accepted: October 20, 2018
Published: November 29, 2018
B Biolayer Interferometry (BLI)
B Recombinant hnRNP H1-His6 Production, Isolation,
and Purification
REFERENCES
Almeida, S., Gascon, E., Tran, H., Chou, H.J., Gendron, T.F., Degroot, S.,
Tapper, A.R., Sellier, C., Charlet-Berguerand, N., Karydas, A., et al. (2013).
Modeling key pathological features of frontotemporal dementia with
C9ORF72 repeat expansion in iPSC-derived human neurons. Acta
Neuropathol. 126, 385–399.
B Affinity of hnRNP H1 for r(G4C2) Repeats
B TR-FRET hnRNP H1 Binding Assay
B Verification of RAN Translation Product in (G4C2)66
NoATG–GFP Plasmid by Western Blot
–
Ambrus, A., Chen, D., Dai, J., Bialis, T., Jones, R.A., and Yang, D. (2006).
Human telomeric sequence forms a hybrid-type intramolecular G-quadruplex
structure with mixed parallel/antiparallel strands in potassium solution. Nucleic
Acids Res. 34, 2723–2735.
B Cell-Based Assay to Assess RAN Translation and Pol-
y(GP) Response
B Effect of Compound Treatment on (G4C2)66–No ATG–
GFP Transcript Levels
Ash, P.E., Bieniek, K.F., Gendron, T.F., Caulfield, T., Lin, W.L., Dejesus-
Hernandez, M., van Blitterswijk, M.M., Jansen-West, K., Paul, J.W., 3rd,
Rademakers, R., et al. (2013). Unconventional translation of C9ORF72
GGGGCC expansion generates insoluble polypeptides specific to c9FTD/
ALS. Neuron 77, 639–646.
B CD Spectroscopy
B NMR Spectroscopy
B Analytical Solution of the Three-State Transition Model
B RNA FISH and Immunostaining
B Foci Imaging by 4-TO-PRO and BG4, a G-Quadruplex
Antibody
Bayly, C.I., Cieplak, P., Cornell, W.D., and Kollman, P.A. (1993). A well-
behaved electrostatic potential based method using charge restraints for
deriving atomic charges - the Resp Model. J. Phys. Chem. 97, 10269–10280.
B Polysome Profiling
Biffi, G., Di Antonio, M., Tannahill, D., and Balasubramanian, S. (2014).
Visualization and selective chemical targeting of RNA G-quadruplex structures
in the cytoplasm of human cells. Nat. Chem. 6, 75–80.
B Verification of RAN Translation Product in (G4C2)66
–
NoATG–NanoLuc Plasmid by SDS-PAGE and LC-MS
B In Vitro Translation Assay
Bugaut, A., and Balasubramanian, S. (2012). 5’-UTR RNA G-quadruplexes:
B Molecular Dynamics & Computational Methods
B Synthetic Methods
translation regulation and targeting. Nucleic Acids Res. 40, 4727–4741.
Bugaut, A., Murat, P., and Balasubramanian, S. (2012). An RNA hairpin
to G-quadruplex conformational transition. J. Am. Chem. Soc. 134,
19953–19956.
B Synthetic Scheme for 1a-TOQ
B Synthetic Scheme for 4-TO-PRO
d QUANTIFICATION AND STATISTICAL ANALYSIS
d DATA AND SOFTWARE AVAILABILITY
Burger, A.M., Dai, F., Schultes, C.M., Reszka, A.P., Moore, M.J., Double, J.A.,
and Neidle, S. (2005). The G-quadruplex-interactive molecule BRACO-19 in-
hibits tumor growth, consistent with telomere targeting and interference with
telomerase function. Cancer Res. 65, 1489–1496.
SUPPLEMENTAL INFORMATION
Cammas, A., and Millevoi, S. (2017). RNA G-quadruplexes: emerging mecha-
Supplemental Information includes seven figures, three tables, and two data
files and can be found with this article online at https://doi.org/10.1016/j.
chembiol.2018.10.018.
nisms in disease. Nucleic Acids Res. 45, 1584–1595.
Carreon, J.R., Mahon, K.P., Jr., and Kelley, S.O. (2004). Thiazole orange-pep-
tide conjugates: sensitivity of DNA binding to chemical structure. Org. Lett. 6,
517–519.
ACKNOWLEDGMENTS
Case, D.A., Betz, R.M., Cerutti, D.S., Cheatham, T.E., Darden, T.A., Duke, R.E.,
Giese, T.J., Gohlke, H., Goetz, A.W., Homeyer, N., et al. (2016). AMBER 16
(University of California).
Thisworkwasfunding bytheNIH(DP1NS096898-02toM.D.D.;P01NS099114-
01 to M.D.D. and L.P.; R35NS097273 to L.P.), Target ALS (to M.D.D.), and the
Nelson Family Fund (to M.D.D). Z.-F.W. is supported in part by the Academia
Sinica-TSRI Postdoctoral Fellowship. A.U. acknowledges the Deutsche For-
schungsgemeinschaft (DFG) for the DFG Postdoctoral Fellowship.
Chen, C.Z., Sobczak, K., Hoskins, J., Southall, N., Marugan, J.J., Zheng, W.,
Thornton, C.A., and Austin, C.P. (2012). Two high-throughput screening as-
says for aberrant RNA-protein interactions in myotonic dystrophy type 1.
Anal. Bioanal. Chem. 402, 1889–1898.
Chew, J., Gendron, T.F., Prudencio, M., Sasaguri, H., Zhang, Y.J.,
Castanedes-Casey, M., Lee, C.W., Jansen-West, K., Kurti, A., Murray, M.E.,
et al. (2015). Neurodegeneration. C9ORF72 repeat expansions in mice cause
TDP-43 pathology, neuronal loss, and behavioral deficits. Science 348,
1151–1154.
AUTHOR CONTRIBUTIONS
M.D.D. directed the study, conceived of the ideas, designed the experiments,
and designed and generated the (G4C2)66-No ATG-GFP and (G4C2)66-No ATG-
NanoLuc plasmids. Z.-F.W. designed and performed the G-quadruplex li-
gands study, NMR kinetic experiments, in vitro translation assay, and image
study. A.U. performed BLI and polysome profiling experiments. A.U., V.B.,
W.-Y.Y., and R.F. synthesized the compounds. R.G. performed Western blot
and TR-FRET experiments. I.Y. performed molecular modeling. S.G.R. and
J.E.R. synthesized the G-quadruplex ligand, PyPx-2-DMA. Y.-J.Z., T.F.G.,
and L.P. generated (G4C2)66 and (TG3C2)62 plasmids and performed PolyGP
activity assay. M.D.D., J.L.C.-D., and B.G.D. wrote the manuscript, edited
by others.
Childs-Disney, J.L., Stepniak-Konieczna, E., Tuan, T., Yildirim, I., Park, H.,
Chen, C.Z., Hoskins, J., Southall, N., Marugan, J.J., Patnaik, S., et al. (2013).
Induction and reversal of myotonic dystrophy type 1 pre-mRNA splicing de-
fects by small molecules. Nat. Commun. 4, 2044.
Childs-Disney, J.L., Yildirim, I., Park, H., Lohman, J.R., Guan, L., Tran, T.,
Sarkar, P., Schatz, G.C., and Disney, M.D. (2014). Structure of the myotonic
dystrophy type 2 RNA and designed small Molecules that reduce toxicity.
ACS Chem. Biol. 9, 538–550.
Chou, M.Y., Rooke, N., Turck, C.W., and Black, D.L. (1999). hnRNP H is a
component of a splicing enhancer complex that activates a c-src alternative
exon in neuronal cells. Mol. Cell. Biol. 19, 69–77.
DECLARATION OF INTERESTS
M.D.D. is a consultant for Expansion Therapeutics.
10 Cell Chemical Biology 26, 1–12, February 21, 2019

Please cite this article in press as: Wang et al., The Hairpin Form of r(G4C2)exp in c9ALS/FTD Is Repeat-Associated Non-ATG Translated and a Target
for Bioactive Small Molecules, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.10.018
Collie, G.W., Campbell, N.H., and Neidle, S. (2015). Loop flexibility in human
telomeric quadruplex small-molecule complexes. Nucleic Acids Res. 43,
4785–4799.
tial G-quadruplex-forming sequences in Saccharomyces cerevisiae. Nucleic
Acids Res. 36, 144–156.
Huang, F.C., Chang, C.C., Lou, P.J., Kuo, I.C., Chien, C.W., Chen, C.T., Shieh,
F.Y., Chang, T.C., and Lin, J.J. (2008). G-quadruplex stabilizer 3,6-bis(1-
methyl-4-vinylpyridinium)carbazole diiodide induces accelerated senescence
and inhibits tumorigenic properties in cancer cells. Mol. Cancer Res. 6,
955–964.
Conlon, E.G., Lu, L., Sharma, A., Yamazaki, T., Tang, T., Shneider, N.A., and
Manley, J.L. (2016). The C9ORF72 GGGGCC expansion forms RNA
G-quadruplex inclusions and sequesters hnRNP H to disrupt splicing in ALS
brains. Elife 5, https://doi.org/10.7554/eLife.17820.
Cornell, W.D., Cieplak, P., Bayly, C.I., Gould, I.R., Merz, K.M., Ferguson, D.M.,
Spellmeyer, D.C., Fox, T., Caldwell, J.W., and Kollman, P.A. (1995). A second
generation force field for the simulation of proteins, nucleic acids, and organic
molecules. J. Am. Chem. Soc. 117, 5179–5197.
Jiang, J., Zhu, Q., Gendron, T.F., Saberi, S., McAlonis-Downes, M., Seelman,
A., Stauffer, J.E., Jafar-Nejad, P., Drenner, K., Schulte, D., et al. (2016). Gain of
toxicity from ALS/FTD-linked repeat expansions in C9ORF72 is alleviated by
antisense oligonucleotides targeting GGGGCC-containing RNAs. Neuron 90,
535–550.
Cornell, W.D., Cieplak, P., Bayly, C.I., and Kollman, P.A. (1993). Application of
RESP charges to calculate conformational energies, hydrogen-bond energies,
and free-energies of solvation. J. Am. Chem. Soc. 115, 9620–9631.
Korobov, V., and Ochkov, V. (2011). Chemical Kinetics with Mathcad and
Maple (Springer-Verlag), pp. 35–72.
DeJesus-Hernandez, M., Mackenzie, I.R., Boeve, B.F., Boxer, A.L., Baker, M.,
Rutherford, N.J., Nicholson, A.M., Finch, N.A., Flynn, H., Adamson, J., et al.
(2011). Expanded GGGGCC hexanucleotide repeat in noncoding region of
C9ORF72 causes chromosome 9p-linked FTD and ALS. Neuron 72, 245–256.
Kovanda, A., Zalar, M., Sket, P., Plavec, J., and Rogelj, B. (2015). Anti-sense
DNA d(GGCCCC)n expansions in C9ORF72 form i-motifs and protonated hair-
pins. Sci. Rep. 5, 17944.
Kuo, M.H., Wang, Z.F., Tseng, T.Y., Li, M.H., Hsu, S.T., Lin, J.J., and Chang,
T.C. (2015). Conformational transition of a hairpin structure to G-quadruplex
within the WNT1 gene promoter. J. Am. Chem. Soc. 137, 210–218.
Disney, M.D., Liu, B., Yang, W., Sellier, C., Tran, T., Charlet-Berguerand, N.,
and Childs-Disney, J.L. (2012). A small molecule that targets r(CGG)exp and im-
proves defects in fragile X-associated tremor ataxia syndrome. ACS Chem.
Biol. 7, 1711–1718.
Lane, A.N., Chaires, J.B., Gray, R.D., and Trent, J.O. (2008). Stability and ki-
netics of G-quadruplex structures. Nucleic Acids Res. 36, 5482–5515.
Dominguez, C., Fisette, J.F., Chabot, B., and Allain, F.H. (2010). Structural ba-
sis of G-tract recognition and encaging by hnRNP F quasi-RRMs. Nat. Struct.
Mol. Biol. 17, 853–861.
Lee, Y.B., Chen, H.J., Peres, J.N., Gomez-Deza, J., Attig, J., Stalekar, M.,
Troakes, C., Nishimura, A.L., Scotter, E.L., Vance, C., et al. (2013).
Hexanucleotide repeats in ALS/FTD form length-dependent RNA foci,
sequester RNA binding proteins, and are neurotoxic. Cell Rep. 5, 1178–1186.
Donnelly, C.J., Zhang, P.W., Pham, J.T., Haeusler, A.R., Mistry, N.A.,
Vidensky, S., Daley, E.L., Poth, E.M., Hoover, B., Fines, D.M., et al. (2013).
RNA toxicity from the ALS/FTD C9ORF72 expansion is mitigated by antisense
intervention. Neuron 80, 415–428.
Lim, K.W., Amrane, S., Bouaziz, S., Xu, W., Mu, Y., Patel, D.J., Luu, K.N., and
Phan, A.T. (2009). Structure of the human telomere in K+ solution: a stable bas-
ket-type G-quadruplex with only two G-tetrad layers. J. Am. Chem. Soc. 131,
4301–4309.
Endoh, T., Kawasaki, Y., and Sugimoto, N. (2013). Translational halt during
elongation caused by G-quadruplex formed by mRNA. Methods 64, 73–78.
Lin, C., Wu, G., Wang, K., Onel, B., Sakai, S., Shao, Y., and Yang, D. (2018).
Molecular recognition of the hybrid-2 human telomeric G-quadruplex by epi-
berberine: insights into conversion of telomeric G-quadruplex structures.
Angew. Chem. Int. Ed. 57, 10888–10893.
Fay, M.M., Anderson, P.J., and Ivanov, P. (2017). ALS/FTD-associated
C9ORF72 repeat RNA promotes phase transitions in vitro and in cells. Cell
Rep. 21, 3573–3584.
Franceschin, M., Rossetti, L., D’Ambrosio, A., Schirripa, S., Bianco, A., Ortaggi,
G., Savino, M., Schultes, C., and Neidle, S. (2006). Natural and synthetic G-quad-
ruplexinteractiveberberinederivatives. Bioorg.Med.Chem. Lett.16, 1707–1711.
Liu, W., Ikeda, Y., Hishikawa, N., Yamashita, T., Deguchi, K., and Abe, K.
(2014). Characteristic RNA foci of the abnormal hexanucleotide GGCCUG
repeat expansion in spinocerebellar ataxia type 36 (Asidan). Eur. J. Neurol.
21, 1377–1386.
Frisch, M.J., Trucks, G.W., Schlegel, H.B., Scuseria, G.E., Robb, M.A.,
Cheeseman, J.R., Montgomery, J.A., Vreven, T., Kudin, K.N., et al. (2004).
Gaussian 03 (Gaussian).
Luu, K.N., Phan, A.T., Kuryavyi, V., Lacroix, L., and Patel, D.J. (2006). Structure
of the human telomere in K+ solution: an intramolecular (3 + 1) G-quadruplex
scaffold. J. Am. Chem. Soc. 128, 9963–9970.
Gendron, T.F., Bieniek, K.F., Zhang, Y.J., Jansen-West, K., Ash, P.E.,
Caulfield, T., Daughrity, L., Dunmore, J.H., Castanedes-Casey, M., Chew, J.,
et al. (2013). Antisense transcripts of the expanded C9ORF72 hexanucleotide
repeat form nuclear RNA foci and undergo repeat-associated non-ATG trans-
lation in c9FTD/ALS. Acta Neuropathol. 126, 829–844.
Marky, L.A., and Breslauer, K.J. (1987). Calculating thermodynamic data for
transitions of any molecularity from equilibrium melting curves. Biopolymers
26, 1601–1620.
Mashimo, T., Sannohe, Y., Yagi, H., and Sugiyama, H. (2008). Folding path-
ways of hybrid-1 and hybrid-2 G-quadruplex structures. Nucleic Acids
Symp. Ser. (Oxf.) 52, 409–410.
Gendron, T.F., Chew, J., Stankowski, J.N., Hayes, L.R., Zhang, Y.J.,
Prudencio, M., Carlomagno, Y., Daughrity, L.M., Jansen-West, K.,
Perkerson, E.A., et al. (2017). Poly(GP) proteins are a useful pharmacodynamic
marker for C9ORF72-associated amyotrophic lateral sclerosis. Sci. Transl
Med. 9, https://doi.org/10.1126/scitranslmed.aai7866.
Mashimo, T., and Sugiyama, H. (2007). Folding pathways of human telomeric
hybrid G-quadruplex structure. Nucleic Acids Symp. Ser. (Oxf.) 41, 239–240.
Matsuzono, K., Imamura, K., Murakami, N., Tsukita, K., Yamamoto, T., Izumi,
Y., Kaji, R., Ohta, Y., Yamashita, T., Abe, K., et al. (2017). Antisense oligonucle-
otides reduce RNA foci in spinocerebellar ataxia 36 patient iPSCs. Mol. Ther.
Nucleic Acids 8, 211–219.
Gray, R.D., and Chaires, J.B. (2008). Kinetics and mechanism of K+- and
Na+-induced folding of models of human telomeric DNA into G-quadruplex
structures. Nucleic Acids Res. 36, 4191–4203.
Gray, R.D., Trent, J.O., and Chaires, J.B. (2014). Folding and unfolding path-
Mizielinska, S., Gronke, S., Niccoli, T., Ridler, C.E., Clayton, E.L., Devoy, A.,
Moens, T., Norona, F.E., Woollacott, I.O., Pietrzyk, J., et al. (2014). C9orf72
repeat expansions cause neurodegeneration in Drosophila through arginine-
rich proteins. Science 345, 1192–1194.
ways of the human telomeric G-quadruplex. J. Mol. Biol. 426, 1629–1650.
Guo, J.U., and Bartel, D.P. (2016). RNA G-quadruplexes are globally unfolded
in eukaryotic cells and depleted in bacteria. Science 353, https://doi.org/10.
1126/science.aaf5371.
Mizielinska, S., and Isaacs, A.M. (2014). C9orf72 amyotrophic lateral sclerosis
and frontotemporal dementia: gain or loss of function? Cur. Opin. Neurol. 27,
515–523.
Haeusler, A.R., Donnelly, C.J., Periz, G., Simko, E.A., Shaw, P.G., Kim, M.S.,
Maragakis, N.J., Troncoso, J.C., Pandey, A., Sattler, R., et al. (2014).
C9orf72 nucleotide repeat structures initiate molecular cascades of disease.
Nature 507, 195–200.
Moye, A.L., Porter, K.C., Cohen, S.B., Phan, T., Zyner, K.G., Sasaki, N.,
Lovrecz, G.O., Beck, J.L., and Bryan, T.M. (2015). Telomeric G-quadruplexes
are a substrate and site of localization for human telomerase. Nat. Commun.
6, 7643.
Hershman, S.G., Chen, Q., Lee, J.Y., Kozak, M.L., Yue, P., Wang, L.S., and
Johnson, F.B. (2008). Genomic distribution and functional analyses of poten-
Cell Chemical Biology 26, 1–12, February 21, 2019 11

Please cite this article in press as: Wang et al., The Hairpin Form of r(G4C2)exp in c9ALS/FTD Is Repeat-Associated Non-ATG Translated and a Target
for Bioactive Small Molecules, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.10.018
Murat, P., Zhong, J., Lekieffre, L., Cowieson, N.P., Clancy, J.L., Preiss, T.,
Balasubramanian, S., Khanna, R., and Tellam, J. (2014). G-quadruplexes
regulate Epstein-Barr virus-encoded nuclear antigen 1 mRNA translation.
Nat. Chem. Biol. 10, 358–364.
biomarker and lead small molecules to target r(GGGGCC)-associated defects
in c9FTD/ALS. Neuron 83, 1043–1050.
Tabet, R., Schaeffer, L., Freyermuth, F., Jambeau, M., Workman, M., Lee,
C.Z., Lin, C.C., Jiang, J., Jansen-West, K., Abou-Hamdan, H., et al. (2018).
CUG initiation and frameshifting enable production of dipeptide repeat pro-
teins from ALS/FTD C9ORF72 transcripts. Nat. Commun. 9, 152.
O’Rourke, J.G., Bogdanik, L., Muhammad, A.K., Gendron, T.F., Kim, K.J.,
Austin, A., Cady, J., Liu, E.Y., Zarrow, J., Grant, S., et al. (2015). C9orf72
BAC transgenic mice display typical pathologic features of ALS/FTD.
Neuron 88, 892–901.
Taylor, J.P. (2014). Neurodegenerative diseases: G-quadruplex poses
quadruple threat. Nature 507, 175–177.
Obayashi, M., Stevanin, G., Synofzik, M., Monin, M.L., Duyckaerts, C., Sato,
N., Streichenberger, N., Vighetto, A., Desestret, V., Tesson, C., et al. (2015).
Spinocerebellar ataxia type 36 exists in diverse populations and can be
caused by a short hexanucleotide GGCCTG repeat expansion. J. Neurol.
Neurosurg. Psychiatry 86, 986–995.
Taylor, J.P., Brown, R.H., Jr., and Cleveland, D.W. (2016). Decoding ALS: from
genes to mechanism. Nature 539, 197–206.
Tran, T., Childs-Disney, J.L., Liu, B., Guan, L., Rzuczek, S., and Disney, M.D.
(2014). Targeting the r(CGG) repeats that cause FXTAS with modularly assem-
bled small molecules and oligonucleotides. ACS Chem. Biol. 9, 904–912.
Onufriev, A., Bashford, D., and Case, D.A. (2004). Exploring protein native
states and large-scale conformational changes with a modified generalized
born model. Proteins 55, 383–394.
Wales, D.J., and Yildirim, I. (2017). Improving computational predictions of sin-
gle-stranded RNA tetramers with revised a/g torsional parameters for the
Amber force field. J. Phys. Chem. B 121, 2989–2999.
Phan, A.T., Kuryavyi, V., Luu, K.N., and Patel, D.J. (2007). Structure of two in-
tramolecular G-quadruplexes formed by natural human telomere sequences in
K+ solution. Nucleic Acids Res. 35, 6517–6525.
Wang, J.M., Huang, F.C., Kuo, M.H., Wang, Z.F., Tseng, T.Y., Chang, L.C.,
Yen, S.J., Chang, T.C., and Lin, J.J. (2014). Inhibition of cancer cell migration
and invasion through suppressing the Wnt1-mediating signal pathway by
G-quadruplex structure stabilizers. J. Biol. Chem. 289, 14612–14623.
Prudencio, M., Belzil, V.V., Batra, R., Ross, C.A., Gendron, T.F., Pregent, L.J.,
Murray, M.E., Overstreet, K.K., Piazza-Johnston, A.E., Desaro, P., et al. (2015).
Distinct brain transcriptome profiles in C9orf72-associated and sporadic ALS.
Nat. Neurosci. 18, 1175–1182.
Wang, J.M., Wolf, R.M., Caldwell, J.W., Kollman, P.A., and Case, D.A. (2005).
Development and testing of a general amber force field, (vol 25, pg 1157,
2004). J. Comput. Chem. 26, 114.
Renton, A.E., Majounie, E., Waite, A., Simon-Sanchez, J., Rollinson, S., Gibbs,
J.R., Schymick, J.C., Laaksovirta, H., van Swieten, J.C., Myllykangas, L., et al.
(2011). A hexanucleotide repeat expansion in C9ORF72 is the cause of chro-
mosome 9p21-linked ALS-FTD. Neuron 72, 257–268.
Xu, H., Di Antonio, M., McKinney, S., Mathew, V., Ho, B., O’Neil, N.J., Santos,
N.D., Silvester, J., Wei, V., Garcia, J., et al. (2017). CX-5461 is a DNA G-quad-
ruplex stabilizer with selective lethality in BRCA1/2 deficient tumours. Nat.
Commun. 8, 14432.
Rode, A.B., Endoh, T., and Sugimoto, N. (2016). tRNA shifts the G-quadruplex-
hairpin conformational equilibrium in RNA towards the hairpin conformer.
Angew. Chem. Int. Ed. 55, 14315–14319.
Yang, W.Y., He, F., Strack, R.L., Oh, S.Y., Frazer, M., Jaffrey, S.R., Todd, P.K.,
and Disney, M.D. (2016). Small molecule recognition and tools to study mod-
ulation of r(CGG)exp in fragile X-associated tremor ataxia syndrome. ACS
Chem. Biol. 11, 2456–2465.
Rodriguez, R., Miller, K.M., Forment, J.V., Bradshaw, C.R., Nikan, M., Britton,
S., Oelschlaegel, T., Xhemalce, B., Balasubramanian, S., and Jackson, S.P.
(2012). Small-molecule-induced DNA damage identifies alternative DNA struc-
tures in human genes. Nat. Chem. Biol. 8, 301–310.
Yang, W.Y., Wilson, H.D., Velagapudi, S.P., and Disney, M.D. (2015). Inhibition
of non-ATG translational events in cells via covalent small molecules targeting
RNA. J. Am. Chem. Soc. 137, 5336–5345.
Rzuczek, S.G., Pilch, D.S., Liu, A., Liu, L., LaVoie, E.J., and Rice, J.E. (2010).
Macrocyclic pyridyl polyoxazoles: selective RNA and DNA G-quadruplex
ligands as antitumor agents. J. Med. Chem. 53, 3632–3644.
Yildirim, I., Stern, H.A., Kennedy, S.D., Tubbs, J.D., and Turner, D.H. (2010).
Reparameterization of RNA c torsion parameters for the AMBER force field
and comparison to NMR spectra for cytidine and uridine. J. Chem. Theory
Comput. 6, 1520–1531.
Sabharwal, N.C., Savikhin, V., Turek-Herman, J.R., Nicoludis, J.M., Szalai,
V.A., and Yatsunyk, L.A. (2014). N-methylmesoporphyrin IX fluorescence as
a reporter of strand orientation in guanine quadruplexes. FEBS J. 281,
1726–1737.
Zamiri, B., Reddy, K., Macgregor, R.B., Jr., and Pearson, C.E. (2014). TMPyP4
porphyrin distorts RNA G-quadruplex structures of the disease-associated
r(GGGGCC)n repeat of the C9orf72 gene and blocks interaction of RNA-bind-
ing proteins. J. Biol. Chem. 289, 4653–4659.
Siddiqui-Jain, A., Grand, C.L., Bearss, D.J., and Hurley, L.H. (2002). Direct ev-
idence for a G-quadruplex in a promoter region and its targeting with a small
molecule to repress c-MYC transcription. Proc. Natl. Acad. Sci. U S A 99,
11593–11598.
Zhang, A.Y., and Balasubramanian, S. (2012). The kinetics and folding path-
ways of intramolecular G-quadruplex nucleic acids. J. Am. Chem. Soc. 134,
19297–19308.
Simone, R., Balendra, R., Moens, T.G., Preza, E., Wilson, K.M., Heslegrave, A.,
Woodling, N.S., Niccoli, T., Gilbert-Jaramillo, J., Abdelkarim, S., et al. (2018).
G-quadruplex-binding small molecules ameliorate C9orf72 FTD/ALS pathol-
ogy in vitro and in vivo. EMBO Mol. Med. 10, 22–31.
Zhang, J.H., Chung, T.D., and Oldenburg, K.R. (1999). A simple statistical
parameter for use in evaluation and validation of high throughput screening as-
says. J. Biomol. Screen. 4, 67–73.
Zu, T., Liu, Y., Banez-Coronel, M., Reid, T., Pletnikova, O., Lewis, J., Miller,
T.M., Harms, M.B., Falchook, A.E., Subramony, S.H., et al. (2013). RAN pro-
teins and RNA foci from antisense transcripts in C9ORF72 ALS and frontotem-
poral dementia. Proc. Natl. Acad. Sci. U S A 110, E4968–E4977.
Simone, R., Fratta, P., Neidle, S., Parkinson, G.N., and Isaacs, A.M. (2015).
G-quadruplexes: emerging roles in neurodegenerative diseases and the
non-coding transcriptome. FEBS Lett. 589, 1653–1668.
Su, Z., Zhang, Y., Gendron, T.F., Bauer, P.O., Chew, J., Yang, W.Y., Fostvedt,
E., Jansen-West, K., Belzil, V.V., Desaro, P., et al. (2014). Discovery of a
Zuker, M. (2003). Mfold web server for nucleic acid folding and hybridization
prediction. Nucleic Acids Res. 31, 3406–3415.
12 Cell Chemical Biology 26, 1–12, February 21, 2019

Please cite this article in press as: Wang et al., The Hairpin Form of r(G4C2)exp in c9ALS/FTD Is Repeat-Associated Non-ATG Translated and a Target
for Bioactive Small Molecules, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.10.018
STAR+METHODS
KEY RESOURCES TABLE
REAGENT or RESOURCE
SOURCE
IDENTIFIER
Antibodies
anti-hnRNP H
Novus Biologicals
Sigma-Aldrich
Cat#: NBP131648; RRID: 10003534
Cat#: MABE917; RRID: 2750936
anti-G-quadruplex, BG4
Chemicals, Peptides, and Recombinant Proteins
1-40
NCI
N/A
TOQ
Carreon et al., 2004
Su et al., 2014
This study
This study
This study
Sigma
N/A
1a-amine
N/A
1a-TOQ
N/A
i1-i8
N/A
4-TO-PRO
N/A
CHX (cycloheximide)
Cat#: 104450
TO-PRO-1
ThermoFisher
FisherScientific
Sigma
Cat#: T3602
TMPyP4
Cat#42-535-0
PDS
Cat#SML0678
Cat#SML1176
Cat#B3251
cPDS
Sigma
BER
Sigma
Braco-19
Sigma
Cat#SML0560
Cat#A11065
CX-5461
AdooQ Bioscience
Frontier Scientific
Rzuczek et al., 2010
Gift
NMM
Cat#NMM580
N/A
PyPx-2-DMA
BMVC
Dr. Ta-Chau Chang (IAMS, Academia Sinica)
Critical Commercial Assays
In vitro transcription assay
In vitro translation assay
TR-FRET assay(StreptavidinXL665)
TR-FRET assay (MAb Anti6His-Tb-cryptate)
Promega
Promega
Cisbio
Cat# P1320
Cat# L4540
Cat# 610SAXLB
Cat# 61HISTLB
Cisbio
Experimental Models: Cell Lines
HEK293T
ATCC
https://www.atcc.org/Products/All/CRL-3216.aspx
Oligonucleotides
DNA
IDT
N/A
N/A
RNA
Dharmoncon
Recombinant DNA
(G4C2)66
Su et al., 2014
This study
This study
This study
Addgene
N/A
(TG4C2)66
N/A
(G4C2)66-NoATG-GFP
N/A
(G4C2)66-NoATG-Nanoluc
N/A
mCherry
Cat#54563
Cat#2302
pet15b expression vector encoding His-tagged hnRNP H1
Addgene
Software and Algorithms
GraphPad 7.03
Origin 7.5
Prism
https://www.graphpad.com/
Originlab
Bruker
https://www.originlab.com/
Topspin 2.1
https://www.bruker.com/products.html
Cell Chemical Biology 26, 1–12.e1–e12, February 21, 2019 e1

Please cite this article in press as: Wang et al., The Hairpin Form of r(G4C2)exp in c9ALS/FTD Is Repeat-Associated Non-ATG Translated and a Target
for Bioactive Small Molecules, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.10.018
CONTACT FOR REAGENT AND RESOURCE SHARING
Further information and requests for resources and reagents should be directed to the corresponding author, Prof. Matthew Disney
(disney@scripps.edu).
EXPERIMENTAL MODEL AND SUBJECT DETAILS
Cell Culture
HEK293T cells (female) were cultured in 13 DMEM containing 4.5 g/L glucose, 10% FBS, 2 mM L-glutamine, and 13 penicillin/strep-
tomycin (growth medium) at 37ꢁC in 5% CO2.
Authentication of Cell Lines Used
HEK293T cells were purchased from ATCC and used without further authentication.
METHOD DETAILS
Compounds
All 1a-like compounds were acquired from the National Cancer Institute (NCI). Compound identity was confirmed by mass spectros-
copy and compound purity was assessed by analytical HPLC. All compounds were >95% pure. Methods describing the synthesis of
1a-TOQ and 4-TO-PRO are provided in ‘‘Synthetic Methods’’, and their characterization is provided in Data S2. The G-quadruplex
ligands, PDS, cPDS (carboxylPDS), BER (Berberine) and Braco-19 were obtained from Sigma. CX-5461 was obtained from Adooq
Bioscience. The NMM (N-Methyl mesoporphyrin IX) and TMPyP4 compounds were obtained from Frontier Scientific and
ThermoFisher, respectively. PyPx-2-DMA was synthesized as previously described (Rzuczek et al., 2010). The BMVC derivatives
were provided by Dr. Ta-Chau Chang (IAMS, Academia Sinica).
Oligonucleotides
All oligonucleotides used in this study have been purchased from GE Dharmacon and de-protected according to the manufacturer’s
recommended procedure. Vivo-Morpholino oligonucleotide was acquired from Gene Tools.
Plasmids
The (G4C2)66, (G4C2)66-NoATG-GFP and (G4C2)66-NoATG-Nanoluc plasmids were constructed using previously described methods
(Gendron et al., 2013; Yang et al., 2015). To generate the (TG3C2)62 plamsid, gDNA from fibroblasts of a Sca36+ patients was used as
a templates in a nested PCR strategy using ThermalAce DNA Polymerase (Invitrogen) or AmpliTaq Gold 360 Polymerase (Thermo
Fisher). The sequence includes 69 bp 5’ of the repeat expansion and 40 bp of 3’ flanking sequence. The PCR products were cloned
into the pAG3 expression vector (gift of T. Golde, University of Florida), then sequentially ligated using TypeIIS restriction enzymes to
generate a (TG3C2)62 fragment. All (TGGGCC)n fragments with 5’ and 3’ flanking sequences were subcloned into the pAG3 expression
vector containing two upstream stop codons in each reading frame, as well as three different C-terminal tags in alternate frames
[i.e., (GP)n-HA, (GL)n-Myc and (WA)n-FLAG].
Affinity of 1a-TOQ for r(G4C2)8
The affinity of 1a-TOQ for r(G4C2)8 was completed by measuring the fluorescence intensity of 1a-TOQ as a function of r(G4C2)8 con-
centration. Briefly, the RNA was folded in 13 Assay Buffer (8 mM K2HPO4, pH 7.0, 185 mM KCl, and 1 mM EDTA) by heating at 95ꢁC
for 3ꢂ5 min and slowly cooling to room temperature for 15 min. BSA and 1a-TOQ were added to final concentration of 40 mg/mL and
1.2 mM, respectively. The RNA was then serially diluted into 13 Assay Buffer supplemented with 40 mg/mL BSA and 1.2 mM 1a-TOQ.
The resulting curve was fit to a one-site binding model to afford the Kd (Data S1).
Dye Displacement Assay to Study Binding of Compounds to r(G4C2)8
The RNA, r(G4C2)8, (600 nM final concentration) was folded as described above. Then BSA and 1a-TOQ were added to final concen-
tration of 40 mg/mL and 1.2 mM, respectively, followed by the compound of interest. The samples were incubated for 15 min at room
temperature and fluorescence was measured (excitation/emission wavelengths of 480/530 nm) using a BioTek FLx800 plate reader.
Background (emission of 1a-TOQ in absence of RNA) was subtracted, and the signal was normalized to 1a-TOQ complex with RNA in
the absence of compound.
IC50s were measured analogously by completing serial dilutions of the compound of interest in 13 Assay Buffer containing
600 nM r(G4C2)8, 1.2 mM 1a-TOQ, and 40 mg/mL BSA. IC50s were calculated from normalized data using a four-parameter logistic
curve fit (SigmaPlot).
Biolayer Interferometry (BLI)
BLI studies were performed similary a previously reported procedure (Su et al., 2014) on an Octet RED96 system. Briefly, 50-bio-
tinylated RNAs (60 nM) were folded by heating at 95ꢁC for 5 min followed by slowly cooling to room temperature in 13 BLI Buffer
e2 Cell Chemical Biology 26, 1–12.e1–e12, February 21, 2019

Please cite this article in press as: Wang et al., The Hairpin Form of r(G4C2)exp in c9ALS/FTD Is Repeat-Associated Non-ATG Translated and a Target
for Bioactive Small Molecules, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.10.018
(80 mM KH2PO4, pH 7.0, 185 mM KCl, 1 mM EDTA). Separately, serial dilutions of the compound of interest (ranging from 0.5 mM to
25 mM) were prepared in 13 BLI Buffer. The following incubation times were used during data acquisition (30ꢁC with shaking at
1000 rpm): baseline step (180 s), loading of RNA (900 s), washing (180 s), association of compound (500 s), and dissociation of com-
pound (350 s). The resulting curves were analyzed and processed using the Octet Data Analysis version 7.1 software by subtracting
the response of the sensors recorded upon incubation with solutions containing compound and no RNA (parallel reference). The data
were then globally fitted assuming reversible binding using the entire time interval for association and dissociation using the 2:1 het-
erogeneous binding or 1:1 binding curve fitting model.
Recombinant hnRNP H1-His6 Production, Isolation, and Purification
The pet15b expression vector encoding His-tagged hnRNP H1 was obtained from Addgene (ID: 23020) (Chou et al., 1999) and trans-
formed into Rosetta 2 (DE3) Escherichia coli cells. The transformed cells were grown in LB medium at 37ꢁC to OD600 = 0.6 before
induction with 0.5 mM IPTG and then cultured at 25ꢁC for 4 h. Cells were pelleted and lysed in 20 mM Tris-HCl, pH 7.5, and
500 mM NaCl (Buffer A) via sonication. The lysate was centrifuged and the pellet was washed with lysis buffer before it was solubilized
in 20 mM Tris-HCl, pH 7.5, 500 mM NaCl, and 6 M Urea (Buffer B). After centrifugation, the supernatant was filtered and loaded onto
His60 Ni Superflow resin (Clontech). The column was washed with 5 column volumes (CVs) of Buffer B. hnRNP H1 was refolded on
the column with a gradient from 100% Buffer B to 100% Buffer A with 20 CVs at a flow rate of 0.1 mL/min. Then the column was
washed with 10 CVs of 90% Buffer B and 10% Buffer C (20 mM Tris-HCl, pH 7.5, 500 mM NaCl, and 500 mM imidazole). The refolded
protein was eluted with a gradient to 100% Buffer C over 10 CVs.
Affinity of hnRNP H1 for r(G4C2) Repeats
The affinity of hnRNP H1 for r(G4C2) repeats was measured by a gel shift assay. The RNA was 50 end labeled with [g-32P]ATP
and T4 polynucleotide kinase and gel purified as previously described (Su et al., 2014). Briefly, r(G4C2)8 was folded in 13 Folding
Buffer (20 mM HEPES, pH 7.5, 110 mM KCl, 10 mM NaCl) by heating to 95ꢁC for 5 min followed by slowly cooling to room tem-
perature. The buffer was adjusted to 13TR-FRET Assay Buffer (20 mM HEPES, pH 7.5, 110 mM KCl, 10 mM NaCl, 0.1% (w/v)
BSA, 2 mM MgCl2, 2 mM CaCl2, 0.05% Tween-20, 5 mM DTT). The RNA was aliquoted and hnRNP H1-His6 (360 nM final
concentration) was added to the first aliquot followed by 1:2 serial dilutions. The samples were incubated at room temperature
for 1 h. The RNA and RNA-hnRNP H1-His6 complex were separated on a native 5% polyacrylamide gel prepared with 13TBE.
The percentage of complex formation as a function of hnRNP H1-His6 concentration was fit to a one-site binding model to afford
the Kd.
TR-FRET hnRNP H1 Binding Assay
The TR-FRET assays were performed as previously described (Disney et al., 2012) except the final concentrations of biotinylated
r(G4C2)8 RNA and hnRNP H1-His6 were 80 nM and 75 nM, respectively. The RNA was folded by heating to 95ꢁC for 5 min in 13
Folding Buffer followed by slowly cooling to room temperature. The buffer was adjusted to 13TR-FRET Assay Buffer followed by
addition of compound. The mixture was incubated for 15 min at room temperature and then hnRNP H1-His6 was added. After an
additional 15 min incubation, streptavidin-XL665 (HTRF, Cisbio Bioassays) and anti-His6-Tb (HTRF, Cisbio Bioassays) were added
to a final concentration of 40 nM and 0.44 ng/mL, respectively. After incubating for 1 h, TR-FRET was measured as previously
described. IC50s were calculated by fitting the resulting curve Equation 1 with GraphPad Prism.
A ꢂ B
IC50
y = B +
(Equation 1)
ꢀ
ꢁ
hillslope
1 +
x
Verification of RAN Translation Product in (G4C2)66–NoATG–GFP Plasmid by Western Blot
HEK293T cells were plated in 60 mm cell culture dishes and transfected with 4 mg (G4C2)66–NoATG–GFP or ATG–GFP plasmid using
JetPrime transfection reagent (Polyplus). After 4 h, cells were trypsinized and plated into 12-well plates in growth medium. After an
additional 2 h, the compound of interest or vehicle was added in growth medium. Cells were lysed in the plate using 300 mL/well of
mammalian protein extraction reagent (MPER, Pierce Biotechnology) containing 1 mL of halt protease inhibitor cocktail (Thermo Sci-
entific). Cellular proteins were separated by SDS-PAGE and then transferred to a PVDF membrane. Western blotting was completed
using anti-GFP (Santa Cruz) as a primary antibody and anti-IgG-horseradish peroxidase conjugate as the secondary antibody.
Chemiluminescent signal was generated by SuperSignal West Pico Chemiluminescent substrate (Thermo Scientific), and the blot
was imaged using X-ray film.
Cell-Based Assay to Assess RAN Translation and Poly(GP) Response
HEK293T cells (80% confluent) were batch-transfected in growth medium for 4 h with (G4C2)66–No ATG–GFP and a plasmid encod-
ing mCherry (for normalization) using JetPrime transfection reagent (Polyplus) according to the manufacturer’s instructions. The cells
were then seeded into a 384-well plate and incubated for 2 h before addition of compound using Biomek NXP Laboratory Automation
Cell Chemical Biology 26, 1–12.e1–e12, February 21, 2019 e3

Please cite this article in press as: Wang et al., The Hairpin Form of r(G4C2)exp in c9ALS/FTD Is Repeat-Associated Non-ATG Translated and a Target
for Bioactive Small Molecules, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.10.018
Workstation that was equipped with a 100 nL 384-pin head. After treatment for 24 h, the medium was removed, and the cells were
lysed in 100 mM potassium phosphate lysis buffer, pH 7.8, and 0.2% Triton X-100. Fluorescence was measured using a BioTek
FLx800 with 530/25 nm (excitation) and 590/35 nm (emission) filters for mCherry and 485/20 nm (excitation) and 528/20 nm (emission)
filters for GFP. Background was determined by measuring the corresponding fluorescence intensities in untransfected cells. The
background-corrected ratio of GFP to mCherry was used to determine the effect of compound on RAN translation. Poly(GP) mea-
surements were performed as previously described (Su et al., 2014). HEK293T cells were batch-transfected with (G4C2)66 or
(TG4C2)62 plasmid and treated with different concentrations of 4.
Effect of Compound Treatment on (G4C2)66–No ATG–GFP Transcript Levels
Approximately 23106 HEK293T cells were plated in 60 mm cell culture dishes and transfected with 3.5 mg (G4C2)66–No ATG–GFP
plasmid using JetPrime transfect agent per manufacturer’s protocol. After 4 h, cells were trypsinized and plated into 48-well plates
in growth medium. After an additional 2 h, the compound of interest or vehicle was added in growth medium. After overnight incu-
bation, the compound-containing medium was removed, and cells were washed with 13 Dulbecco’s phosphate-buffered saline
(DPBS). Total RNA was harvested with a Quick RNATM Mini-Prep Kit (Zymo Research) per the manufacturer’s recommended
protocol.
The cDNA was generated from 100 ng RNA using a QScript cDNA Synthesis Kit (Quanta Biosciences) according to manufacturer’s
recommended procedure; qPCR was completed with Power SYBR Green PCR Master Mix (Life Technologies) and 1 mL of 10-fold
diluted cDNA on an Applied Biosystems 7900HT Fast Real Time PCR System. The levels of (G4C2)66–No ATG–GFP were measured
with primers specific for GFP and normalized to 18S or b-actin values. Table S3 lists the sequences of all PCR primers.
CD Spectroscopy
CD spectroscopic experiments were conducted using a J-815 Jasco spectropolarimeter with a bandwidth of 2 nm at a scan speed of
50 nm/min and a step resolution of 0.2 nm over the spectral range of 210–450 nm. The RNA sample concentration was 10 to 20 mM in
10 mM Na2HPO4 buffer with 50 mM LiCl. Thermal melting curves were recorded by a peltier thermal coupler chamber (Jasco), moni-
tored at 265 nm between 20 and 95ꢁC with a temperature ramping rate of 1ꢁC/min. The observed signals were baseline subtracted,
and the first derivative zero points were defined as the melting temperature. Shape analysis of the melting curves yielded Van’t Hoff
enthalpy (DHvH) using standard procedures (Marky and Breslauer, 1987). The cooperativity of 4 binding to r(G4C2) repeats was
measured by plotting the normalized signal intensity at 300 nm or 330 nm in CD spectra as a function of the molar ratio of ([4]/
[RNA]) and fitted to the following equation:
nHill
y = x0 + ðxn Þ=ðx + K
nHill
Þ
(Equation 2)
Hill
where y is the normalized signal intensity, xo is initial molar ratio of ([4]/ [RNA]); x is the molar ratio of ([4]/ [RNA]), nHill is Hill coefficient,
and K is the Hill binding constant.
NMR Spectroscopy
All NMR spectroscopic experiments were performed on a Bruker AVIII 700 MHz NMR spectrometer equipped a cryoprobe (Bruker,
USA). 1D imino proton NMR spectra were recorded by a WATERGATE pulse sequence for water suppression. RNA concentrations
were typically 0.1 mM for 1D experiments and 0.4 mM for the 2D experiments, and studies were completed in 10 mM sodium phos-
phate buffer with 100 mM LiCl. Ligand were prepared 10 mM in D2O or d6-DMSO, and then added into a solution of RNA. For kinetic
experiments, RNA samples were prepared in 10 mM sodium phosphate buffer with 50 mM LiCl as the hairpin folded state, followed
by addition of 150 mM KCl. The time for recording an individual NMR spectrum was 30 min. Kinetic curves were fit to a three-state
kinetic model and were analyzed by Origin 7.0 and Topspin 2.0 (see below). An internal reference of 0.1 mM 4,4-dimethyl-4-silapen-
tane-1-sulfonic acid was used. The resonances of exchangeable protons were assigned using 2D NOESY with a mixing time
of 400 ms.
Analytical Solution of the Three-State Transition Model
A three-state transition model for hairpin state as H, unfolding state as U, and G-quadruplex state as Q was previously described
(Korobov and Ochkov, 2011) and is as follows:
k1
%
k3
%
HðtÞ
UðtÞ
QðtÞ
k2
k4
d½Hꢃ
dt
= ꢂ k1½Hꢃ + k2½Uꢃ
(Equation 3.1)
d½Uꢃ
dt
= k1½Hꢃ + k4½Qꢃ ꢂ ðk2 + k3Þ½Uꢃ
= k3½Uꢃ ꢂ k4½Qꢃ
d½Qꢃ
dt
e4 Cell Chemical Biology 26, 1–12.e1–e12, February 21, 2019

Please cite this article in press as: Wang et al., The Hairpin Form of r(G4C2)exp in c9ALS/FTD Is Repeat-Associated Non-ATG Translated and a Target
for Bioactive Small Molecules, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.10.018
Where k1 and k2 are the forward and reverse transition rates associated with the first step between the initial and intermediate states,
respectively, while k3 and k4 are the forward and reverse transition rates associated with the second step between the intermediate
and final states, respectively. The analytical solution for this three-state transition model is given as:
ꢂ
ꢃ
k2k4 k1ðg1 ꢂ k3 ꢂ k4Þ
k1ðk3 + k4 ꢂ g Þ
g1ðg2 ꢂ g1Þ2
1t
2t
HðtÞ = H0
ꢂ
eꢂg
ꢂ
eꢂg
(Equation 3.2)
(Equation 3.3)
(Equation 3.4)
g1g2
g1ðg2 ꢂ g1Þ
ꢂ
ꢃ
k4
k4 ꢂ g1
k4 ꢂ g2
1t
2t
UðtÞ = H0k1
ꢂ
eꢂg
ꢂ
eꢂg
g1g2 g1ðg2 ꢂ g1Þ
g2ðg2 ꢂ g1Þ
ꢂ
ꢃ
1
1
1
1t
2t
QðtÞ = H0k1k3
ꢂ
eꢂg
ꢂ
eꢂg
g1g2 g1ðg2 ꢂ g1Þ
g2ðg2 ꢂ g1Þ
Where g1g2 = k1k3+k2k4+k1k4 and g1 + g2 = k1 + k2 + k3+ k4. The initial conditions were H(0) = H0, U(0) = Q(0) = 0. The experimental
quantities associated with the initial and final states were extracted from the time-resolved spectra, which were used as input for
nonlinear regression to extract k1, k2, k3, and k4 in the Origin 7.5 software (OriginLab Corp., Northampton, MA, USA).
Kinetic data were obtained by analytical solution of the three-state transition model, where normalized kinetic curves showed the
imino proton signals of initial state of hairpin (H) and final state of G-quadruplex (Q) as a function of time, which were fitted by
Equations 3.2 and 3.4.
RNA FISH and Immunostaining
Nuclear Foci
The RNA FISH and hnRNP H immunostaining were completed as previously described with modifications (Lee et al., 2013). Briefly,
HEK293T cells (80% confluent) were batch-transfected as described above. The cells were then seeded into 12-well plates with cov-
erslips and incubated for 2 h before addition of compound. After incubation with compound for 24 h, the cells were fixed in 4% para-
formaldehyde for 15 min and washed 5 times with 13 DPBS. After washing with 13 DPBS for 15 min at room temperature, the cells
were washed with 0.1% Triton X-100 in 13 DPBS for 5 min at room temperature followed by washing with 40% formamide in 23 SSC
buffer for 15 min at room temperature. The RNA FISH probe was then added (5 ng/mL of 5’-Cy3-d(G2C4)8-3’) in 40% formamide in 23
SSC containing 8 mg/mL BSA, 66 mg/mL yeast tRNA, and 2 mM vanadyl complex. The samples were incubated at 48ꢁC and 5% CO2
overnight. Following overnight hybridization, the cells were washed three times with 23 SSC and then three times with 13 DPBS at
room temperature for 15 min each. To immunostain hnRNP H, cells were fixed with 2% (v/v) formaldehyde in 13 DPBS, permeabi-
lized with 0.1% (v/v) Triton X-100 in 13 DPBS, blocked with 5% goat serum for 40 min, and then incubated with a 1:250 dilution of
anti-hnRNP H (Novus Bio) in 13 DPBS for 1 h at room temperature. After washing with 13 DPBS twice, the cells were incubated with
a 1:250 dilution anti-rabbit IgG DyLight 650 in 13 DPBS at room temperature for 1 h. After washing (13 DPBS three times at room
temperature for 15 min each), cells were stained with 1 mg/mL DAPI in 13 DPBS for 10 min at room temperature and then washed
again with 13 DPBS immediately before imaging. Images were collected using an Olympus Fluoview 1000 confocal microscope. A
cell was considered foci-positive if green (RNA) and red (hnRNP H) fluorescence signal overlapped in punctate spots. ‘‘% Foci-pos-
itive Cells’’ was therefore calculated by dividing the number of foci-positive cells by the total number of cells analyzed and multiplying
by 100.
Foci Imaging by 4-TO-PRO and BG4, a G-Quadruplex Antibody
Briefly, HEK293T cells (80% confluent) were batch-transfected as described above. The cells were then seeded into 12-well plates
with coverslips and incubated for 2 h before addition of compound. After incubation with compound for 24 h, the cells were fixed in
2% paraformaldehyde for 15 min and washed five times with 13 DPBS. The cells were then washed with 0.1% Triton X-100 in 13
DPBS for 5 min and three times with 13 DPBS. 4-TO-PRO (0.5 mM) was added to the cells for 2 h at 37ꢁC.
For experiments employing 4 or RNase, the cells were first incubated with 4 (5 mM for 24 h) or RNase (25 mg/mL in 13 DPBS) at
37ꢁC for 2 h and washed two times with 13 DPBS before treatment with 4-TO-PRO (5 mM for 24 h). For BG4 studies (Moye et al.,
2015), the cells were fixed with 2% (v/v) formaldehyde in 13 DPBS, permeabilized with 0.1% (v/v) Triton X-100 in PBS, and blocked
with 2% (w/v) milk in 13 DPBS. After washing with 13 DPBS twice, the slides were incubated with BG4 primary antibody (600 nM) for
1 h at 37ꢁC, and washed three times for 5 min with 0.01% Tween-20 in 13 DPBS, and then overlaid with secondary antibody (Rabbit
anti-flag Tag antibody, Cell Signaling; 1:800 dilution) for 1 h at 37ꢁC. After washing three times with 0.01% Tween-20 in 13 DPBS, the
cells were incubated with anti-rabbit IgG DyLight 650. The cells were stained with 1 mg/mL DAPI in 13 DPBS for 10 min at room tem-
perature and then washed with 13 DPBS. Images were collected using an Olympus Fluoview 1000 confocal microscope. A cell was
considered 4-TO-PRO foci-positive if the cell contained punctate spots with green fluorescence. ‘‘% 4-TO-PRO Foci-positive Cells’’
was therefore calculated by dividing the number of 4-TO-PRO foci-positive cells by the total number of cells analyzed and multiplying
by 100.
Cell Chemical Biology 26, 1–12.e1–e12, February 21, 2019 e5

Please cite this article in press as: Wang et al., The Hairpin Form of r(G4C2)exp in c9ALS/FTD Is Repeat-Associated Non-ATG Translated and a Target
for Bioactive Small Molecules, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.10.018
Polysome Profiling
Polysome profiling studies were completed similarly to previously described (Yang et al., 2015, 2016). Briefly, 100 mm dishes were
plated with 4.53106 HEK293T cells and batch-transfected as described above. Approximately 2 h after plating, cells were treated
with vehicle or 4 (5 mM) and incubated for 16-18 h. Cells were then lysed with 250 mL of ice-cold Cell Lysis Buffer [10 mM NaCl,
10 mM MgCl2, 10 mM Tris, pH 7.5, 1% Triton X-100, 1% sodium deoxycholate, 1 mM DTT supplemented with 0.1 mg/L cyclohex-
imide and 0.2 U/mL RNAsin (Promega)]. The lysate was transferred to an Eppendorf tube, gently vortexed and centrifuged for 5 min at
13,200 rpm and 4ꢁC. The supernatant was transferred to a new tube and stored at ꢂ80ꢁC until further use.
Cellular lysates were separated using a 10-50% sucrose gradient in 500 mL fractions. A 100 mL aliquot from each fraction was
removed and the RNA isolated by using a Quick RNATM Mini-Prep (Zymo Research) according to manufacturer’s protocol. cDNA
was generated from 250 ng of RNA using a QScript cDNA Synthesis Kit according to manufacturer’s recommended procedure.
The levels of (G4C2)66–No ATG–GFP mRNA in each fraction were measured by qPCR as described above with primers specific
for GFP (Table S3). Data from vehicle- and compound-treated samples were normalized as follows: triplicate Ct values were aver-
aged and DCt values were calculated by comparison to b-actin; DDCt values were afforded by comparing treated to untreated
DCt values. Data were then normalized to the fraction with the lowest abundance of GFP.
Verification of RAN Translation Product in (G4C2)66–NoATG–NanoLuc Plasmid by SDS-PAGE and LC-MS
HEK293T cells were prepared as described above and transfected (G4C2)66–NoATG–NanoLuc or ATG-NanoLuc (pNL1.1) plasmid
using JetPrime transfection reagent. Cells were lysed in the plate 300 mL/well of MPER containing 1 mL of halt protease inhibitor cock-
tail. Cellular proteins were separated by SDS-PAGE. After coomassie staining, the bands correspoding to the molecular weight of
NanoLuc (ꢀ21 kDa) and Poly(GP)-NanoLuc (ꢀ35 kDa) were excised. The protein-containing gel was treated with 10 mM DTT
followed by 50 mM iodoacetamide, and subjected to trypsin digestion overnight. Prior to mass spectrometry analysis, the peptide
pools were acidified, desalted with Zip-Tip C18 tip columns and lyophilized to dryness.
Each sample was then reconstructed in 100 mL of 0.1% formic acid, and a 13 mL aliquot was analyzed by an Orbitrap Fusionꢀ
Tribridꢀ Mass Spectrometer (Thermo Fisher Scientific) coupled to an EASY-nLC 1000 system. Peptides were in-line eluted on an
analytical RP column (0.075 x 250 mm Acclaim PepMap RLSC nano Viper, Thermo Fisher Scientific), operating at 300 nL/min using
the following gradient: : 5–25% B for 40 min, 25–44% B for 20 min, 44-80% B in 10 s, 80% B for 5 min, 80-5% in 10 s, and 5% B for
20 min (solvent A: 0.1% formic acid (v/v); solvent B: 0.1% formic acid (v/v), 80% CH3CN (v/v)). The Orbitrap Fusion was operated in a
data-dependent MS/MS mode using the 10 most intense precursors detected in a survey scan from 380 to 1,400 m/z performed at
120K resolution. Tandem MS was performed by HCD fragmentation with normalized collision energy (NCE) of 30.0%. Protein iden-
tification was carried out using Mascot and Sequest algorithms, allowing Oxidation (Met) and Deamination (Q) as variable modifica-
tions. Other settings included Carbamidomethylation of Cys as a fixed modification, three missed cleavages, and mass tolerance of
10 and 20 ppm for precursor and fragment ions, respectively. MS/MS raw files were searched against a human database along with
porcine trypsin and sequence of protein of interest (NanoLuc).
In Vitro Translation Assay
The RNA template for translational assays was transcribed from the (G4C2)66–No ATG–NanoLuc plasmid using an in vitro transcrip-
tion system (RiboMax, Promega). In vitro translation reactions (50 mL) using the Flexi Rabbit Reticulocyte Lysate System (Promega)
were programmed with 4 mg (G4C2)66–No ATG–NanoLuc RNA and 1 mg Firefly Luciferase RNA (Promega) as templates. The RNA was
folded into the G-quadruplex form in buffer containing KCl (384 mM) by heating at 95ꢁC for 10 min followed by slow cooling to 25ꢁC
in a total volume of 10 mL. The RNA was folded into the hairpin form in buffer by placing on ice. (NMR spectroscopy shows that
only the hairpin is formed at low temperatures.) After folding, the RNA was added to the reticulocyte lysate, amino acid mixtures,
and RNasin ribonuclease inhibitor (Promega) in a total volume of 50 mL. Translation was allowed to proceed for 90 min at 30ꢁC,
and then luciferase activity was measured by diluting the reactions in GloLysis Buffer followed by addition of Nano-Glo substrate
(Promega).
Molecular Dynamics & Computational Methods
Parameterization of 4
A previously described protocol (Childs-Disney et al., 2013, 2014) was applied to study the binding modes of 4 to r(G4C2)exp, including
force field parameters for 4. The AMBER GAFF force field (Wang et al., 2005) was used to define the atom types while RESP charges
were derived following the RESP protocol (Bayly et al., 1993; Cornell et al., 1993). The molecule was optimized and the electrostatic
potential as a set of grid points was calculated at the HF level using the 6-31G* basis set, where Gaussian03 was used to perform
these calculations (Frisch et al., 2004).
Binding Studies
ThefoldedhairpinstructureofG4C2 hastwounique131GGmotifs;(i)5ʹ-CGG/5ʹ-CGG, and(ii)5ʹ-GGC/5ʹ-GGC. Thus, twomodelsystems
were designed to study the binding modes of 4 to RNA G4C2: (i) r(CCGCGGCGG)2, and (ii) r(CCGGGCCGG)2. For each case, the lowest
freeenergyboundstates of 4 were calculatedusing a dynamic docking methodologywepreviouslydesigned and successfully applied to
study binding modes of two compounds targeting RNA CUG and CCUG repeat expansions (Childs-Disney et al., 2013, 2014). For the
RNA, the amber99 force field (Cornell et al., 1995) with revised c (Yildirim et al., 2010) and a/g (Wales and Yildirim, 2017) torsional param-
eters was used. A modified implicit solvent model (GBOBC)(Onufriev et al., 2004) with 0.3 M salt concentration was used in all binding
e6 Cell Chemical Biology 26, 1–12.e1–e12, February 21, 2019

Please cite this article in press as: Wang et al., The Hairpin Form of r(G4C2)exp in c9ALS/FTD Is Repeat-Associated Non-ATG Translated and a Target
for Bioactive Small Molecules, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.10.018
simulations. The stem regions of RNA were restrained to sample around the A-form RNA by inclusion of Watson-Crick (WC) base pairing,
and torsional restraints, as well as chirality restraints to keep the chiral centers in proper orientations. The dynamic docking methodology
can be summarized as follows: The initial structures of the model RNAs with 131 GG motifs have been designed to be in A-form confor-
˚
mations where the 131 GG loop residues are in anti-anti state. Compound 4 molecules were placed 40 A away from RNA loops to create
the initial structure for the RNA-4 complexes. First, we created initial bound states by slowly moving 4 to the 131 GG loops of the RNA by
˚
increments of 1 A. The distance between the center of mass (COM) of the heavy atoms of the closing GC bases and the heavy atoms of
ring atoms of 4 was used as the reaction coordinate. During this step, WC base pairing and torsional restraints representing A-form RNA
˚
was imposed on the RNA except the loop regions to allow 4 to interact with the RNA freely. Once 4 was 0 A away from the RNA loops as
˚
described above, we slowly moved the compound away from the RNA by increments of 1 A. In this second step, WC base pairing and
torsional restraints representing A-form RNA was imposed including the loop regions. This process was repeated 50 times continuously,
which provided 50 initial bound states for our next step. We then ran 50 independent MD simulations in implicit solvent using the 50 initial
structures. Again, WC base pairing and torsional restraints representing A-form RNA were imposed on the RNA except the loop regions
so that 4 was free to sample around the initial conformations. The simulation time of each independent MD was 120 ns creating a total of
6 ms combined trajectory, which was used in cluster analyses. During the whole process, Langevin dynamics with collision frequency of 1
˚
was used with a long-range cutoff of 999 A. We utilized the pmemd.cuda to perform the MD simulations (Case et al., 2016).
Analyses
An in-house code was written to perform the cluster analyses. First, the 50 MD trajectories were combined for the cluster analyses
where root-mean-square deviation (rmsd) was used as the parameter to cluster the snapshots in the combined trajectories. Starting
˚
from the first structure in the combined trajectory, rmsd analyses were performed and snapshots with 1 A were clustered. Note that
the RNA duplexes are symmetric, which were considered during the cluster analyses. Once the clusters were created, MMPBSA
(Case et al., 2016) analyses were performed to calculate the relative binding free energies (see table below).
Atom names for 4 (see for atom types and charges).
Atom numbers, names, types, and charges for 4.
Atom
Atom
Atom
Type
Atom
Atom
Atom
Atom
Type
Atom
Number
Name
Charge
Number
Name
Charge
1
N1
C1
H1
C2
H2
C3
C4
C5
H3
C6
C7
H4
H5
H6
C8
C9
C10
C11
H7
nb
ca
h4
ca
ha
ca
ca
ca
h4
ca
c3
hc
hc
hc
ca
cp
ca
c3
hc
-0.697242
0.379613
0.056724
-0.471363
0.159024
0.281386
-0.328324
0.490060
0.030793
-0.081640
-0.180363
0.067851
0.067851
0.067851
0.084953
-0.109642
0.133682
-0.355354
0.119629
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
H8
hc
hc
na
hn
ca
cp
ca
ha
ca
ha
ca
os
c3
h1
h1
h1
ca
ha
0.119629
0.119629
-0.449642
0.382690
0.124979
0.194609
-0.300241
0.189262
-0.260461
0.169908
0.399860
-0.368024
0.025384
0.058026
0.058026
0.058026
-0.419302
0.182153
2
H9
3
N2
4
H10
C12
C13
C14
H11
C15
H12
C16
O1
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
C17
H13
H14
H15
C18
H16
Cell Chemical Biology 26, 1–12.e1–e12, February 21, 2019 e7

Please cite this article in press as: Wang et al., The Hairpin Form of r(G4C2)exp in c9ALS/FTD Is Repeat-Associated Non-ATG Translated and a Target
for Bioactive Small Molecules, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.10.018
A sample input file used to perform MMPBSA analysis.
============ mmpbsa.in ==============
MMPBSA.py input file for running PB and GB
&general
startframe=1, endframe=100000, interval=1,
keep_files=1, verbose =2,
/
&gb
igb=5, saltcon=0.300,surften=0.0072,surfoff=0.00, /
&pb
istrng=0.300, fillratio=4.0,inp=1,radiopt=0, indi=1.0,exdi=80,scale=2,linit=1000,prbrad=1.4, cavity_surften=0.0072,cavity_offset=0.00,
/
=====================================
Synthetic Methods
General Synthetic Methods
All reagents and solvents used for chemical synthesis were purchased from commercial sources and used without further purification
unless specified. NMR solvents were acquired from Cambridge Isotope Labs and used as received. Compound characterization is
provided in Data S2.
Instrumentation
1H- and 13C-NMR spectra were acquired on 400 MHz or 700 MHz Bruker Avance spectrometers. Chemical shifts (d) are reported in
ppm relative to tetramethylsilane or the respective NMR solvent; coupling constants (J) are in Hertz (Hz). Abbreviations used are s,
singlet; bs, broad singlet; d, doublet; dd, doublet of doublets; t, triplet; dt, doublet of triplets; td, triplet of doublets; tt, triplet of triplets;
bt, broad triplet; q, quartet; m, multiplet; and bm, broad multiplet.
Reverse-phase HPLC was completed using a Waters 1525 binary HPLC pump equipped with a Waters 2487 dual absorbance de-
tector system. Preparative HPLC separations were completed using Atlantis PrepT3 OBDTM 5 mM 193150mm column on a gradient
of 20% to 100% methanol (MeOH) + 0.1% trifluoroacetic acid (TFA) in H2O + 0.1% TFA over 60 min and a flow rate of 5 mL/min.
Analytical HPLC separations were completed using SunFire C18 3.5mM 4.63150mm column on a gradient of 20% to 100% meth-
anol (MeOH) + 0.1% TFA in H2O + 0.1% TFA over 60 min and a flow rate of 1 mL/min.
Flash chromatography was performed on a Biotage Isolera instrument using pre-packed silica columns purchased from Agela
Technologies.
Mass spectra were recorded on a Varian 500-MS IT mass spectrometer. High resolution mass spectra were obtained at the
Scripps Florida Mass Spectrometry and Proteomics Laboratory.
Synthetic Scheme for 1a-TOQ
Synthesis of 1a-TOQ
1a-amine (Su et al., 2014) and TOQ (Carreon et al., 2004) were synthesized as previously described. TOQ (12.5 mg, 22.5 mmol, 1.0 eq)
and HATU (8.3 mg, 22.5 mmol, 1.0 eq) were dissolved in anhydrous N,N-dimethylformamide (DMF; 50 mL), followed by the addition of
DIPEA (16 mL, 92 mmol, 4.1 eq). The mixture was vortexed and incubated for 1 min at room temperature. Then an aliquot from a 50 mM
1a-amine (22.5 mmol, 1.0 eq) stock solution in DMF was added. The mixture was stirred for 16 h at room temperature. The reaction
mixture was diluted with MeOH and water and purified by preparative HPLC. The product was isolated as a dark red solid in 63%
yield (11.4 mg).
1H NMR (400 MHz, DMSO-d6) d 11.97 (s, 1H), 10.02 (s, 1H), 9.42 (s, 1H), 8.75 (d, J = 8.0 Hz, 1H), 8.58 (d, J = 7.3 Hz, 1H), 8.48
(d, J = 7.1 Hz, 1H), 8.42 (d, J = 7.2 Hz, 1H), 8.11 – 8.00 (m, 2H), 8.01 – 7.91 (m, 1H), 7.83 – 7.77 (m, 2H), 7.73 (dd, J = 10.0,
4.6 Hz, 2H), 7.65 – 7.58 (m, 1H), 7.49 (d, J = 8.6 Hz, 1H), 7.42 (t, J = 7.6 Hz, 1H), 7.30 (d, J = 7.2 Hz, 1H), 7.13 (dd, J = 8.6,
2.3 Hz, 1H), 6.87 (s, 1H), 4.67 (t, J = 7.3 Hz, 2H), 4.51 (t, J = 7.3 Hz, 2H), 4.00 (s, 3H), 3.27 (s, 3H), 3.03 (q, J = 6.6 Hz, 2H), 2.80
(s, 3H), 2.01 (dt, J = 15.0, 7.4 Hz, 2H), 1.92 (t, J = 7.4 Hz, 2H), 1.80 (dt, J = 15.0, 7.5 Hz, 2H), 1.48 – 1.42 (m, 2H), 1.35 – 1.09 (m, 16H).
e8 Cell Chemical Biology 26, 1–12.e1–e12, February 21, 2019

Please cite this article in press as: Wang et al., The Hairpin Form of r(G4C2)exp in c9ALS/FTD Is Repeat-Associated Non-ATG Translated and a Target
for Bioactive Small Molecules, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.10.018
13C NMR (176 MHz, DMSO-d6) d 172.01, 160.03, 158.12 (q, TFA), 157.94 (q, TFA), 157.76 (q, TFA), 157.58 (q, TFA), 152.12, 148.48,
146,45, 146.35, 144.89, 144.25, 140.49, 136.96, 136.16, 133.54, 133.28, 132.10, 130.62, 128.25, 126.80, 126.15, 125.82, 124.57,
124.21, 123.90, 123.03, 122.90, 120.29, 119.75, 118.13 (CN, CH3CN), 118.03, 117.92 (q, TFA), 117.48 116.28 (q, TFA), 113.05,
112.14, 110.07, 109.79, 107.82, 88.08, 59.30, 54.94 (CH2, CH2Cl2), 54.14, 37.95, 35.39, 33.81, 30.44, 28.83, 28.78, 28.75, 28.65,
28.59, 28.53, 25.91, 25.25, 23.00, 14.92, 11.96, 1.18 (CH3, CH3CN).
HRMS: calculated for [C51H59N5O2S2+]: 402.7189; found: 402.7191
Synthetic Scheme for 4-TO-PRO
Synthesis of i1
A solution of 5-methoxyindole (5 g, 33.9 mmol, 1.0 eq), hexane-2,5-dione (6.4 mL, 54.2 mmol, 1.6 eq) and p-toluenesulfonic acid
(34 mg, 1.7 mmol, 0.05 eq) was sealed in a 20 mL microwave tube and irradiated for 20 min at 160ꢁC. The reaction mixture was trans-
ferred to a 250 mL round-bottom flask and recrystallized from ethanol. The crystallized product was washed with ice-cold ethanol
and dried under vacuum to provide i1 as a slightly red solid (5.11 g, yield = 67%).
1H NMR (400 MHz, CDCl3) d 7.93 (bs, 1H), 7.70 (d, J=2.4 Hz, 1H), 7.38 (d, J=8.7 Hz, 1H), 7.12 (d, J=7.3 Hz, 1H), 7.08 (dd, J=8.7,
2.4 Hz, 1H), 6.91 (d, J=7.3 Hz, 1H), 3.95 (s, 3H), 2.85 (s, 3H), 2.52 (s, 3H).
13C NMR (101 MHz, CDCl3) d 207.26, 153.67, 139.70, 134.44, 130.78, 126.14, 125.00, 121.38, 120.49, 117.17, 113.48, 110.94,
106.38, 56.21, 36.94, 29.96, 20.49, 16.60.
HRMS (ESI) calc. for C15H15NO [M+H]+, 226.1154; found 226.1228.
Synthesis of i2
Trimethylacetyl chloride (1.56 mL, 12.72 mmol, 1.5 eq) and sodium iodide (2.6 g, 17.3 mmol, 2.04 eq) were added to a solution of i1
(1.91 g, 8.48 mmol, 1.0 eq) in 21.4 mL acetonitrile at room temperature. The reaction mixture was heated to reflux (112ꢁC) for 3 h, the
solvent was removed under reduced pressure and the crude product was dissolved in ethyl acetate (100 mL). The solution was
Cell Chemical Biology 26, 1–12.e1–e12, February 21, 2019 e9

Please cite this article in press as: Wang et al., The Hairpin Form of r(G4C2)exp in c9ALS/FTD Is Repeat-Associated Non-ATG Translated and a Target
for Bioactive Small Molecules, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.10.018
washed with water (50 mL), a saturated solution of sodium thiosulfate (50 mL) and brine (20 mL). The organic layer was dried over
sodium sulfate, filtered and the solvent was removed under reduced pressure. The crude product was loaded on a 100 g pre-filled
Biotage cartridge and was eluted by applying a linear gradient from 4-34% EtOAc in hexanes providing i2 as a slightly brown solid
(1.56 g, yield = 63%).
1H NMR (400 MHz, CDCl3) d 8.10 (s, 1H), 7.73 – 7.69 (m, 1H), 7.34 – 7.28 (m, 1H), 7.11 (d, J=7.3 Hz, 1H), 7.06 – 7.01 (m, 1H), 6.88
(d, J=7.3, 1H), 2.72 (s, 3H), 2.50 (s, 3H), 1.48 – 1.43 (m, 9H).
13C NMR (101 MHz, CDCl3) d 178.49, 144.17, 139.70, 137.17, 130.96, 126.51, 124.73, 121.28, 120.83, 118.63, 117.24, 114.73,
111.04, 39.24, 27.45, 20.43, 16.68.
HRMS (ESI) calcd. for C19H21NO2 [M+H]+, 296.1572; found 296.1651.
Synthesis of i3
Phosphorous oxychloride (679 mL, 7.47 mmol, 1.5 eq) was added dropwise to N-methylformanilide (922 mL, 7.47 mmol, 1.5 eq) at
room temperature. After stirring for 30 min, a solution of i2 (1.47 g, 4.98 mmol, 1.0 eq) was added and the reaction mixture was heated
to reflux (140ꢁC) for 3 h. The solvent was removed under reduced pressure and the crude product was dried under vacuum for 90 min.
An aqueous solution of potassium acetate (10%) was added and the mixture was stirred for 16 h at room temperature. The product
was filtered and washed with 1% aqueous HCl (10 mL), water (10 mL) and a mixture of hexanes and toluene (1:1, 10 mL). After drying
under vacuum, i3 was obtained as a brown solid (1.53 g, yield = 95%).
1H NMR (400 MHz, DMSO-d6) d 11.82 (s, 1H), 10.36 (s, 1H), 7.89 (d, J=2.3 Hz, 1H), 7.68 (s, 1H), 7.59 (d, J=8.7 Hz, 1H), 7.18
(dd, J=8.7, 2.3 Hz, 1H), 3.07 (s, 3H), 2.55 (s, 3H), 1.36 (s, 9H).
13C NMR (101 MHz, DMSO-d6) d 191.47, 177.05, 144.12, 143.02, 138.05, 135.90, 128.25, 125.70, 123.59, 121.03, 119.75, 118.35,
115.17, 111.83, 38.54, 26.91, 16.54, 14.61.
HRMS (ESI) calcd. for C20H21NO [M+H]+, 324.1521; found 324.1602.
Synthesis of i4
A mixture of i3 (1.43 g, 4.42 mmol, 1.0 eq), aminoacetaldehyde diethylacetal (804 mL, 5.53 mmol, 1.25 eq), acetic acid (443 mL,
7.74 mmol, 1.75 eq) and sodium cyanoborohydride (486 mg, 7.74 mmol, 1.75 eq) in ethanol (34 mL) was heated to 83ꢁC for 5 h.
The reaction was quenched by the addition of a saturated aqueous solution of NaHCO3 (30 mL) and the product was extracted
with chloroform (3 x 30 mL). The combined organic extracts were washed with brine (20 mL), dried over sodium sulfate and the sol-
vent was removed under reduced pressure giving intermediate product as a brown oil (1.91 g, yield = 98%), which was used without
any further purification for the next step.
The intermediate product described above was dissolved in 11 mL THF, triethylamine (1.22 mL, 8.84 mmol, 2.0 eq) and p-tolue-
nesulfonyl chloride (927 mg, 4.86 mmol, 1.1 eq) were added at 0ꢁC. The cooling bath was removed, and the reaction mixture was
stirred at room temperature for 16 h. After the addition of ethyl acetate and water (each 20 mL), the phases were separated
and the aqueous layer was extracted with ethyl acetate (3 x 20 mL). The combined organic extracts were washed with 0.1 M HCl
(1 x 50 mL), saturated aqueous NaHCO3 solution (1 x 50 mL) and brine (1 x 50 mL), dried over sodium sulfate, filtered and the solvent
was removed under reduced pressure. The crude product was co-evaporated with silica, loaded on a 100 g pre-filled Biotage car-
tridge and was eluted by applying a linear gradient from 8-66% EtOAc in hexanes giving i4 (1.56 g, 59%) as a slightly brown solid.
(58% yield over two steps)
1H NMR (400 MHz, CDCl3) d 8.14 – 8.05 (m, 1H), 7.79 – 7.70 (m, 3H), 7.39 – 7.34 (m, 1H), 7.30 – 7.27 (m, 2H), 7.08 – 7.04 (m, 1H), 6.96
(s, 1H), 4.65 (s, 2H), 4.42 (t, J=5.4 Hz, 1H), 3.57 – 3.46 (m, 2H), 3.34 – 3.25 (m, 2H), 3.19 (d, J=5.4 Hz, 2H), 2.71 (s, 3H), 2.39 (s, 3H), 2.37
(s, 3H), 1.43 (s, 9H), 1.07 (t, J=7.0 Hz, 6H).
13C NMR (101 MHz, CDCl3) d 178.29, 144.29, 143.23, 139.40, 137.54, 137.48, 130.68, 129.62, 129.28, 127.41, 124.76, 124.27,
121.91, 118.83, 117.01, 115.19, 110.90, 102.07, 63.21, 50.74, 49.52, 39.23, 27.43, 21.58, 16.55, 15.92, 15.37.
HRMS (ESI) calcd. for C33H42N2O6S [M]+, 594.2764; found 594.2720.
Synthesis of i5
Sodium methoxide (78 mg, 1.45 mmol, 2.0 eq) was added to a solution of i4 (430 mg, 0.72 mmol, 1.0 eq) in 3.6 mL methanol and the
reaction mixture was stirred for 3 h at room temperature. The reaction was quenched by the addition of water (10 mL) and the product
was extracted with ethyl acetate (3 x 20 mL). The combined organic extracts were dried over Na2SO4 and the solvent was removed
under reduced pressure. The crude product was co-evaporated with silica and loaded on a 50 g pre-filled Biotage cartridge and was
eluted by applying a linear gradient from 10-80% EtOAc in hexanes providing i5 as a brown solid (274 mg, yield = 74%).
1H NMR (400 MHz, DMSO-d6) d = 10.82 (s, 1H), 8.87 (s, 1H), 7.77 (d, J=8.4 Hz, 2H), 7.53 (d, J=2.3 Hz, 1H), 7.44 (d, J=8.4 Hz, 2H),
7.32 (d, J=8.6 Hz, 1H), 6.91 – 6.87 (m, 2H), 4.47 (s, 2H), 4.04 (t, J=5.4 Hz, 1H), 3.37 – 3.26 (m, 5H), 3.09 (m, 2H), 3.02 (d, J=5.4 Hz, 2H),
2.69 (s, 3H), 2.41 (s, 3H), 2.38 (s, 3H), 0.90 (t, J=7.0 Hz, 6H).
13C NMR (101 MHz, DMSO-d6) d 150.29, 143.15, 139.47, 136.35, 134.23, 129.75, 129.53, 128.20, 127.15, 123.91, 122.26, 120.99,
116.82, 114.10, 111.22, 107.31, 100.53, 61.98, 50.77, 49.17, 20.98, 16.64, 15.33, 14.99.
HRMS (ESI) calcd. for C28H34N2O5S [M+H]+, 511.2188; found 511.2153.
e10 Cell Chemical Biology 26, 1–12.e1–e12, February 21, 2019

Please cite this article in press as: Wang et al., The Hairpin Form of r(G4C2)exp in c9ALS/FTD Is Repeat-Associated Non-ATG Translated and a Target
for Bioactive Small Molecules, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.10.018
Synthesis of i6
i5 (100 mg, 0.196 mmol, 1.0 eq) was dissolved in 2 mL acetone. After the addition of K2CO3 (81 mg, 0.392 mmol, 3.0 eq) and propargyl
bromide (80% in toluene, 37 mL, 0.392 mmol, 2.0 eq) the reaction mixture was heated to 80ꢁC and stirred for 16 hours. The reaction
was quenched by the addition of water (5 mL) and the product was extracted with EtOAc (3 x 10 mL). The combined organic extracts
were dried over Na2SO4, filtered and the solvent was removed under reduced pressure. The resulting yellow oil was purified by flash
chromatography applying a linear gradient from 8-66% EtOAc in hexanes, which provided i6 as a slightly yellow solid (93 mg, yield =
86% yield).
1H NMR (400 MHz, CDCl3) d 8.00 (s, 1H), 7.82 (d, J=2.4, 1H), 7.74 (d, J=8.3 Hz, 2H), 7.37 (d, J=8.8 Hz, 1H), 7.27 (d, J=8.0 Hz, 2H),
7.11 (dd, J=8.8, 2.4 Hz, 1H), 6.94 (s, 1H), 4.78 (d, J=2.4 Hz, 2H), 4.65 (s, 2H), 4.42 (t, J=5.4 Hz, 1H), 3.52 (dq, J=9.4 Hz, 7.0, 2H), 3.30
(dq, J=9.4 Hz, 7.0, 2H), 3.20 (d, J=5.4Hz, 2H), 2.77 (s, 3H), 2.55 (t, J=2.4 Hz, 1H), 2.39 (s, 3H), 2.36 (s, 3H), 1.07 (t, J=7.0 Hz, 6H).
13C NMR (101 MHz, CDCl3) d 151.68, 143.24, 139.53, 137.42, 135.48, 130.68, 129.59, 129.14, 127.39, 124.92, 123.91, 121.98,
116.97, 114.59, 111.06, 109.01, 102.07, 79.34, 75.46, 63.24, 57.49, 50.90, 49.52, 21.56, 16.53, 15.90, 15.35.
HRMS (ESI) calcd. for C31H36N2O5S [M+H]+, 549.2345; found 549.2286.
Synthesis of i7
A solution of i6 (111 mg, 0.202 mmol) in 1.5 mL dioxane and 220 mL of 6 M HCl was heated to reflux (120ꢁC) for 45 min. The solvent was
evaporated, 6 mL 1% aqueous NaCl solution were added and the suspension was stirred for 1 h. The product was filtered, washed
with water and rinsed with acetone. The product was removed from the frit and transferred to an Eppendorf vial, suspended in
acetone, centrifuged and the acetone was decanted. The acetone washing was repeated three times, and the product was dried
in vacuo to yield i7 as a yellow orange solid (39 mg, yield = 57%).
1H NMR (400 MHz, DMSO-d6) d 12.13 (s, 1H), 9.94 (s, 1H), 8.51 – 8.25 (m, 2H), 8.01 (d, J=2.4, 1H), 7.61 (d, J=8.8 Hz, 1H), 7.35
(dd, J=8.8, 2.4 Hz, 1H), 4.98 (d, J=2.4 Hz, 2H), 3.63 (t, J=2.4 Hz, 1H), 3.31 (s, 3H), 2.84 (s, 3H).
13C NMR (176 MHz, DMSO-d6) d 151.95, 144.77, 144.00, 137.66, 133.77, 133.65, 127.57, 125.65, 122.59, 119.56, 119.44, 117.73,
112.12, 110.12, 109.80, 79.60, 78.35, 56.54, 14.93, 12.06.
HRMS (ESI) calcd. for C20H16N2O [M+H]+, 301.1263; found 301.1340.
Synthesis of i8
i7 (10 mg, 33.3 mmol, 1 eq) and 3-azido-1-propanamine (10 mg, 100 mmol, 3 eq) were dissolved in a mixture of DMF:H2O=1:1 (4 mL in
total). Then ascorbic acid (2.64 mg, 15 mmol, 0.45 eq) and Cu(I) complex (3 mg, 5 mmol, 0.15 eq) were added under inert atmosphere
and the mixture was left to react overnight at room temperature. The solvent was removed and to the solid residue was added 5%
ethanolic HCl (5 mL) and the mixture was refluxed overnight at 98oC. Afterwards, the solvent was removed in vacuo and the resulting
product was re-dissolved in MeOH:H2O=1:4 and purified using preparative HPLC yielding the desired product as a yellow solid
(5.5 mg, yield = 41%).
1H NMR (700 MHz, DMSO-d6) d 9.97 (s, 1H), 8.42 (q, J = 6.9 Hz, 2H), 8.34 – 8.31 (s, 1H), 8.08 (s, 1H), 7.89 (s, 3H), 7.60
(d, J = 8.7 Hz, 1H), 7.37 (dd, J = 8.7, 2.3 Hz, 1H), 5.34 (s, 2H), 4.51 (t, J = 6.9 Hz, 2H), 3.32 (s, 3H), 2.83 (s, 5H), 2.18 – 2.07 (m, 2H).
13C NMR (176 MHz, DMSO-d6) d 158.70 (q, TFA), 158.50 (q, TFA), 158.29 (q, TFA), 158.09 (q, TFA), 152.91, 144.83, 144.15, 143.13,
137.42, 133.87, 133.66, 127.59, 125.78, 124.73, 122.71, 119.57, 119.48, 118.27 (q, TFA), 117.61, 116.61 (q, TFA), 114.95 (q, TFA),
113.30 (q, TFA), 112.21, 110.12, 109.33, 62.09, 46.65, 36.40, 27.89, 14.96, 11.99.
HRMS (ESI) calcd. for C23H25N6O [M+H]+ , 401.4940; found 401.2058.
Synthesis of 4-TO-PRO
TOQ (6.43 mg, 13.5 mmol, 1 eq) and HATU (5.14 mg, 13.5 mmol, 1 eq) were dissolved in N,N-dimethylformamide (DMF; 200 mL), fol-
lowed by addition of N-ethyldiisopropylamine (DIPEA; 5.2 mL, 3.85 mg, 29.76 mmol, 2.2 eq). The mixture was left to react for 15 min at
room temperature. Then, i8 (5.95 mg, 14.88 mmol, 1.1 eq) dissolved in DMF (400 mL) previously neutralized with DIPEA (2.2 eq) was
added and the mixture was stirred overnight at room temperature. The reaction mixture was concentrated under reduced pressure
and re-dissolved in a mixture of MeOH:H2O. The desired product was obtained using preparative HPLC as a dark red solid (7.9 mg,
yield = 68%).
1H NMR (700 MHz, DMSO-d6) d 12.04 (s, 1H), 9.91 (s, 1H), 8.68 (d, J = 8.7 Hz, 1H), 8.56 (d, J = 7.2 Hz, 1H), 8.36 (dd, J = 20.4, 6.9 Hz,
2H), 8.32 (s, 1H), 8.07 (d, J = 8.7 Hz, 1H), 8.03 (d, J = 2.1 Hz, 1H), 7.98 – 7.92 (m, 3H), 7.70 (t, J = 7.4 Hz, 2H), 7.57 – 7.55 (m, 1H), 7.51
(d, J = 8.7 Hz, 1H), 7.37 (t, J = 7.5 Hz, 1H), 7.28 (dd, J = 8.7, 2.3 Hz, 1H), 7.19 (d, J = 7.1 Hz, 1H), 7.04 (s, 1H), 6.76 (s, 1H), 5.30 (s, 2H),
4.52 (t, J = 7.5 Hz, 2H), 4.40 (t, J = 7.1 Hz, 2H), 3.93 (s, 3H), 3.29 (s, 3H), 3.07 (dd, J = 12.6, 6.6 Hz, 2H), 2.81 (s, 3H), 2.05 (t, J = 7.3 Hz,
2H), 1.97 (p, J = 6.9 Hz, 2H), 1.83 – 1.78 (m, 2H), 1.49 – 1.45 (m, 2H), 1.35 – 1.31 (m, 2H), 1.22 (s, 10H).
13C NMR (176 MHz, DMSO-d6) d 172.33, 159.72, 158.27 (q, TFA), 158.09 (q, TFA), 157.90 (q, TFA), 157.72 (q, TFA), 152.84, 148.27,
144.75, 144.13, 142.91, 140.29, 137.36, 136.89, 133.78, 133.58, 133.18, 128.09, 126.70, 125.74, 125.70, 124.57, 124.41, 124.13,
123.80, 122.77, 122.66, 119.52, 119.37, 117.99, 117.57 (q, TFA), 117.47, 115.88 (q, TFA), 112.87, 112.12, 110.02, 109.25, 107.76,
87.88, 62.08, 54.12, 47.35, 40.02, 35.63, 35.34, 33.68, 30.09, 28.77, 28.75, 28.70, 28.63, 28.50, 28.44, 25.82, 25.18, 19.70,
14.96, 11.99.
HRMS (ESI) calcd. for C52H57N8O2S+, 857.4320; found 857.4278.
Cell Chemical Biology 26, 1–12.e1–e12, February 21, 2019 e11

Please cite this article in press as: Wang et al., The Hairpin Form of r(G4C2)exp in c9ALS/FTD Is Repeat-Associated Non-ATG Translated and a Target
for Bioactive Small Molecules, Cell Chemical Biology (2018), https://doi.org/10.1016/j.chembiol.2018.10.018
QUANTIFICATION AND STATISTICAL ANALYSIS
Sample size was determined by studying variation in untreated samples. Each sample was randomly assigned as either treated or
untreated and were unblinded. Statistical significance was calculated by using a two-tailed Student t test (Prism’s GraphPad v. 7.03).
Differences between treatment groups were considered statistically significant when p < 0.05 (*p < 0.05, **p < 0.01, ***p < 0.001).
Total RNA harvested from cells was excluded from further analysis if the ratio of Abs260/Abs280 was <1.8, which indicates poor
quality. The number of replicates for each experiment is reported in the figure caption (nR3).
DATA AND SOFTWARE AVAILABILITY
N/A.
e12 Cell Chemical Biology 26, 1–12.e1–e12, February 21, 2019