Mechanistic Studies of a Small-Molecule Modulator of SMN2 Splicing

Mechanistic studies of a small-molecule modulator of SMN2 splicing

Jingxin Wanga, Peter G. Schultza,b,c,1, and Kristen A. Johnsona,1

aCalifornia Institute for Biomedical Research, La Jolla, CA 92037; bDepartment of Chemistry, The Scripps Research Institute, La Jolla, CA 92037; and cSkaggs Institute for Chemical Biology, The Scripps Research Institute, La Jolla, CA 92037

Contributed by Peter G. Schultz, April 2, 2018 (sent for review January 9, 2018; reviewed by Peter B. Dervan and Adrian R. Krainer) RG-7916 is a first-in-class drug candidate for the treatment of intronic splicing silencer (ISS) binding site at intron 7, leading to spinal muscular atrophy (SMA) that functions by modulating pre- an increase of SMN protein levels in motor neurons (19). Sig- mRNA splicing of the SMN2 gene, resulting in a 2.5-fold increase in nificant efforts have been made to develop an oral small mole- survival of motor neuron (SMN) protein level, a key protein lacking cule that can correct the splicing error of SMN2 exon 7 (12, 20, in SMA patients. RG-7916 is currently in three interventional phase 21). The pyrido-pyrimidinone RG-7916 is currently in phase 2 clinical trials for various types of SMA. In this report, we show that 2 clinical trials for various types of SMA. A pair of close analogs of SMN-C2 and -C3, close analogs of RG-7916, act as selective RNA- RG-7916, SMN-C2 and SMN-C3, correct exon 7 splicing with an binding ligands that modulate pre-mRNA splicing. Chemical proteo- EC50 ∼ 100 nM, presumably through the same mechanism (Fig. 1 mic and genomic techniques reveal that SMN-C2 directly binds to the A and B) (21). In type 2 and 3 SMA patients, RG-7916 was shown AGGAAG motif on exon 7 of the SMN2 pre-mRNA, and promotes a to induce a 2.5-fold increase in SMN protein levels in peripheral conformational change in two to three unpaired nucleotides at the blood cells (22). To understand the mechanism of SMN-C2/C3 in junction of intron 6 and exon 7 in both in vitro and in-cell models. exon 7 splicing and to guide the rational design of future splicing This change creates a new functional binding surface that increases modulators, we undertook an effort to elucidate the cellular target binding of the splicing modulators, far upstream element binding of SMN-C2/C3. Chemical proteomic and genomic studies re- protein 1 (FUBP1) and its homolog, KH-type splicing regulatory pro- vealed that SMN-C2/C3 directly binds to the AGGAAG motif on tein (KHSRP), to the SMN-C2/C3–SMN2 pre-mRNA complex and en- exon 7 of the SMN2 pre-mRNA, and promotes a conformational hances SMN2 splicing. These findings underscore the potential of change in two to three unpaired nucleotides at the junction of in- BIOCHEMISTRY small-molecule drugs to selectively bind RNA and modulate pre- tron 6 and exon 7 in both in vitro and in-cell models. We hypoth- mRNA splicing as an approach to the treatment of human disease. esize that this change increases binding of the splicing modulators, FUBP1 and its homolog, KHSRP, to the SMN-C2/C3–SMN2 pre- RG-7916 | RNA splicing | spinal muscular atrophy | RNA binding | mRNA complex and enhances SMN2 splicing. chemical proteomics SMN-C2/C3 Interacts with the Pre-mRNA of SMN2 pinal muscular atrophy (SMA) is one of the most common SMN-C3 has been documented to selectively regulate splicing Slethal genetic diseases in newborns (1). In the most severe of SMN2 along with only a handful other genes, such as STRN3 form of SMA (type I), infants usually do not survive beyond their first 2 y of life due to progressive hypotonia, leading to respiratory Significance failure (2). The cause of SMA in most type I patients is a recessive SMN homozygous deletion of the survival of motor neuron ( ) The development of small-molecule therapeutics that act by 1 gene in chromosome 5. With a reduced level of functional SMN targeting defined DNA or RNA sequences associated with hu- protein, the size of motor neurons is smaller, eventually causing man disease remains a challenge. RG-7916, a small-molecule muscle weakness (1). The exact mechanism of SMN in motor drug candidate for the treatment of spinal muscular atrophy neuron maintenance and survival has not been fully elucidated. (SMA), selectively regulates the alternative splicing (AS) of the One strategy to treat SMA is to promote increased expression SMN2 gene. Herein, we show that SMN-C2 and -C3, close an- SMN2 SMN1 of the endogenous gene , which is nearly identical to , alogs of RG-7916, act by binding SMN2 pre-mRNA and thereby to compensate for loss of the latter. However, the endogenous increasing the affinity of the RNA binding proteins far up- expression of SMN from SMN2 is ∼85% reduced (compared with stream element binding protein 1 (FUBP1) and KH-type splicing SMN1) primarily due to a pre-mRNA splicing error (3). The regulatory protein (KHSRP) to the SMN2 pre-mRNA complex. major transcription product of SMN2 leads to a shorter and These results suggest that nucleic acid targeted small molecules nonfunctional SMN protein primarily because exon 7 of SMN2 may have untapped potential for modulating disease processes pre-mRNA (abbreviated throughout as exon 7) is skipped in the at the level of pre-mRNA splicing. process of splicing (9), as a result of a C-to-T transition at position +6 on exon 7 (4). Despite the challenges associated with the de- Author contributions: J.W., P.G.S., and K.A.J. designed research; J.W. performed research; velopment of small molecule therapeutics targeting nucleic acids J.W. contributed new reagents/analytic tools; J.W. analyzed data; and J.W., P.G.S., and (5–12), several exon 7 splicing regulators have been identified K.A.J. wrote the paper. including the splicing activator, SRSF1 (13), the splicing repressor, Reviewers: P.B.D., California Institute of Technology; and A.R.K., Cold Spring Harbor Laboratory. hnRNP A1 (14), an SR-like protein, Tra2β1 (15), and various The authors declare no conflict of interest. members in hnRNP family (16). The secondary structure or higher-order folding of SMN2 pre-mRNA is also crucial for ac- Published under the PNAS license. curate mRNA splicing: a stem-loop structure at the 5′-splice site Data deposition: Next-generation sequencing data reported for Chem-CLIP and in-cell ′ SHAPE-MaP in this paper have been deposited in the Sequence Read Archive (SRA) da- (5 -ss) of exon 7, namely terminal stem-loop 2 (TSL2), has been tabase, https://www.ncbi.nlm.nih.gov/sra (accession no. SRP126430). shown to be a negative regulatory element for exon 7 splicing (17). 1To whom correspondence may be addressed. Email: [email protected] or kjohnson@ To date, Nusinersen is the only Food and Drug Administration- calibr.org. approved therapeutic for SMA and has been shown to improve This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. motor function in SMA patients after 15 mo of treatment (18). 1073/pnas.1800260115/-/DCSupplemental. Nusinersen is an antisense oligomer (ASO) that blocks an

www.pnas.org/cgi/doi/10.1073/pnas.1800260115 PNAS Latest Articles | 1of9 Downloaded by guest on September 30, 2021 F A N N N N N N N N N N N N N O O N N N N O N O SMN-C2 SMN-C3 SMN-C5

B exon 6 exon 7 exon 8 full length SMN2 mRNA functional SMN protein SMN-C2/ SMN-C3/ SMN-C5

SMN2 pre-mRNA

exon 6 exon 8 SMN2 Δexon 7 mRNA nonfunctional SMN protein DMSO

C N

H N H N O N HN S O O N O O H H O N N N N H 4 H O N N SMN-C2-BD

Fig. 1. (A) The structures of SMN-C2, SMN-C3, and SMN-C5 (the structure of RG-7916 has not been disclosed); (B) SMN-C2, SMN-C3, and SMN-C5 correct exon 7 splicing. (C) Structure of a biotin-diazirine bifunctional probe, SMN-C2-BD.

(21). Therefore, the direct target of SMN-C3 is unlikely to be active than its parent compound as demonstrated by Western the global splicing machinery. We hypothesized two potential blot analyses of SMN protein in SMNΔ7 mouse astrocytes (SI targets for SMN-C3: (i) a sequence of RNA on or close to exon Appendix,Fig.S1). 7, or (ii) a splicing regulatory protein or protein complex that is To capture the interacting RNA fragment, 293T cells were specific to exon 7. We therefore carried out a pull-down ex- treated with 2 μM SMN-C2-BD, in the absence or presence of periment to determine whether SMN-C3 directly binds to an 50× SMN-C3, and then cross-linked by irradiation (365 nm) of RNA target using the photo–cross-linking probe, SMN-C2-BD the live cells. Total RNA was extracted, pulled down by strep- (Fig. 1C), in which a biotin-diazirine bifunctional handle re- tavidin beads, washed, and bound RNA was released under de- places the ethyl group of SMN-C2. SMN-C2-BD is ∼20-fold less naturing conditions. The released RNA fraction was reverse

A 2

1 bits

0 12345678 B 7 6 CUUCCUUUAUUUUCCUUACAGGGUUUUAGACAAAAUCAAAAAGAAGGAAGGUGCUCACAUUCCUUAAAUUAAGGAGUAAGU 1 2 3 4 5 C Fluorescence polarization of 300 SMN-C2 and RNA 15-mers

oligo-1 200 oligo-2 oligo-3

mP oligo-4 100 oligo-5 oligo-6 oligo-7 0 -6.0 -5.5 -5.0 -4.5 -4.0 -3.5 Log[M]

− Fig. 2. (A) A purine-rich RNA binding motif of SMN-C2 identified by MEME analysis (24) with an E value of 1.3 × 10 151.(B) Sequences of synthetic RNA 15- mers. The coverage of each 15-mer is indicated with a bar above or below the sequences. Exon 7 is highlighted in bold with the putative binding sequence AGGAAG of SMN-C2 and a similar AAGGAG sequence at the 5′-ss in red and green, respectively. (C) Fluorescence polarization assays with SMN-C2 (200 nM)

and seven RNA 15-mers that cover the whole exon 7, part of the polypyrimidine tract of intron 6, and the junction of exon 7/intron 7 (n = 2). The Kd values of SMN-C2 binding to oligo-4 and -7 were determined to be 16 ± 2 and 46 ± 3 μM, respectively.

2of9 | www.pnas.org/cgi/doi/10.1073/pnas.1800260115 Wang et al. Downloaded by guest on September 30, 2021 Downloaded by guest on September 30, 2021 1lde G;8 laiehdoyi adr ,prfe 5 purified 9, ladder; hydrolysis alkaline 8, (G); ladder T1 es RA . C,1 M5 nM 10 KCl, M 0.5 tRNA, yeast Appendix i.3. Fig. age al. et Wang + 2o xn7 1∼ 7. exon of 32 ( A o h ulsqec fRNA. of sequence full the for tutrso SMN-C2– of Structures ) ,cevg ecinwt 0,4,ad0n SMN-C2 nM 0 and 40, 400, with reaction cleavage 3, ′ - B A 32 M-2PeC SMN-C2-GGH-Cu SMN-C2-Phe-Cu P 123456789 – hnadSMN-C2 and Phen aee N,01 H 0.1% RNA, labeled O N Cu HN H N 2 Cu N N NH NH O H N – ′ G opee ihcpe.( copper. with complexed GGH - 32 2 O P O SMN-C2-GGH-Cu – SMN-C2-Phe-Cu 2 aee N.Terbnces ecincnitdo 0m ee,p .,2 MMgCl mM 20 8.0, pH Hepes, mM 60 of consisted reaction ribonuclease The RNA. labeled H N Msdu sobt,adcpe-opee hmclpoea ifrn ocnrtos See concentrations. different at probe chemical copper-complexed and ascorbate, sodium mM 1 , N N N – Phen N – u 4 Cu; O ∼ ,cevg ecinwt 0,4,ad0n SMN-C2 nM 0 and 40, 400, with reaction cleavage 6, O N O B SMN-C2 ) N O N N – N Phen N – uadSMN-C2 and Cu ’-ACAAGGGTTTTAAGACAAAATCAAAAAAGAAAAGGAAAAGGTGCTCACATTCCTTAAATTAAGGAGT-3’ cleavage exon 7 start exon 7 bindingbi di siteit ends exon 7 – GGH – uecuieycev position cleave exclusively Cu NSLts Articles Latest PNAS – GGH – u ,RNase 7, Cu; 2 ,0.2 | μ 3of9 g/mL SI

BIOCHEMISTRY transcribed to cDNA by random hexamer extension and analyzed duct to form a truncated product at the primer extension stage by sequencing. Using the competitively treated cells as reference (30). The resulting band intensity of an unpaired nucleotide is control, alignment of the sequencing reads of the probe-only higher than a paired one by PAGE analysis of the radiolabeled sample to the human genome did not reveal an enrichment of cDNA fragments. exon 7. However, we were able to identify a binding motif for A 140-nt RNA clone of SMN2 exon 7 and adjacent regions was SMN-C2-BD in a genome-wide analysis of the captured pre- used as a template. This 140-nt RNA was imbedded within mRNA. A de novo motif searching for common sequences was a stabilizing cassette for in vitro SHAPE (30). The in vitro performed within the peaks of the aligned reads using MEME SHAPE-directed modeling revealed the following: (i) a stem-loop motif discovery algorithm (24). A purine-rich binding motif, structure (loop 2, Fig. 4 A and B)atthe5′-ss of exon 7 that was GAGGAAGA (Fig. 2A), was discovered with a satisfying E value previously reported in literature as TSL2 (17); (ii) a secondary − of 1.3 × 10 151. structure at the 3′-ss of exon 7 consisting of two bulges and a loop This putative binding sequence highly resembled the +24 to (Fig. 4 A and B), designated previously as TSL1 (17); and (iii)a +29 (AGGAAG) region of exon 7 (Fig. 2B). To validate the distal stem-loop structure on intron 7 (SI Appendix, Fig. S2 for full- binding specificity, seven RNA 15-mers were synthesized, which length SHAPE profile). Addition of SMN-C2 (50 μM) did not cover the entire exon 7 and adjacent intron 6 and 7 regions (Fig. change the overall secondary structure of exon 7 as determined by 2B). Each adjacent oligomer was designed to have a 5-nt overlap. in vitro SHAPE (Fig. 4A, lane 2). Importantly, no band was ob- An analog of SMN-C3, SMN-C2, contains a coumarin fluo- served for the sequence complementary to the AGGAAG binding rophore so that fluorescence polarization could be used to motif in the in vitro SHAPE analysis, suggesting that binding of measure the binding affinity between SMN-C2 and its target SMN-C2 to AGGAAG does not displace the complementary (Fig. 2C). As expected, the sequence containing the putative strand. However, the intensity of the bands corresponding to three binding site (oligo-4, Fig. 2 B and C), AGGAAG, showed more nucleotides in loop 1 and bulge 2 increased compared with DMSO than 10-fold decrease in Kd compared with sequences 2, 3, 5, and control (Fig. 4A), suggesting these nucleotides adopt a more 6 (Fig. 2 B and C). The lower binding affinity of oligo-3 to SMN- flexible conformation in the presence of SMN-C2. C2 suggests that high purine content alone is probably not an In vitro RNA secondary structure can be different from that in SMN-C2 recognition pattern. The pyrimidine-rich sequence, a cellular context primarily because of the equilibrium between oligo-1, showed the lowest binding affinity to SMN-C2 (Fig. 2 B naked RNA and RNA/RNA-binding protein complexes (31). and C). It has been shown that SMN-C5 also binds to an RNA Therefore, an in-cell SHAPE-mutation profiling (MaP) experi- duplex of 5′-ss of exon 7 and U1 snRNA (23). Although oligo- ment was carried out to investigate structural differences in exon 7 at the 5′-ss of exon contains an AAGGAG sequence that is 7 upon treatment with SMN-C2. First, 293T cells were trans- similar to the putative binding sequence AGGAAG, the binding fected with an SMN2 minigene to enrich the target RNA. The affinity of SMN-C2 to oligo-7 (Kd,46± 3 μM) is somewhat lower cells were subsequently treated with SMN-C2, a less active an- compared with binding of SMN-C2 to oligo-4 (Kd,16± 2 μM). alog, SMN-C2-Ac (SI Appendix, Fig. S3) or DMSO, and NAI. Next, we attempted to confirm the binding site of SMN- Total RNA was then reverse transcribed with a gene-specific + + C2 using an RNA footprinting experiment in the context of a primer. Replacing Mg2 with Mn2 in the buffer system results longer RNA sequence containing the entire exon 7. Iron or in an error-prone reverse transcription at the NAI-addition po- copper chelates in the presence of hydrogen peroxide have sition, creating mutations (32). Finally, a region of 276 nt con- previously been used as an artificial chemical ribonuclease to taining exon 7 was amplified by PCR with high fidelity for probe the nucleic acid binding sites of molecules (25). Conju- sequencing analysis. Analogous to in vitro SHAPE, a high mu- gation of the metal ion chelate to a nucleic acid binding molecule tation rate or high SHAPE reactivity translates to an unpaired leads to oxidative cleavage of the phosphate sugar backbone nucleotide or a more flexible conformation [see SI Appendix, Fig. adjacent to the binding site (26). Therefore, a conjugate of SMN- S4, for SHAPE reactivity and mutation rate calculated by C2 and 1,10-phenanthroline, SMN-C2–Phen (Fig. 3A), was ShapeMapper software (32) at single-nucleotide resolution]. synthesized with a two-carbon linker (27). To eliminate a po- The secondary structure within this 276-nt amplicon was pre- tential bias by the phenanthroline moiety in binding to dicted by matching the SHAPE reactivity for each nucleotide RNA, another copper ligand, Gly–Gly–His (28), conjugate with a software package, SuperFold (SI Appendix, Figs. S5 and was also synthesized, SMN-C2–GGH (Fig. 3A). In the presence of S6) (32). Specifically, the TSL2 region remains unchanged in the 32 sodium ascorbate and H2O2,a5′- P–labeled 120-nt RNA, which in-cell model but TSL1 undergoes some base pair rearrangement spans exon 7 (54 nt) and adjacent intron 6 and 7 regions, was around bulge 2 and loop 1 (Fig. 4B). RNA-binding proteins, such treated with submicromolar concentrations (40∼400 nM) of the as Tra2β1 (15) and hnRNP A1 (14), are known to bind to sub- SMN-C2–Phen- and SMN-C2–GGH–copper complexes. As sequences within TSL1 and may contribute to the base pair predicted, both SMN-C2–Phe–copper and SMN-C2–GGH– rearrangement. Despite this arrangement, the base pairing be- copper exclusively cleaved the RNA three nucleotides away tween AGGAAG and polypyrimidine tract is retained in the (Fig. 3B) from the putative binding site (29), AGGAAG. This cellular context (Fig. 4B). Like the in vitro SHAPE experiment, sequence-specific cleavage data confirmed the binding site treatment with 10 μM SMN-C2 results in significant SHAPE of SMN-C2. reactivity changes only in TSL1, but not TSL2 (SI Appendix, Fig. S7). An increase of SHAPE reactivity in position 34 and a de- SMN-C2 Does Not Disrupt the Secondary Structure of SMN2 crease in position 41 near the AGGAAG binding site, compared Pre-mRNA upon Binding with DMSO, indicates a more flexible conformation in position After the AGGAAG binding motif was identified for SMN-C2/ 34 and a more constrained one in position 41 (Fig. 4C). SHAPE C3, the next step was to determine whether any structural al- reactivity of a less active analog, SMN-C2-Ac, resembles that of terations are induced on exon 7 by SMN-C2 binding. The se- DMSO at 10 μM and does not show significant changes in po- lective 2′-hydroxyl acylation analyzed by primer extension sitions 34 and 41 (Dataset S1). In the 276-nt sequence, there is (SHAPE) method was used to interrogate the local nucleotide only one other secondary structure upstream of TSL1 which conformation of exon 7. SHAPE uses a 2′-OH alkylation re- showed significant SHAPE reactivity changes (SI Appendix, Fig. agent, 2-methylnicotinic acid imidazolide (NAI), to modify the S7). Thus, under both in vitro and in-cell conditions, the base- backbone ribose of an RNA sequence. A base-paired nucleotide paired SMN-C2–binding site AGGAAG on exon 7 and its is less accessible than an unpaired one to NAI modification; as a complementary sequence in the polypyrimidine tract on intron consequence, reverse transcriptase stops at the ribose-NAI ad- 6 forms a stem-loop structure, TSL1, at the junction of intron

4of9 | www.pnas.org/cgi/doi/10.1073/pnas.1800260115 Wang et al. Downloaded by guest on September 30, 2021 A B

Fig. 4. (A) In vitro SHAPE experiment with NAI and a 140-nt-long RNA template containing exon 7. 1, DMSO; 2, SMN-C2 (50 μM); 3, SMN-C2 (5 μM); 4, lacking BIOCHEMISTRY of NAI; 5∼8, ladders generated by addition of ddATP, ddTTP, ddCTP, and ddGTP during primer extension. PAGE was carried out on a TBE–urea sequencing gel at 60 W for 3 h. Red asterisks indicate increased band intensity with 50 μM SMN-C2. See SI Appendix for the full sequence of the RNA template. (B) In vitro and in-cell SHAPE-directed modeling of exon 7 and adjacent regions. For in vitro RNA model, SHAPE stabilizing cassette (orange) and nucleotides 1∼19 (blue) are shown in sketch. For in-cell RNA model, nucleotide numbering is aligned with in vitro SHAPE template. Nucleotides 1∼18 and 120∼140 were omitted. Sig- nificant reactivity changes are indicated in red and green asterisks for in vitro and in-cell SHAPE, respectively. The secondary structures that were previously named TSL1 and TSL2 (17) are enclosed in blue boxes. (C) Differential in-cell SHAPE reactivity in SMN2 minigene-transfected 293T cells for 10 μM SMN-C2 and DMSO in TSL1. SHAPE reactivity in single-nucleotide resolution and its SD were calculated by ShapeMapper software (32). Green asterisks indicate significant SHAPE reactivity change induced by 10 μM SMN-C2.

6 and exon 7. SMN-C2 treatment only affected the SHAPE re- SMN-C3. The pull-down of FUBP1 is dependent on endogenous activity of some unpaired nucleotides on TSL1 but not TSL2 or RNA as evidenced by a reduced FUBP1 signal when treating the other areas on exon 7. These changes revealed by SHAPE in- protein lysate with a mixture of RNase A/T1 (Fig. 5A). dicated that SMN-C2 does not disrupt the secondary structure of To confirm binding of SMN-C3 to FUBP1, a cellular thermal TSL1, but tunes the conformation of the nucleotides in the bulge shift assay (CETSA) (Fig. 5B) was conducted (35). In CETSA, and loop of TSL1. small molecules bind and protect their specific cellular protein targets from unfolding and aggregation when the cell suspension SMN-C2, SMN2 Pre-mRNA Exon 7, and FUBP1 Form a Ternary is heated (36). THP-1 cells were treated with SMN-C3 at various Structure concentrations or DMSO control and heated at 76 °C. After cell The interaction of SMN-C2/C3 with its binding motif, AGGAAG, lysis, cell debris and aggregated protein were removed by cen- does not explain the selectivity of SMN-C3 as a pre-mRNA splicing trifugation. The amount of FUBP1 in the supernatant increased modulator. There are more than 15 sequences of AGGAAG in 58% in the presence of 25 μM SMN-C3 compared with DMSO the SMN2 gene alone, in addition to a larger number within the control and was dose dependent as evidenced by Western blot whole human genome. To look for additional cofactors that may with an anti-FUBP1 antibody (Fig. 5B). These data suggest contribute to selectivity, a proteomic analysis was performed by SMN-C3 and FUBP1 interact in live cells. photo–cross-linking cell lysates in the presence of 2 μM biotin- Recombinant FUBP1 was then expressed and purified to diazirine bifunctional probe, SMN-C2-BD, followed by immu- further probe its interaction with SMN-C2 in vitro. A fluores- noblotting for biotinylated proteins (33, 34). After ammonium cence polarization experiment using recombinant FUBP1 and sulfate fractionation, the cell lysate was further resolved by two- SMN-C2 (Fig. 5C) revealed no apparent binding between up to dimensional SDS/PAGE. Western blots with anti-biotin antibody 20 μM FUBP1 and 200 nM SMN-C2. In contrast, FUBP1 in- revealed two associated proteins, FUBP1 and hnRNP A1 (SI duced higher polarization for SMN-C2 in the presence of either Appendix, Fig. S8). FUBP1 and hnRNP A1 are both known to a 15-mer containing the AGGAAG binding motif (oligo-4) or a play a role in pre-mRNA splicing. To confirm specificity for 120-nt RNA covering exon 7 and adjacent regions (Fig. 5C). This SMN-C2/C3, a protein pull-down was carried out using SMN-C2- result suggests a ternary complex is formed by FUBP1, SMN-C2, BD with or without unlabeled SMN-C3 competitor (Fig. 5A). and exon 7. To compare, 20 μM hnRNP A1 did not induce a FUBP1 was consistently competed with SMN-C3, whereas significant polarization compared with FUBP1 in the presence of hnRNP A1 only showed a basal signal level compared with 1% of oligo-4, but demonstrated a similar polarization increase with the the input control (Fig. 5A). The cross-linking of hnRNP A1 likely longer 120-nt RNA. This is likely due to the fact the longer se- results from the high abundance of this protein in the nucleus. quence covers both SMN-C2 and multiple hnRNP A1 binding This result suggests that FUBP1 is a potential protein target for sites (16). To confirm the ternary interaction, a 500-nt RNA

Wang et al. PNAS Latest Articles | 5of9 Downloaded by guest on September 30, 2021 ACQuantification of bands in pulldown Fluorescence polarization of SMN-C2 60 250 50 RNA: 40 200 FUBP1 oligo-4 30 hnRNP A1 20 120 nt RNA 150 10 mP

percentage of 1 % input 0 probe − + + + + 100 competition − − 10X 40X − RNase − − − − + pulldown 50 FUBP1 RNA +++− −− 1% input FUBP1 − +−+−− hnRNPA1 − −+− +− pulldown hnRNPA1 D SMN-C3 (μM) 1% input 12.5 0.39 0 B Quantification of FUBP1 bands in CETSA FUBP1 bound RNA 70000 Free RNA 60000 50000

RFL E SMN-C3 (nM) 40000 1000 333 111 37 12 4 0 30000 FL SMN 20000 FUBP1/ KHSRP Δ7 SMN dual knockdown SMN-C3 (μM)

Random siRNA FL SMN anti-FUBP1 Δ7 SMN

Fig. 5. (A) Biotin–streptavidin pull-down with SMN-C2-BD (2 μM)–cross-linked cell lysate visualized by Western blots with anti-FUBP1 and anti-hnRNP A1 antibody. The 20 or 80 μM SMN-C3 was used as a competitor. RNase A (10 μg/mL)/T1 (25 U/mL) mix was used to digest the endogenous RNA. The quantification of bands in Western blots was by the ratio of pull-down signal over 1% input signal in the same blot (n = 2). (B) Cellular thermal shift assay (CETSA) with THP-1 cells and a serial dilution of SMN-C3. The dose–response was observed optimal at 76 °C. FUBP1 that remained in the supernatant was visualized by Western blot with anti-FUBP1 antibody, and the bands were integrated by ImageStudio (n = 2). (C) Fluorescence polarization with SMN-C2 (200 nM), purified recombinant FUBP1 (20 μM) or hnRNP A1 (20 μM) and a 15-mer oligo-4 (used in Fig. 2C) or a 120-nt RNA that contains exon 7 (used in Fig. 3B). (D)EMSAwith10pmolof3′-biotin–labeled RNA (500 nt containing exon 7) and a serial dilution of SMN-C3. The gel was visualized by Northern blot with streptavidin–HRP (n = 2). (E)Dose–response of SMN-C3 (24 h) for end-point RT-PCR of FUBP1/KHSRP dual knockdown in SMN2 minigene-transfected 293T cells compared with a random siRNA control. FL SMN and Δ7 SMN were amplified by PCR with minigene vector-specific primers and resolved in a denaturing TBE–urea PAGE.

covering exon 7 was in vitro transcribed, 3′-labeled with biotin, SMN-C3 in a dose–response manner as analyzed by Western and subjected to an electrophoretic mobility shift assay (EMSA) blotting with an anti-KHSRP antibody (SI Appendix, Fig. S10). visualized by Northern blot with streptavidin–HRP conjugate. KHSRP also contains four KH domains and has been shown to SMN-C3 increased the bound fraction of the RNA (10 pmol) in regulate pre-mRNA splicing through cooperation with other the presence of a low concentration of recombinant FUBP1 splicing regulatory proteins (41, 42). The third and fourth KH (120 nM) in a dose–response manner (Fig. 5D). At this con- domains of KHSRP have been shown to independently interact centration of FUBP1, no protein-bound RNA was observed in with different regions of the AU-rich elements and recognize a EMSA without SMN-C3. Taken together, we conclude that broad set of mRNAs (43–45). SMN-C3 increases the binding affinity of FUBP1 and exon 7 by To further explore their roles, we knocked down FUBP1 and forming a ternary structure. KHSRP individually or in combination with siRNAs in SMN2 minigene-transfected 293T cells. The ratio of full-length SMN FUBP1 and KHSRP Are SMN2 Splicing Activators (FL SMN) to that which lacks exon 7 (Δ7 SMN) was analyzed by Next, the role of FUBP1 in SMN2 splicing was investigated. end-point RT-PCR with a pair of minigene vector-specific pri- FUBP1 has four KH domains that can bind either single- mers. Although FUBP1 and KHSRP alone did not significantly stranded DNA or RNA. The optimal recognition site of multi- affect SMN2 splicing (SI Appendix, Fig. S11), dual knockdown of ple KH-domain protein can be long and secondary structure both genes decreased the basal FL SMN mRNA level and shifted dependent. For example, a multi-KH domain protein, vigilin, has the dose–response curve of SMN-C3 (Fig. 5E). In summary, a 75-nt-long recognition sequence (37). In the literature, FUBP1 and KHSRP are both activators for exon 7 splicing. FUBP1 has been reported as both a pre-mRNA splicing activator and repressor (38–40). In the case of triadin exon 10 splic- Discussion ing, FUBP1 binds to an AU-rich motif downstream of a stem- Recently, small-molecule RNA-targeted therapeutics have drawn loop structure at the 3′-ss and inhibits the second step of splicing, considerable attention from both the pharmaceutical industry and that is, nucleophilic attack of the exon 3′-OH on the 3′-splice site academic laboratories (11, 46–48). One such molecule, RG-7916, to release the lariat (40). Due to the high homology (61%) of was originally identified through a phenotypic high-throughput FUBP1 and KH-type splicing regulatory protein (KHSRP) (SI screen. An optimized analog is currently being evaluated in type Appendix, Fig. S9), especially at the conserved RNA binding KH I, II, and III SMA patients in multiple phase 2 clinical studies. domains, we hypothesized that the two proteins likely play a Comprehensive proteomic, genomic, and structural studies by similar role in modulating SMN2 splicing by SMN-C3. Indeed, our laboratory and Sivaramakrishnan et al. have led to an un- photo–cross-linking pull-down experiments confirmed that SMN- derstanding of the selective actions of RG-7916 on splicing C2-BD can also specifically bind to KHSRP and is competed by of SMN2 pre-mRNA.

6of9 | www.pnas.org/cgi/doi/10.1073/pnas.1800260115 Wang et al. Downloaded by guest on September 30, 2021 Sivaramakrishnan et al. (23) recently proposed that the RG- were immediately washed with PBS once on ice and then collected by 7916 analog, SMN-C5, binds to two distinct sites of the SMN2 scraping. After centrifugation, total RNA of the cell pellets was extracted pre-mRNA and stabilizes an unidentified ribonucleoprotein by using the TRIzol reagent (Ambion) and RNeasy mini columns (Qiagen) (RNP) complex that is specific to SMN-C5–SMN2 pre-mRNA with on-column treatment of DNase (#79254; Qiagen). The cross-linked RNA (100 μL in TBE buffer) was captured by incubation with prewashed complex. NMR analysis showed that SMN-C5 induces chemical μ ′ streptavidin agarose (20 L; #S1638; Sigma) in the presence of RNasin ribo- shift perturbations in 7 nt in the 5 -ss of exon 7-U1 snRNA duplex nuclease inhibitor (40 U/mL; #N2111; Promega) for 4 h at 4 °C with rotation. and causes the broadening of one imino signal. In addition, SMN- The beads were washed with TBE buffer (100 μL × 3) and heated with 1× C5 was shown to bind to immobilized ESE2 fragment of exon elution buffer (50 μL; 95% formamide, 10 mM EDTA, pH 8.2) at 80 °C for 7 using label-free surface plasmon resonance (SPR) technology. 5 min. The released RNA was purified by RNeasy mini kit by adding 300 μLof With this ESE2 fragment, several imino signals in 1H NMR were RLT buffer and completed by following the manufacturer’s protocol. cDNA broadened in the ESE2 region with chemical shift perturbations, was produced from 1 μg of RNA using a qScript cDNA Synthesis Kit (#95048; which indicated that the conformation of ESE2 was likely changed Quantabio). The resulting cDNA was desalted by NucAway columns in the presence of SMN-C5. The authors argued that binding of (#AM10070; Thermo Fisher). Chem-CLIP-seq samples were prepared by using ′ ScriptSeq, version 2, RNA-Seq library preparation kit (Epicentre) following SMN-C5 at the 5 -ss region is more important than ESE2 based manufacturer’s protocol. The libraries were sequenced on a NextSeq on mutagenesis of the SMN2 minigene. 500 using 1 × 75 single-end reads to generate ∼17 million reads per sample. In our study, the AGGAAG motif, which is almost identical to The sequencing result was mapped and aligned with human genome by ESE2, was identified by RNA pull-down and RNA affinity Bowtie2 algorithm and analyzed by motif discovery function (MEME) in cleavage experiments using probes based on the structure of MEME suite (meme-suite.org) (24). SMN-C3, a close analog of RG-7916. In-cell SHAPE-MaP analysis showed that the most significant conformational Protein Pull-Down. Four 15-cm dishes of 293T cells (90% confluency) were change induced by SMN-C2 is in TSL1, which contains ESE2, trypsinized, pelleted, and washed with PBS once before addition of 2 mL of · but not the 5′-ss. SMN-C2 does not disrupt the secondary ice-cold lysis buffer [50 mM Tris HCl (pH 7.5), 100 mM NaCl, 2 mM MgCl2, 2 mM CaCl2, 0.2% Nonidet P-40, 0.8% Triton X-100, 1 mM β-mercaptoe- structure of TSL1 but tunes the conformation of nucleotides in thanol, 1× protease inhibitor (Roche), and 40 U/mL RNasin]. The cells were the bulge and loop of TSL1. In agreement with previous publi- suspended in the lysis buffer, incubated for 10 min at 4 °C, and homoge- cations, mutagenesis by deleting AGGAAG on exon 7, which is a nized by sonication (output 2, duty cycle 10%) for 1 min. The cell lysate was conserved sequence for the splicing activator Tra2β1 (15), or its cleared by centrifugation (18,000 × g) at 4 °C for 10 min, and the superna- complementary sequences at polypyrimidine tract in intron 6 in a tant was rotated with 20 μL of prewashed streptavidin agarose (#S1638; SMN2 minigene, resulted in abrogation of exon 7 splicing (SI Sigma) at 4 °C for 2 h. The total protein concentration of the resulting su- BIOCHEMISTRY Appendix, Fig. S12). In in vitro experiments, we measured pernatant was measured by BCA protein assay kit (#23225; Thermo Fisher) ′ and normalized to ∼5 mg/mL. The resulting cell lysate was aliquoted into binding of SMN-C2 to oligos containing the 5 -ss, but it is weaker μ μ than that to TSL1 site. 400 L per sample and incubated with or without RNase A/T1 mix (2 L; #EN0551; Thermo Fisher) at 4 °C with rotation for 30 min. The protein Because of the relatively large number of AGGAAG motifs in mixture was then transferred into a 24-well plate, compounds added (2 μM the human genome, and the selective effect of RG-7916 on SMN-C2-BD, the designated concentrations of SMN-C3 or DMSO control) and mRNA splicing, it is likely that other factors must influence the shaken at room temperature (80 rpm) for 30 min before UV radiation selectivity of the drug for SMN2. To identify effector proteins (365 nm) for 20 min in a UV cross-linker (StrataLinker). The mixture in each that may also play a role in RG-7916 action, Sivaramakrishnan well was collected in 1.5-mL Eppendorf tubes, added 40 μL of prewashed et al. used immobilized RNA fragment of ESE2 and identified its streptavidin agarose beads and rotated overnight at 4 °C. Four microliters of interaction with hnRNP G. The binding of hnRNP G was par- protein solution from each sample served as the input control. The beads μ × tially blocked by SMN-C5 in an SPR-based binding assay, yet the were washed three times with lysis buffer before boiling with 80 Lof2 Laemmli buffer at 95 °C for 10 min with occasional gentle flicks. The 40-μL functional role of hnRNP G in pre-mRNA splicing was not fully sample was loaded in each well of a denaturing SDS/PAGE gel (#NP0323PK2; elucidated (23). In our study, the splicing factor FUBP1 was Thermo Fisher). The samples were processed according to the methods de- – identified in a whole-cell photo cross-linking experiment using scribed in SI Appendix, Western Blot. Anti-FUBP1 and anti-KHSRP antibodies SMN-C2-BD. We hypothesize that conformational changes in- were purchased from EMD Millipore (#ABE1330 and MABE987). duced by SMN-C2 or -C3 binding lead to the partial displace- ment of hnRNP G, and enhance pre-mRNA recognition by both Fluorescence Polarization Assay. For measurement of the binding affinity of FUBP1 and the highly homologous splicing factor KHSRP, as 15-mer RNA and SMN-C2, synthetic 15-mers were reconstituted in DEPC- demonstrated by fluorescence polarization and EMSA. The treated water at 500 μM. A serial dilution of each RNA oligomer was pre- splicing activation effect of FUBP1/KHSRP was further vali- pared in 20 μL of DEPC-treated water. The samples were heated at 80 °C for × μ dated by siRNA knockdown in the presence of SMN-C3. 3 min and snap cooled on ice for at least 1 min. Binding buffer (2 ;20 L, 100 mM Hepes, 300 mM NaCl, 8 mM MgCl2) with 200 nM SMN-C2 was added In conclusion, the combination of our studies and the previous into each sample, mixed well by pipetting up and down, and transferred work of Sivaramakrishnan et al. suggest that binding of SMN-C2/ into a clear-bottom 96-well plate. For measurement of the fluorescence C3 to the exon 7 AGGAAG motif and the resulting effect on anisotropy to FUBP1 and hnRNPA1, 20 μM oligo-4 or 1 μM 120-nt RNA (for hnRNP G/FUBP1/KHSRP binding contribute for the specificity sequence, see SI Appendix) were snap-cooled and mixed with 200 nM SMN- of SMN-C2/C3 in modulating SMN2 splicing. In the literature, C2 and 20 μM recombinant FUBP1 (SI Appendix) or hnRNP A1 (#ab123212; there are 46 other proteins shown to be relevant for exon Abcam) in a 96-well plate. The working solution of the proteins was ex- 7 splicing (16). It cannot be ruled out that other proteins might changed into 1× binding buffer using a Zeba spin desalting column (#89882; ’ also recognize the binding of SMN-C2/C3 and exon 7 and reg- Thermo Fisher) according to the manufacturer s protocol before use. The ulate SMN2 splicing. Further mechanistic studies should include plate was read after 5-min equilibration at room temperature using a plate reader with fluorescence polarization function (excitation/emission, 400/ the elucidation of the recognition sequence of FUBP1 and 480 nm). Sequences of synthetic RNA oligomers are listed in SI Appendix. KHSRP and their regulatory splicing mechanism, but are beyond the scope of this paper. Chemical Ribonuclease Cleavage with Hydroxyl Radical. A small-molecule probe–copper complex was made by mixing SMN-C2–Phe or SMN-C2–GGH Materials and Methods (2 mM, 2 μL) with freshly prepared CuSO4 (1 mM, 2 μL), and diluted to 2 or RNA Pull-Down (Chem-CLIP)-Seq. Chem-CLIP protocol was modified from 0.2 μM working solution. In each reaction sample, purified 5′-32P –labeled published procedures (49). Briefly, a 10-cm dish of 293T cells (80% con- RNA (1 μL, >50 cpm; see SI Appendix for sequence and labeling protocols) fluency) were treated with SMN-C2-BD (2 μM) or SMN-C2-BD (2 μM) plus was incubated at 80 °C for 3 min and snap cooled on ice for at least 3 min SMN-C3 (50 μM). The cells were incubated at 37 °C for 30 min before ex- before mixing with the copper complex (0.5 μL), Hepes (1 μL, pH 7.5,

posure to UV (365 nm) for 20 min in a UV cross-linker (StrataLinker). The cells 300 mM), MgCl2 (1 μL, 100 mM), yeast tRNA (1 μL, 1 μg/μL), and KCl (0.5 μL,

Wang et al. PNAS Latest Articles | 7of9 Downloaded by guest on September 30, 2021 3.0 M). The mixture was incubated at 37 °C for 20 min and H2O2 (0.5%, 1 μL), S2 sonicator (intensity setting, 5; duty, 10%; burst cycles, 200; 5 min with and sodium ascorbate (5 mM, 1 μL) was added in order. The mixture was frequency sweeping mode). Ten nanograms of fragmented DNA products incubated at 37 °C for another 20 min before being quenched with thiourea were then treated with NEBNext Ultra II DNA Library Prep Kit for Illumina (50 mM, 1 μL) and immediately precipitated with ammonium acetate (1 μL, platform following manufacturer’s protocol with 9 cycles of PCR. The li- 3 M) and ice-cold EtOH (100 μL). The mixture was placed at −20 °C for 30 min braries were cleaned up using 0.9× AmpureXP beads and then sequenced on and subsequently centrifuged (18,000 × g) at 4 °C for 10 min. The super- a NextSeq 500 using 1 × 75 single-end reads to generate ∼10 million reads natant was removed and the RNA pellet was air-dried for 5 min before per sample. The analysis pipeline was performed using the software pack- loading onto a 10% TBE–urea sequencing gel. The electrophoresis was ages developed by Weeks and coworkers (32). performed at 60 W for 90 min. The gel was subsequently dried on a gel heater and exposed overnight to a phosphor storage screen. CETSA. The experimental protocol from literature (35) was followed except that THP-1 cells were used with a denaturing temperature at 76 °C. Anti- In Vitro SHAPE. The experimental protocol from literature (30) was followed FUBP1 antibody was purchased from EMD Millipore (#ABE1330). except that NAI was used for RNA 2′-OH acylation at 37 °C for 30 min. See SI Appendix for the sequence of 140-nt RNA template flanked by stabilization FUBP1/KHSRP Knockdown. Thirty picomoles of siRNA of FUBP1 (#s16967; cassettes and a primer binding sequence at 3′-end. Thermo Fisher), KHSRP (#s16322; Thermo Fisher), or randomized RNA control (#4390843; Thermo Fisher) were cotransfected with 0.2 μg of SMN2 mini- In-Cell SHAPE-MaP. The experimental protocol for in-cell SHAPE-MaP exper- gene (pCI-SMN2; #72287; Addgene) and 5 μL of Lipofectamine RNAiMAX iment was modified from the one developed by Weeks and coworkers (32). (#13778030; Thermo Fisher) into one well of 293T cells (70% confluency) in a Briefly, a 10-cm dish of 293T cells (70% confluency) was transfected with six-well plate following manufacturer’s protocol. After 24 h, the cells were SMN2 minigene (10 μg; pCI-SMN2) using FuGENE HD (50 μL; Promega) with trypsinized and replated into a 48-well plate containing a serial dilution of manufacturer’s protocol. At 24-h posttransfection, the cells were trypinsized SMN-C3. After 24-h incubation with the compound, RNA was extracted from and aliquoted into 106 cells per sample (200 μL of DMEM with 1% FBS) in each well by RNeasy mini kit and amplified by SuperScript III Platinum One- 1.5-mL Eppendorf tubes that contain compounds (in DMSO, final concentra- Step qRT-PCR Kit (#11732020; Thermo Fisher) with primers (forward, 5′- tion at 2 μM) or DMSO control. The tubes were incubated at 37 °C for 30 min TACTTAATACGACTCACTATAGGCTAGCCTCG; reverse, 5′-GTATCTTATCATG-

with gentle vortexing every 5 min. The suspension of cells in each Eppendorf TCTGCTCG) for 35 cycles (Ta = 60 °C). The PCR product was resolved on 6% tube was then immediately transferred into a new tube containing 10 μLof TBE–urea gel followed by soaking with SYBR-Safe dye for visualization. NAI solution (2 M in DMSO, #03-310; EMD Millipore). The resulting suspension was incubated at 37 °C for 15 min with gentle vortexing every 5 min. EMSA. LightShift Chemiluminescent EMSA kit (Thermo Fisher) was used for × The samples were centrifuged at 300 g for 2 min, and the supernatant was EMSA with 120 nM purified recombinant FUBP1, 10 pmol of biotinylated RNA μ removed. The cells were washed twice with PBS (500 L) to remove the (500 nt), and a 1:2 serial dilution of SMN-C3 following manufacturer’s pro- μ residue of NAI before homogenizing with TRIzol (800 L). To each sample, tocol. See SI Appendix for the sequence and the expression protocol of the μ CHCl3 (200 L) was added and the tubes were vigorously vortexed and recombinant FUBP1 and the sequence of the RNA. centrifuged at 12,000 × g for 5 min. The aqueous partition in each tube containing the total RNA was mixed with 1.5 vol of EtOH and extracted with Note Added in Proof. While this manuscript was in preparation, Sivaramakrishnan RNeasy mini kit (#74104; Qiagen) following the manufacturer’s protocol. + et al. (23) showed that another active analog of RG-7916, SMN-C5 (Fig. 1A), Reverse transcription was performed with Mn2 /SuperScript II reverse tran- interacts with both exonic splicing enhancer 2 (ESE2) and the 5′-ss of exon 7. scriptase (32), 1 μg of RNA, and a gene-specific primer (5′-TGTTTTACAT- The location of ESE2 on exon 7 is almost identical to the AGGAAG motif TAACCTTTCAACT). Amplicons containing SMN2 exon 7 were then amplified identified in this report. These results collectively provide important insights through PCR with the primer pair (5′-AATGTCTTGTGAAACAAAATGCT and into the mechanism of this potential SMA therapeutic. 5′-AACCTTTCAACTTTCTAACATCT), AccuPrime pfx DNA polymerase, and the cDNA as template. The amplicons were purified by agarose gel (1.8%) and ACKNOWLEDGMENTS. We thank Drs. James Williamson, John Hammond, reconstituted in deionized water. Four hundred nanograms of each PCR Rebecca Berlow (The Scripps Research Institute), Kevin Weeks, Steve Busan amplicon was concatenated for 20 °C for 2 h before EtOH precipitated at (University of Northern Carolina, Chapel Hill), Gene Yao, Stefan Aigner, Frederick −80 °C for overnight. DNA pellets were washed with 80% ice-cold EtOH and Tan (University of California, San Diego), and Shoutian Zhu (California Institute resuspended in 50 μL of buffer containing 10 mM Tris·HCl, pH 7.5, 20 mM for Biomedical Research) for discussion and help on experimental techniques. NaCl, and 0.1 mM EDTA. DNA were then fragmented by a Covaris This work was made possible by NIH Grant R01 NS094721 (to P.G.S.).

1. Lunn MR, Wang CH (2008) Spinal muscular atrophy. Lancet 371:2120–2133. 15. Hofmann Y, Lorson CL, Stamm S, Androphy EJ, Wirth B (2000) Htra2-beta 1 stimulates 2. D’Amico A, Mercuri E, Tiziano FD, Bertini E (2011) Spinal muscular atrophy. Orphanet an exonic splicing enhancer and can restore full-length SMN expression to survival J Rare Dis 6:71. motor neuron 2 (SMN2). Proc Natl Acad Sci USA 97:9618–9623. 3. Kolb SJ, Kissel JT (2011) Spinal muscular atrophy: A timely review. Arch Neurol 68: 16. Wee CD, Havens MA, Jodelka FM, Hastings ML (2014) Targeting SR proteins improves 979–984. SMN expression in spinal muscular atrophy cells. PLoS One 9:e115205. 4. Lorson CL, Hahnen E, Androphy EJ, Wirth B (1999) A single nucleotide in the SMN 17. Singh NN, Singh RN, Androphy EJ (2007) Modulating role of RNA structure in alter- gene regulates splicing and is responsible for spinal muscular atrophy. Proc Natl Acad native splicing of a critical exon in the spinal muscular atrophy genes. Nucleic Acids – Sci USA 96:6307 6311. Res 35:371–389. 5. Howe JA, et al. (2015) Selective small-molecule inhibition of an RNA structural ele- 18. Biogen, Inc (2014) A study to assess the efficacy and safety of IONIS-SMN Rx in patients – ment. Nature 526:672 677. with later-onset spinal muscular atrophy (CHERISH). NLM Identifier: NCT02292537. 6. Wemmer DE, Dervan PB (1997) Targeting the minor groove of DNA. Curr Opin Struct Available at https://clinicaltrials.gov/ct2/show/NCT02292537. Accessed November Biol 7:355–361. 15, 2017. 7. Yang F, et al. (2013) Antitumor activity of a pyrrole-imidazole polyamide. Proc Natl 19. Hua Y, et al. (2010) Antisense correction of SMN2 splicing in the CNS rescues necrosis Acad Sci USA 110:1863–1868. in a type III SMA mouse model. Genes Dev 24:1634–1644. 8. Kurmis AA, Yang F, Welch TR, Nickols NG, Dervan PB (2017) A pyrrole-imidazole 20. Makhortova NR, et al. (2011) A screen for regulators of survival of motor neuron polyamide is active against enzalutamide-resistant prostate cancer. Cancer Res 77: protein levels. Nat Chem Biol 7:544–552. 2207–2212. 21. Naryshkin NA, et al. (2014) Motor neuron disease. SMN2 splicing modifiers improve 9. Lintner NG, et al. (2017) Selective stalling of human translation through small- motor function and longevity in mice with spinal muscular atrophy. Science 345: molecule engagement of the ribosome nascent chain. PLoS Biol 15:e2001882. – 10. Velagapudi SP, et al. (2016) Design of a small molecule against an oncogenic non- 688 693. coding RNA. Proc Natl Acad Sci USA 113:5898–5903. 22. PTC Therapeutics, Inc. (2017) RG7916 increased SMN protein production in SUNFISH 11. Childs-Disney JL, Disney MD (2016) Approaches to validate and manipulate RNA clinical trial in patients with Type 2/3 spinal muscular atrophy (PTC Therapeutics, Inc., targets with small molecules in cells. Annu Rev Pharmacol Toxicol 56:123–140. South Plainfield, NJ). Available at ir.ptcbio.com/news-releases/news-release-details/ 12. Palacino J, et al. (2015) SMN2 splice modulators enhance U1-pre-mRNA association rg7916-increased-smn-protein-production-sunfish-clinical-trial?releaseid=1042582. and rescue SMA mice. Nat Chem Biol 11:511–517. Accessed October 29, 2017. 13. Cartegni L, Krainer AR (2002) Disruption of an SF2/ASF-dependent exonic splicing 23. Sivaramakrishnan M, et al. (2017) Binding to SMN2 pre-mRNA-protein complex elicits enhancer in SMN2 causes spinal muscular atrophy in the absence of SMN1. Nat Genet specificity for small molecule splicing modifiers. Nat Commun 8:1476. 30:377–384. 24. Bailey TL, et al. (2009) MEME SUITE: Tools for motif discovery and searching. Nucleic 14. Cartegni L, Hastings ML, Calarco JA, de Stanchina E, Krainer AR (2006) Determinants Acids Res 37:W202–W208. of exon 7 splicing in the spinal muscular atrophy genes, SMN1 and SMN2. Am J Hum 25. Pope LE, Sigman DS (1984) Secondary structure specificity of the nuclease activity of Genet 78:63–77. the 1,10-phenanthroline-copper complex. Proc Natl Acad Sci USA 81:3–7.

8of9 | www.pnas.org/cgi/doi/10.1073/pnas.1800260115 Wang et al. Downloaded by guest on September 30, 2021 26. Schultz PG, Dervan PB (1983) Sequence-specific double-strand cleavage of DNA by 38. Jacob AG, Singh RK, Mohammad F, Bebee TW, Chandler DS (2014) The splicing factor penta-N-methylpyrrolecarboxamide-EDTA X Fe(II). Proc Natl Acad Sci USA 80:6834–6837. FUBP1 is required for the efficient splicing of oncogene MDM2 pre-mRNA. J Biol 27. Putnam WC, Bashkin JK (2000) De novo synthesis of artificial ribonucleases with be- Chem 289:17350–17364. nign metal catalysts. Chem Comm 2000:767–768. 39. Miro J, et al. (2015) FUBP1: A new protagonist in splicing regulation of the DMD gene. – 28. Jin Y, Cowan JA (2005) DNA cleavage by copper-ATCUN complexes. Factors influ- Nucleic Acids Res 43:2378 2389. 40. Li H, et al. (2013) Far upstream element-binding protein 1 and RNA secondary encing cleavage mechanism and linearization of dsDNA. J Am Chem Soc 127: structure both mediate second-step splicing repression. Proc Natl Acad Sci USA 110: 8408–8415. E2687–E2695. 29. Murakawa GJ, Chen CHB, Kuwabara MD, Nierlich DP, Sigman DS (1989) Scission of 41. Markovtsov V, et al. (2000) Cooperative assembly of an hnRNP complex induced by a RNA by the chemical nuclease of 1,10-phenanthroline-copper ion: Preference for tissue-specific homolog of polypyrimidine tract binding protein. Mol Cell Biol 20: – single-stranded loops. Nucleic Acids Res 17:5361 5375. 7463–7479. ′ 30. Wilkinson KA, Merino EJ, Weeks KM (2006) Selective 2 -hydroxyl acylation analyzed 42. Min H, Turck CW, Nikolic JM, Black DL (1997) A new regulatory protein, KSRP, me- by primer extension (SHAPE): Quantitative RNA structure analysis at single nucleotide diates exon inclusion through an intronic splicing enhancer. Genes Dev 11:1023–1036. resolution. Nat Protoc 1:1610–1616. 43. García-Mayoral MF, et al. (2007) The structure of the C-terminal KH domains of KSRP 31. Spitale RC, et al. (2015) Structural imprints in vivo decode RNA regulatory mecha- reveals a noncanonical motif important for mRNA degradation. Structure 15:485–498. nisms. Nature 519:486–490. 44. Gherzi R, et al. (2004) A KH domain RNA binding protein, KSRP, promotes ARE- 32. Smola MJ, Rice GM, Busan S, Siegfried NA, Weeks KM (2015) Selective 2′-hydroxyl directed mRNA turnover by recruiting the degradation machinery. Mol Cell 14: – acylation analyzed by primer extension and mutational profiling (SHAPE-MaP) for 571 583. direct, versatile and accurate RNA structure analysis. Nat Protoc 10:1643–1669. 45. Trabucchi M, et al. (2009) The RNA-binding protein KSRP promotes the biogenesis of a subset of microRNAs. Nature 459:1010–1014. 33. Bollong MJ, et al. (2017) Small molecule-mediated inhibition of myofibroblast trans- 46. Morgan BS, Forte JE, Culver RN, Zhang Y, Hargrove AE (2017) Discovery of key differentiation for the treatment of fibrosis. Proc Natl Acad Sci USA 114:4679–4684. physicochemical, structural, and spatial properties of RNA-targeted bioactive ligands. 34. Johnson K, et al. (2012) A stem cell-based approach to cartilage repair. Science 336: Angew Chem Int Ed Engl 56:13498–13502. 717–721. 47. Disney MD, et al. (2016) Inforna 2.0: A platform for the sequence-based design of 35. Jafari R, et al. (2014) The cellular thermal shift assay for evaluating drug target in- small molecules targeting structured RNAs. ACS Chem Biol 11:1720–1728. – teractions in cells. Nat Protoc 9:2100 2122. 48. Mullard A (2017) Small molecules against RNA targets attract big backers. Nat Rev 36. Martinez Molina D, et al. (2013) Monitoring drug target engagement in cells and Drug Discov 16:813–815. tissues using the cellular thermal shift assay. Science 341:84–87. 49. Yang W-Y, Wilson HD, Velagapudi SP, Disney MD (2015) Inhibition of non-ATG 37. Kanamori H, Dodson RE, Shapiro DJ (1998) In vitro genetic analysis of the RNA translational events in cells via covalent small molecules targeting RNA. JAm binding site of vigilin, a multi-KH-domain protein. Mol Cell Biol 18:3991–4003. Chem Soc 137:5336–5345. BIOCHEMISTRY

Wang et al. PNAS Latest Articles | 9of9 Downloaded by guest on September 30, 2021