LY3023414 is a small molecule capable of increasing transcription of NMD targets.

A) Schematic of HTS used to identify inhibitors of NMD. Mutant transcripts are represented by a smaller length in the cartoon for illustrative purposes only. All small molecules were tested at 10μM. B) Mutant RNA reads relative to wild-type reads for the top 8 hits from the HTS. The dotted line represents the minimum fraction required to be considered a hit (> 5 standard deviations above DMSO control). Full screen results are presented in Figure S4. C) Targeted RNA sequencing results of isogenic RPTec knockout clones treated with the 8 best hits from the HTS at 10µM. The dotted line represents a relative RNA expression level of 1, equal to that of DMSO treated wells. Data for ceritinib, which did not validate on any line, is presented only in Figure S8. D) TP53 western blot on RPE TP53 224, containing a homozygous TP53 mutation, using the four hit compounds that validated in RPTec isogenic lines at 10μM. E) Western blot analysis of full-length TP53α and isoform TP53β after treatment with two NMD inhibitor lead candidates at 10μM. TP53β (expression known to be controlled by NMD) as well as mutant TP53 are prominently induced by LY3023414 whereas full-length is not. Note that RPE TP53 223 is a heterozygous knockout clone with one near wild-type allele whereas RPTec TP53 588 contains a homozygous TP53 indel mutation. F) qPCR showing 10μM LY3023414 treatment causes increased expression of the NMD controlled alternative transcript for TP53, TP53β, in parent cell lines for RPE1 and RPTec. Significance determined by Student’s t-test. Unless indicated otherwise cells were exposed to test compound for 16 hours.

Inhibiting NMD in cancer cells increases broad expression of truncated gene mRNA and protein.

A) Mutant transcript recovery rates for genes containing heterozygous indel mutations based on RNA-sequencing results in cell lines treated with 5µM LY3023414 for 16 hours. Strict inclusion criteria were used, such that only mutations with sufficient sequencing coverage are shown (see methods). Recovery is defined as at least 2-fold increase over DMSO treatment. B) Targeted high coverage RNA-sequencing confirms recovery of mutant transcript levels in NCI-H358 and LS180 cancer cell lines treated with 5µM LY3023414. RNF43 and DROSHA contain common heterozygous SNPs and the mutant allele refers to the non-reference genome allele. Error bars indicate 95% confidence limits. C) Western blot analyses of NCI-H358 cells showing mutant and wild-type protein levels in EXOC1 and D) SPTAN1 with and without 5µM LY3023414 treatment. The black arrow indicates the expected size of the mutant protein. The C-terminal SPTAN1 antibody is downstream of the out-of-frame indel mutation and is not expected to identify the mutant allele. E) Fold change in the number of mutant RNA transcripts from deep-targeted RNA sequencing of heterozygous mutated genes in NCI-H358 and LS180 xenografts treated by oral gavage with 60mg/kg LY3023414 assayed 16 hours post-treatment. Student’s t-test for target genes are all p < 0.05, while the null hypothesis holds for RNF43 (common SNP).

Novel NMD inhibitor KVS0001 is SMG1 specific and induces expression of NMD targeted genes in vitro and in vivo.

A) Fraction of mutant allele transcripts in genes with heterozygous indels previously established in this study as sensitive to NMD inhibition. Results show mutant levels after siRNA treatment targeting kinases inhibited by LY3023414. RNF43 and DROSHA are common heterozygous SNPs (shaded gray) and serve as negative controls. B) Fraction of mutant allele transcripts in genes with truncating mutations known to be sensitive to NMD inhibition after siRNA treatment with siUPF1 or non-targeting siRNA. Data from deep-targeted RNA-sequencing. C) Structure of novel NMD inhibitor KVS0001. D) Targeted RNA-sequencing on three genes with heterozygous, out-of-frame, indel mutations in LS180 cancer cells treated in a dose-response with KVS0001 or SMG1i-11. RNF43 serves as a control (common heterozygous SNP) and the mutant allele refers to the non-reference genome allele. E) Western blot of EXOC1 protein in NCI-H358 cells treated with 5µM novel inhibitor KVS0001, LY3023414 or SMG1i-11 for 24 hours. F) Western blot of phosphorylated UPF1 on three cell lines treated with 5µM KVS0001, SMG1i-11, or DMSO. Note that total UPF1 and p-UPF1 were run on different gels, loading controls correspond to indicated gel. G) Fold change in the number of mutant allele transcripts measured by targeted RNA-Seq in genes containing heterozygous out-of-frame indel mutations in NCI-H358 or H) LS180 subcutaneous xenografts in bilateral flanks of nude mice. Mice were treated once with IP injection of vehicle or 30mg/kg KVS0001 and tumors harvested 16hrs post IP treatment. All genes shown contain heterozygous out-of-frame truncating mutations except RNF43 and DROSHA which serve as controls (contain heterozygous SNP’s).

KVS0001 treatment induces targetable cell surface presentation of peptides known to be downregulated by NMD.

A) MHC class I HLA presentation of mutant specific peptide sequences from NCI-H358 and B) LS180 cells by quantitative HPLC-Mass Spectrometry treated with DMSO or 5µM KVS0001. The gene name, type of mutation (in parenthesis) and presented peptide are shown on the y-axis for each gene. Colors indicate different ions. C) TP53 gene structure and mutant DNA sequence for NCI-H716 and NCI-H2228 cancer cell lines, both contain a homozygous splice site mutation in TP53. Capital letters represent exonic sequence; lowercase letters represent intronic sequence. DNA mutation reflected by gold bases. D) Western blot against TP53 in the presence or absence of 5µM NMD inhibitor in NCI-H716_A24 and NCI-H2228 cell lines. NCI-H2228 has an expected size of 46.6 kDa and NCI-H716 of 34.7 kDa. E) IFN-γ levels over baseline based on ELISA in a co-culture assay with NCI-H716_A24 and NCI-H2228 cells, 1.25µM NMD inhibitor, human CD3+ T-cells, and bispecific antibody for TP53 and CD3. Chemotherapy (5-Fluorouracil) is shown as a control. F) Cell killing based on luciferase levels in a co-culture assay in NCI-H716 cells with and without A24 expression, treated with TP53-CD3 bispecific antibody, 1.25µM NMD inhibitor and human CD3+ T-cells.

In vivo treatment of murine tumors with KVS0001 yield differential tumor growth compared with vehicle treatment.

A) Fold change in RNA transcript levels in LLC or B) RENCA cells treated in-vitro with 5µM of NMD inhibitor KVS0001 or DMSO. Orange bars indicate genes with homozygous indel mutations potentially targeted by NMD. Purple bars show genes with no mutations but that are known to have their normal transcription levels controlled by NMD. Green bar is a control gene that should not change with treatment. The dotted line shows relative expression of DMSO treatment (equal to 1). C) Treatment schedule for syngeneic tumor mouse experiments. D) Average tumor size of LLC (left) and RENCA (right) syngeneic tumors in immune-competent mice (n=8) treated with 30 mg/kg KVS0001 or vehicle control. Difference is statistically significant after day 10 based on one-way ANOVA with Dunnett’s test p<0.001 (p<0.05 for day 23 RENCA data point) for both tumors tested. E) Average tumor size of LLC (left) and RENCA (right) in immunodeficient mice (n=8) treated with 30 mg/kg KVS0001 or vehicle control. Error bars show 95% confidence intervals in all plots.

Next generation sequencing results depicting genomic mutations at the CRISPR target area in STAG2 (top left) and TP53 (top right) isogenic cell line clones in the RPE1 cell line which were used in subsequent experiments. Western blot showing STAG2 protein loss in four independent clones of which only clones 2 and 8 were used in subsequent experiments (bottom left). Note that STAG2 clone 8 has two independent biallelic indel events. The mutation in exon 15 was used for scoring relative transcript levels in the screen and for subsequent work. IHC results confirming p53 protein loss in RPE TP53 221 (bottom right).

Fold change in RNA expression levels from whole transcriptome RNA-sequencing data for STAG2 and TP53 knockout clones in the RPE1 cell line background. RPE1 TP53 clone 221, RPE1 STAG2 clone 2 and RPE1 STAG2 clone 8 were used in the HTS screen. RPE1 TP53 clone 223 and RPE1 TP53 clone 224 were used in Figure S6. Note clone 223 has a 9bp in-frame deletion in one allele and an out of frame deletion on the other allele, presumably accounting for the higher level of expression.

RNA transcript level changes based on qPCR in STAG2 and TP53 knockout clones treated with the known NMD inhibitor emetine at 12 mg/mL. Error bars show standard deviation of three biological replicates. All changes are statistically significant by student t-test (p<0.05).

Primary screen results from high throughput assay. The x-axis represents all 2,658 compounds screened at 10µM, y-axis shows ratio of mutant to wild type reads for each of the three isogenic cell lines. Higher values indicate more mutant RNA reads, representing inhibition of NMD. The dotted line at 0.46 is the cutoff for a hit to be called (5 standard deviations above DMSO treated samples). Colored data points demarcate the eight hit compounds in the three screened cell lines.

Emetine and DMSO control sample data from the HTS for each of the three clones used as measured by deep-targeted RNA sequencing. 6 DMSO and 2 emetine (12mg/mL) samples were included in each dosing plate for a total of 198 DMSO and 66 emetine measurements in each boxplot.

A) IHC staining of TP53 in RPTec TP53 knockout clones. B) Western blot against STAG2 on RPTec STAG2 knockout clones demonstrating successful knockout at the protein level of all ten clones. Out of frame indels were confirmed in all clones by NGS (data not shown). The arrow indicates the expected size for full-length STAG2 protein. C) IHC staining of TP53 protein on RPE1 TP53 knockout clones. Clone 223 contains the same truncating mutation found in clone 224 on allele 1 and has a 9 bp in-frame deletion that preserves some full length TP53α and TP53β isoform expression on allele 2 (See Figure 1E).

Protein schematic cartoons showing indel mutation site and expected size of various TP53 knockout clones used in this study. Note, RPE TP53 223 has an in-frame deletion in the DNA binding domain.

Fold change in mutant RNA transcription levels for STAG2 and TP53 in three knockout cell lines from the RPtec background containing out of frame indels targeted by NMD. Each line was treated with each of the eight hit compounds from the screen. The 10µM dose is also shown in figure 1C.

RNA expression levels of kinases post siRNA targeting in NCI-H358 and LS180 cells by qPCR. Cells are treated with siRNAs targeting genes known to be inhibited by LY3023414. Error bars represent 95% confidence intervals.

Slider plots showing mutant allele fraction relative to total reads measured by deep-targeted RNA-sequencing for in vitro treated LS180 (top in red) or NCI-H358 (bottom in blue) cells with 5µM LY3023414 (labeled LYO) or a previously described SMG1 inhibitor SMG1i-11 at 1µM. Gene names are shown in the boxes above each slider plot, genes highlighted in gray are common SNPs and serve as a negative control (not expected to change). In the case of the control SNPs, the mutant allele refers to the non-reference genome allele. TRIM21 did not show a large change in expression with either small molecule, while ANLN did not show a difference with LY3023414 but did respond to SMG1i-11. The remaining genes responded to NMD inhibition by both LY3023414 and SMG1i-11.

A) Western blot showing SPTAN1 expression after treatment with DMSO, 5µM LY3023414, or 1µM SMG1 inhibitor SMG1i-11 (lanes 1,2,3 respectively for each antibody). The arrow indicates the expected size of the mutant (NMD targeted) protein. Antibody ab75755 (Abcam) binds C-terminal to the out of frame indel and does not show mutant protein as expected. Antibody A301-249 (Bethyl) is polyclonal and also did not bind mutant protein. Antibody ab11755 (Abcam) is located N-terminal to the indel and does display mutant protein expression. B) Western blot showing EXOC1 expression after treatment with DMSO, 5µM LY3023414, or 1µM SMG1i-11 (lanes 1,2,3 respectively). The arrow indicates the expected size of the mutant (NMD targeted) protein.

Biophysical properties for novel SMG1 inhibitor KVS0001.

A) Kinativ™ Assay results for KVS0001 at 100 nM and B) 1µM run with biological replicates showing KVS0001 specificity against the known kinome. Results are based on the average between two unique peptides for each kinase.

A) Slider plots showing mutant allele fraction measured by deep-targeted RNA-sequencing for genes from NCI-H358 (top in red) and B) LS180 (bottom in blue) treated in a dose response in vitro with novel NMD inhibitor KVS0001. DROSHA and RNF43 are common heterozygous SNPs and serve as a negative control. In the case of the control SNPs, the mutant allele refers to the non-reference genome allele. Only the highest two concentrations were tested on LY3023414 which served as a positive control in this experiment. The dotted line indicates the mutant expression with DMSO treatment and the solid line is a reference for equal expression of both the wild type and mutant alleles.

A) Western blot showing expression of LMAN1 in LS180 cells treated with DMSO (lane 1), KVS0001 at 5µM (lane 2) or SMG1i-11 at 1µM (lane 3). Arrows indicate expected size of wild type and mutant LMAN1 protein. B) Stain free loading control image for gel.

DNA and protein sequences for the wild type and mutant alleles of A) EXOC1 B) RAB14 and C) ZDHHC16 genes. The mutant protein sequences caused by the out of frame indel are highlighted in red and the boxes indicate the peptides presented on the cell surface and identified by mass spectrometry in cells treated with 5µM of KVS0001 (see Figures 4A and 4B).

Heavy peptide loading controls and endogenous (light) peptide presentation of genes in LS180 and NCI-H358 treated with DMSO or 5µM KVS0001. Data is from quantitative HPLC-Mass Spectrometry. Note y-axis scale changes between samples. Tables on the right show relative increase in peptide presentation with KVS0001.

Waterfall plot of publicly available TP53 RNA expression (as shown by FPKM) for 675 cancer cell lines. The two cell lines used in this study are highlighted in red and were in the bottom quartile of TP53 expression.

Western blot of TP53 on NCI-H716 and NCI-H2228 cells treated with 5µM or 7.5µM of NMD inhibitor, 1µM SMG1i-11, or 200mg/mL chemotherapy, showing controls related to Figure 4D. LYO is LY3023414 and 5-FU is 5-fluorouracil. HEK293 parent cells are shown as a control (wild type TP53 protein).

Mouse weights for A) C57BL/6N and B) BALB/c tumor bearing mice treated with 30mg/kg KVS0001 or vehicle control IP daily. N=8 for all arms.

Mouse tumor size as measured by calipers for mammary fat pad placed syngeneic tumor models treated daily with 30mg/kg KVS0001 or vehicle. A) Tumors with low/moderate and B) high indel mutational loads are shown. No results presented in this supplemental figure are statistically significant. Error bars show 95% confidence intervals.