SMG5 and SMG7 help SMG6 cut faulty mRNA to trigger its breakdown

To gain insights into the transcriptome-wide effects of the SMG7 depletion, we sequenced poly(A)+ enriched mRNA from both SMG7 KO clones and identified differentially expressed genes (Extended Data Fig. 2a and Supplementary Data 1). Consistent with the mRNA-degrading function of SMG7 in NMD, more than twice as many genes were upregulated than downregulated in the SMG7 KO cells (Extended Data Fig. 2b). Compared to a recently published study using SMG7 KD in HeLa cells¹⁴, this ratio of upregulated vs. downregulated genes was higher (Extended Data Fig. 2b–e). We observed a substantial overlap between upregulated genes in both SMG7 KO cell lines, indicating that these genes are high-confidence SMG7 targets. In contrast, only a limited overlap between downregulated genes could be detected, suggesting these are rather clone-specific effects or off-targets (Extended Data Fig. 2b). From these analyses, we conclude that the KO of SMG7 leads to stronger NMD inhibition than the KD.

Next, we quantified alternative splicing events (Supplementary Data 2), as well as differential transcript usage, and identified significant isoform switches (Extended Data Fig. 2a and Supplementary Data 3). Isoform switches are characterized by significant changes in the relative contribution of isoforms to the overall gene expression when comparing two conditions⁷⁵. Because of the identification at the isoform level, this approach allows the identification of PTC-containing transcripts that are upregulated upon NMD inhibition. As a specific example of such an isoform switch, we visualized the read coverage for the previously used bona fide NMD target SRSF2. While the overall SRSF2 expression remained nearly unchanged, we detected prominent NMD-inducing exon inclusion and 3′ UTR splicing events in the SMG7 KO but not in the SMG7 KD conditions (Fig. 1e). We verified the SMG7-dependent upregulation of additional examples by end-point PCR (Extended Data Fig. 2f–h). On a transcriptome-wide scale, NMD-sensitive isoforms with annotated PTC were almost exclusively detected to be upregulated in the SMG7 KO cells, which was not the case in the SMG7 KD in HeLa cells¹⁴ (Fig. 1f–i). Collectively, the RNA-Seq data analysis supported the initial observation that NMD is markedly impaired when SMG7 is knocked out. Due to the clear effect on NMD upon complete loss of SMG7, the KO cells provide an ideal background to examine further mechanistic details of NMD, which could not be studied before.

To gain further mechanistic insight into the cooperation between SMG5, SMG6, and SMG7, we performed complementation experiments in SMG7 KO cells with additional SMG7 rescue constructs and in combination with SMG5 or SMG6 KDs (Extended Data Fig. 3c). Three main clusters of rescue conditions were identified according to their NMD activity (Fig. 2f and Extended Data Fig. 3d). Consistent with our earlier observations, SMG7 WT rescued the SMG7 defect in all conditions and restored high NMD activity. The SMG7 1-633 deletion mutant (reported to be unable to recruit the CCR4-NOT complex; Fig. 2a)⁶⁴ also conferred high NMD activity in SMG7 KO cells, even when SMG5 or SMG6 were depleted in addition (Fig. 2f and Extended Data Fig. 3c, d). This result indicates that the accelerated deadenylation of NMD substrates by SMG7-recruited CCR4-NOT is not required for their efficient degradation. In contrast to a previous report⁶⁴, the C-terminus of SMG7 is not required for NMD even when SMG6 is downregulated. However, by combining the C-terminal truncation with either 14-3-3^mut or G100E mutations, these SMG7 variants became less NMD-competent, especially in the SMG5 KD condition (Fig. 2f and Extended Data Fig. 3c, d). This observation suggests that the deadenylation-inducing activity might serve as an additive but dispensable feature that helps to clear NMD targets, especially when other features of SMG7 are inactivated. In the full-length SMG7 context, the 14-3-3^mut or G100E mutations displayed intermediate NMD rescue activity in the SMG5 and SMG6 KD conditions, whereas a 14-3-3^mut/G100E double-mutation was completely inactive in all conditions (Fig. 2f and Extended Data Fig. 3d). All KD/rescue experiments were also performed in the SMG7 KO clone 34 and showed similar functional outcomes (Extended Data Fig. 3e). In summary, these results reveal the synergistic effect of simultaneous binding of SMG7 to p-UPF1 and SMG5, whereas the lack of both interactions incapacitates the function of SMG7 in NMD.

Next, we performed co-immunoprecipitation (co-IP) experiments with the functionally tested set of FLAG-tagged SMG7 constructs to analyze their steady-state interaction with UPF1 (Extended Data Fig. 3f). The full-length or truncated 14-3-3^mut SMG7 protein showed strongly impaired ability to co-immunoprecipitate UPF1, consistent with the role of the 14-3-3-like domain to mediate the stable interaction with phosphorylated UPF1 (Extended Data Fig. 3f; lane 10 vs. lanes 11 and 15)^63,77. In turn, this result also indicates that SMG5 does not bridge the 14-3-3^mut SMG7 protein to UPF1. The G100E mutants co-immunoprecipitated similar UPF1 levels than the respective WT construct (Extended Data Fig. 3f; lanes 10,14 vs. lanes 12,16), confirming that SMG7 does not require the heterodimerization with SMG5 to interact with UPF1⁶². In conjunction with the impaired functional rescue of the G100E mutants (Fig. 2f and Extended Data Fig. 3c–e), this result implies that disrupting the interaction of SMG7 with SMG5 impairs NMD even when the interaction between SMG7 and UPF1 is maintained. In conclusion, our functional studies reveal the previously underestimated requirement for SMG7 to interact with both p-UPF1 and SMG5 in order to support full NMD activity.

The unexpected relevance of SMG5 prompted us to investigate the molecular properties of SMG5 in more detail. Similar to our analysis of SMG7, we aimed to understand which functions of SMG5 are required to maintain NMD competence in the presence or absence of SMG7. The first striking result was the almost complete rescue of the SMG7 depletion phenotype by the overexpression (~50-fold) of SMG5 WT or G120E mutant (reduced interaction with SMG7)⁶² in control or SMG5 KD conditions (Fig. 3a, b and Extended Data Fig. 4a–c). This finding was very surprising since SMG5 is not expected to be able to directly carry out RNA degradation. Although SMG5 harbors a PilT N-terminus (PIN) domain that structurally resembles the functional PIN domain of SMG6, two of the three required catalytic residues are missing in the inactive C-terminal SMG5 PIN domain (Fig. 3a)⁷⁸. SMG5 was found in one study to promote degradation by interacting with the decapping factor PNRC2 (Fig. 3a), which could explain the NMD activity we observe in the rescue assays⁷⁹. However, we did not observe any effects of PNRC2 KDs on the ability of SMG5 to rescue the NMD defects (Fig. 3c), consistent with other studies which failed to confirm this PNRC2-dependent path of SMG5-dependent degradation^64,65.

Collectively, the observation of SMG5 expression rescuing SMG7 loss once again calls into question the relevance of the SMG7 deadenylation-promoting function for NMD. We hypothesized that in the absence of SMG7, SMG5 interacts with p-UPF1 via its N-terminal 14-3-3-like domain to activate NMD. In line with this hypothesis, mutating three residues in the potential phosphopeptide binding pocket of SMG5 severely affected the ability of SMG5 to rescue the SMG7 KO phenotype in control or SMG5 KD conditions (Fig. 3b and Extended Data Fig. 4a, b). Of note, we could not detect any stable binding of SMG5 to UPF1 above background levels in SMG7 KO cells (Extended Data Fig. 4d), confirming that strong SMG5-UPF1 interactions require SMG7. However, in WT cells the 14-3-3^mut SMG5 protein showed reduced binding to p-UPF1 (Fig. 3d).

We hypothesized that besides the interaction with UPF1 another function of SMG5 may be required to maintain NMD activity. To test this idea, we generated an SMG5 deletion mutant (1-853) lacking the C-terminal PIN domain (Fig. 3a). This SMG5 mutant, albeit being able to interact with UPF1 (Fig. 3d), was unable to restore NMD activity (Fig. 3b and Extended Data Fig. 4a, b), suggesting that the catalytically inactive PIN domain of SMG5 carries out an essential function during NMD. In search of an explanation, we considered earlier reports that the C-terminus of SMG5 is involved in the dephosphorylation of UPF1, likely by direct recruitment of the protein phosphatase 2 (PP2A) complex (Fig. 3a)^53,60. Appropriately, we observed that the expression of the SMG5 1-853 construct leads to overall increased UPF1 phosphorylation, and consequently the SMG5 1-853 protein co-immunoprecipitated more p-UPF1 (Fig. 3d), indicating that the C-terminal PIN domain could play a role in the dephosphorylation of UPF1. To address whether the SMG5 PIN domain indeed enables the dephosphorylation of UPF1 via the direct recruitment of PP2A, we performed co-immunoprecipitation of FLAG-tagged SMG5 WT or 1-853 constructs in WT 293 cells, followed by label-free mass spectrometry (Supplementary Data 4). Although both SMG5 constructs co-immunoprecipitated the direct binding partner SMG7 efficiently, we did not detect a significant enrichment of neither PP2A nor PP1 protein phosphatase complex components (Fig. 3e–g). We detected increased levels of UPF1 and UPF2 in the SMG5 1-853 co-immunoprecipitation compared to the SMG5 WT, whereas PNRC2 or other NMD components such as SMG6 were not enriched in any condition (Fig. 3e–g and Supplementary Data 4). Collectively, these findings do not provide evidence for but rather against the direct recruitment of PP2A via the SMG5 PIN domain, which questions the previously assumed role of SMG5 in UPF1 dephosphorylation. An alternative explanation could be that the SMG5 PIN domain is critical to promote NMD execution once SMG5 is bound to p-UPF1. Accordingly, the absence of the PIN domain would lead to the accumulation of NMD-incompetent UPF1 complexes. Supporting evidence for this hypothesis is the failed functional rescue of SMG5 1-853 (Fig. 3b and Extended Data Fig. 4a, b), the increased interaction with UPF1 and UPF2 (Fig. 3e–g), and the apparent hyper-phosphorylation of UPF1 (Fig. 3d).

The finding of SMG5-dependent alterations of UPF1 phosphorylation levels prompted us to test whether, in cells lacking either SMG5, SMG7, or both, the general UPF1 phosphorylation status is changed. These alterations could reflect stalled UPF1-containing complexes in an arrested NMD processes. To this end, we expressed FLAG-tagged UPF1 in WT or SMG7 KO cells and detected its overall S/T phosphorylation level after immunoprecipitation (Extended Data Fig. 4e). The overexpressed UPF1 showed elevated phosphorylation in the SMG7 KO cells, which were not further increased when SMG5 was depleted (Extended Data Fig. 4e). We also assessed the SMG5-SMG7 dependent phosphorylation status of endogenous UPF1 using the phospho-S1127 specific antibody. Whereas the isolated KD of SMG5 had no effect on the phosphorylation status, similarly increased p-UPF1/UPF1 ratios were detected in SMG7 KO cells with or without SMG5 KD (Fig. 3h). Together, this indicates that the severe NMD defects observed in SMG5-SMG7 depleted cells compared to SMG7 KO cells do not coincide with further increased accumulation of hyper-phosphorylated UPF1.

Searching for other possible molecular explanations, we considered previous reports that SMG5 and SMG7 stabilize p-UPF1 binding to target mRNAs⁶¹ and ATPase-deficient UPF1 mutants accumulate in a hyper-phosphorylated, SMG5-SMG7 bound state⁴⁸. Inspired by these results, we investigated whether the ability of UPF1 to bind to and/or to recognize NMD targets is impaired in SMG5-SMG7 depleted cells. To this end, we employed UPF1 RNA immunoprecipitation (RIP) assays to study NMD target discrimination by UPF1⁴⁸. Binding of UPF1 to two NMD targets, which displayed increased mRNA levels upon NMD inhibition (Supplementary Data 1), as well as to two non-NMD targets remained unchanged in control, SMG5 KD, SMG7 KO, or SMG7 KO plus SMG5 KD conditions (Extended Data Fig. 4f–i). These results suggest that UPF1 can identify and bind to NMD-targeted transcripts, although their degradation cannot be executed anymore.

Intrigued by the remarkable changes in gene expression and isoform usage when NMD was effectively inhibited, we wanted to re-examine the statement that up to 10% of genes are affected by NMD. To this end, we calculated how many genes showed single or combined differential gene expression (DGE), differential transcript usage (DTU), or alternative splicing (AS) events when the SMG7 KO was combined with the KD of SMG5 or SMG6 (Fig. 4h). With this approach, we find that about 40% of the expressed genome in the Flp-In-T-REx-293 cells is directly or indirectly altered by strong NMD inhibition. With more stringent cutoffs for the analyses, still, around 20% of all expressed genes were affected (Extended Data Fig. 5i), mostly those with medium to high expression levels (Extended Data Fig. 5j). Collectively, the RNA-Seq analysis confirmed that SMG5 KD, as well as SMG6 KD, have similar effects on NMD in SMG7 KO cells. It also provided global evidence that the loss of the SMG5-SMG7 heterodimer effectively inactivates NMD and leads to massive changes in the expressed transcriptome. In combination with the functional studies (Figs. 2–3), these transcriptome-wide observations profoundly question the independence of the SMG5-SMG7 and SMG6 decay pathways and suggest a functional connection during NMD execution.

Finally, we were intrigued by the fact that both SMG5 and SMG7 wild-type proteins can individually rescue the SMG7 KO phenotype. Specifically, we wondered if the main NMD-supporting function of both factors could be to enable and/or sustain SMG6-mediated endonucleolytic cleavage of the target mRNA. All SMG5 and SMG7 rescue proteins that restored full NMD activity also resulted in a normal NOP56 cleavage pattern, indicating that SMG6 was reactivated in these cells (Fig. 5d). Based on these results, we postulate that SMG5 and SMG7 maintain NMD activity by permitting the activation of SMG6.

Flp-In-T-REx-293 (human, female, embryonic kidney, epithelial; Thermo Fisher Scientific, RRID:CVCL_U427) cells were cultured in high glucose, GlutaMAX DMEM (Gibco) supplemented with 9% fetal bovine serum (Gibco) and 1 × Penicillin Streptomycin (Gibco). The cells were cultivated at 37 °C and 5% CO₂ in a humidified incubator. The generation of knockout and stable cell lines is described below and all cell lines are summarized in Supplementary Data 6.

The knockouts were performed using the Alt-R CRISPR-Cas9 system (Integrated DNA Technologies) and reverse transfection of a Cas9:guideRNA ribonucleoprotein complex using Lipofectamine RNAiMAX (Thermo Fisher Scientific) according to the manufacturer’s protocol. The crRNA sequence (Design ID: Hs.Cas9.SMG7.1.AD; Integrated DNA Technologies) to target SMG7 was /AlTR1/rGrArArArArUrGrCrUrArGrUrUrArCrCrGrArUrUrGrUrUrUrUrArGrArGrCrUrArUrGrCrU/AlTR2/. Reverse transfection was performed on 1.5 × 10⁵ cells per crRNA in 12-well plates. 48 h after transfection the cells were trypsinized, counted, and seeded at a mean density of a single cell per well in 96-well plates. Cell colonies originating from a single clone were then screened via western blot and genome editing of SMG7 was analyzed on the genomic level via DNA extraction and Sanger sequencing (Eurofins Genomics). Alterations on the transcript level were analyzed via RNA extraction (see below) followed by reverse transcription and Sanger sequencing.

One day prior to DNA extraction, 2.5 × 10⁵ cells were seeded in a 6-well plate. To extract DNA, QuickExtract DNA Extraction Solution (Lucigen) was used following the manufacturer’s instructions. RNA was isolated using peqGOLD TriFast (VWR Peqlab) or RNA-Solv (Omega Bio-Tek) following the manufacturer’s instructions. Following changes were made: Instead of 200 μl chloroform, 150 μl 1-Bromo-3-chloropropane (Molecular Research Center, Inc.) was used. RNA was resuspended in 20 μl RNase-free water.

SDS-polyacrylamide gel electrophoresis and immunoblot analysis were performed using protein samples harvested with RIPA buffer (50 mM Tris/HCl pH 8.0, 0.1% SDS, 150 mM NaCl, 1% IGEPAL CA 630, 0.5% deoxycholate) or samples eluted from Anti-FLAG M2 magnetic beads (Sigma-Aldrich). For protein quantification, the Pierce Detergent Compatible Bradford Assay Reagent (Thermo Fisher Scientific) was used. All antibodies were used at the indicated dilutions in 50 mM Tris [pH 7.2], 150 mM NaCl with 0.2% Tween-20, and 5% skim milk powder. Amersham ECL Prime or Select Western Blotting Detection Reagent (GE Healthcare) in combination with the Fusion FX-6 Edge system (Vilber Lourmat) was used for visualization. All antibodies used in this study are listed in Supplementary Data. Protein bands detected with the Fusion FX-6 Edge system (Vilber Lourmat) using the Evolution-Capt Edge software (version 18.05) were quantified in a semi-automated manner using the ImageQuant TL 1D software (version 8.1) with a rolling-ball background correction. The control condition was set to unity, quantification results are shown as data points and mean. 6

To measure the growth and mortality rate of cells, CytoTox-Glo Cytotoxicity Assay (Promega) was performed. 10,000 cells/well were seeded in a 96-well plate and the assay was performed after 0–4 days using luminometer Centro XS3 LB 960 (Berthold Technologies) and the MikroWin software (version 5.14) following the manufacturer’s instructions.

The point and deletion mutants of SMG7 were PCR amplified using Q5 polymerase (New England Biolabs) and inserted with an N-terminal FLAG-tag via NheI and NotI (both New England Biolabs) restriction sites into PB-CuO-MCS-IRES-GFP-EF1α-CymR-Puro (System Biosciences). Accordingly, N-terminally FLAG-tagged GST or UPF1, as well as FLAG-TurboID or FLAG-TurboID-UPF1 (generated with Integrated DNA Technologies gBlocks and/or PCR amplification) were cloned via NheI and NotI into PB-CuO-MCS-BGH-EF1-CymR-Puro, which was modified from the original vector by replacing the IRES-GFP cassette with a BGH polyA signal.

The point and deletion mutants of SMG5 were PCR amplified using Q5 polymerase and inserted with an N-terminal FLAG-tag via NheI and NotI restriction sites into the tetracycline-inducible pcDNA5/FRT/TO vector (Thermo Fisher Scientific). The mRNA reporter constructs TPI-WT and TPI-PTC160 in the pcDNA5/FRT/TO vector are available on Addgene (IDs 108377-108378).

The cells were stably transfected using the PiggyBac Transposon system (SMG7, UPF1, GST) or using the Flp-In T-REx system (SMG5, mRNA reporter). 2.5–3 × 10⁵ cells were seeded 24 h before transfection in 6-wells. For PiggyBac stable cells, 2 µg of PiggyBac construct was transfected together with 0.8 µg of the Super PiggyBac Transposase expressing vector and for Flp-In T-REx stable cells, 1-2 µg of pcDNA5 construct was transfected together with 1 µg of the Flp recombinase expressing plasmid pOG44, using the calcium phosphate method. Forty-eight hours after transfection, the cells were transferred into 10 cm dishes and selected with 2 µg ml⁻¹ puromycin (InvivoGen) for PiggyBac or 100 µg ml⁻¹ hygromycin (InvivoGen) for Flp-In T-REx. After 7–10 days, the colonies were pooled. Expression of the PiggyBac constructs was induced with 30 µg ml⁻¹ cumate, Flp-In T-REx constructs were induced with 1 µg ml⁻¹ doxycycline. All vectors used in this study are listed in Supplementary Data 6.

Mycoplasma contamination was tested by PCR amplification of mycoplasma-specific genomic DNA⁹⁴ or by using the Mycoplasmacheck service (Eurofins Genomics).

1–4 µg of total RNA was reverse-transcribed in a 20 µl reaction volume with 10 µM VNN-(dT)₂₀ primer using the GoScript Reverse Transcriptase (Promega). 2% of cDNA was used as template in end-point PCRs using the GoTaq Green Master Mix (Promega) or MyTaq Red Mix (Bioline) and 0.2–0.6 µM final concentration of sense and antisense primer (see Supplementary Data 6 for sequences). After 30 PCR cycles, the samples were resolved by electrophoresis on ethidium bromide-stained, 1–2% agarose TBE gels and visualized by trans-UV illumination using the Gel Doc XR+ (Bio-Rad) and Image Lab software (version 5.1).

Bands detected in agarose gels from the indicated biological replicates of end-point PCRs were quantified using the Image Lab software (version 6.0.1). Results of the indicated band % per lane are shown as data points and mean. Sanger sequencing of individual bands was performed using the service of Eurofins Genomics.

Quantitative RT-PCR was performed with the GoTaq qPCR Master Mix (Promega) using 2% of cDNA in 10 µl reactions, 0.2–0.6 µM final concentration of sense and antisense primer (see Supplementary Datafor sequences), and the CFX96 Touch Real-Time PCR Detection System (Bio-Rad) with Bio-Rad CFX Manager software (version 3.0). The reactions for each biological replicate were performed in duplicates or triplicates and the average Ct (threshold cycle) value was measured. For alternative splicing events, values for canonical isoforms were subtracted from values for NMD-sensitive isoforms to calculate the ΔCt. For differentially expressed targets, the values for the housekeeping genes C1orf43 were subtracted from values for the target to calculate the ΔCt. The mean log2 fold changes were calculated from three biologically independent experiments. Log2 fold change results are shown as data points and mean. 6

Cells were seeded in 6-well plates at a density of 2–3 × 10⁵ cells per well and reverse transfected using 2.5 µl Lipofectamine RNAiMAX and 60 pmol of the respective siRNA(s) according to the manufacturer’s instructions. In preparation for UPF1 phosphorylation and RNA immunoprecipitation (RIP) assays, 3 × 10⁶ cells were reverse transfected in 10 cm dishes using 6.25 µl Lipofectamine RNAiMAX and 150–200 pmol siRNA. All siRNAs used in this study are listed in Supplementary Data 6.

RNA-Seq experiments were carried out with Flp-In-T-REx-293 wild-type (WT) cells transfected with Luciferase or SMG5 siRNA and the SMG7 KO clones 2 and 34 transfected with either Luciferase, SMG5, or SMG6 siRNAs. Three biological replicates were analyzed for each sample. Total RNA was extracted using peqGOLD TriFast (VWR Peqlab) as described above.

The Lexogen SIRV Set1 Spike-In Control Mix (SKU: 025.03) that provides a set of external RNA controls was added to the total RNA to enable performance assessment. Mix E0 was added to samples with Luciferase siRNA, mix E1 was added to samples with SMG5 siRNA, and mix E2 samples with SMG6 siRNA. The Spike-Ins were used for quality control purposes, but not used for the final analysis of DGE, DTU, or AS.

The library preparation was performed with the TruSeq mRNA Stranded kit (Illumina). After poly-A selection (using poly-T oligo-attached magnetic beads), mRNA was purified and fragmented using divalent cations under elevated temperatures. The RNA fragments underwent reverse transcription using random primers. This is followed by second-strand cDNA synthesis with DNA Polymerase I and RNase H. After end repair and A-tailing, indexing adapters were ligated. The products were then purified and amplified to create the final cDNA libraries. After library validation and quantification (Agilent tape station), equimolar amounts of the library were pooled. The pool was quantified by using the Peqlab KAPA Library Quantification Kit and the Applied Biosystems 7900HT Sequence Detection System and sequenced on an Illumina NovaSeq6000 sequencing instrument and a PE100 protocol.

Reads were aligned against the human genome (version 38, GENCODE release 33 transcript annotations⁹⁵ supplemented with SIRVomeERCCome annotations from Lexogen; obtained from https://www.lexogen.com/sirvs/download/↗) using the STAR read aligner (version 2.7.3a)⁹⁶. Transcript abundance estimates were computed with Salmon (version 1.3.0)⁹⁷ with a decoy-aware transcriptome. After the import of transcript abundances, differential gene expression analysis was performed with the DESeq2⁹⁸ R package (version 1.28.1) with the significance thresholds |log2FoldChange|> 1 and adjusted p-value (padj) <0.05. Differential splicing was detected with LeafCutter (version 0.2.9)⁹⁹ with the significance thresholds |deltapsi| >0.1 and adjusted p-value (p.adjust) <0.05. Differential transcript usage was computed with IsoformSwitchAnalyzeR (version 1.10.0) and the DEXSeq method^75,100–104. Significance thresholds were |dIF| >0.1 and adjusted p-value (isoform_switch_q_value) <0.05.

PTC status of transcript isoforms with the annotated open reading frame was determined by IsoformSwitchAnalyzeR using the 50 nucleotides (nt) rule of NMD^75,105–107. Isoforms with no annotated open reading frame in GENCODE were designated “NA” in the PTC analysis.

The control, SMG7, and SMG6/7 knockdown datasets (Gene Expression Omnibus, GEO: GSE86148↗)¹⁴ were processed and analyzed with the same programs, program versions, and scripts as the SMG7 KO dataset, with minor changes due to the different sequencing method (paired-end vs. single-end). All scripts and parameters for the RNA-Seq analysis are available at GitHub [https://github.com/boehmv/SMG5-SMG7↗]. Overlaps of data sets were represented via nVenn¹⁰⁸ or the ComplexHeatmap package (version 2.6.2)¹⁰⁹. Integrative Genomics Viewer (IGV) (version 2.8.12)¹¹⁰ snapshots were generated from mapped reads (BAM files) converted to binary tiled data (tdf), using Alfred¹¹¹ with resolution set to 1 and IGVtools. Mean junction counts were obtained from sashimi plots generated using ggsashimi¹¹².

The structure of human SMG7 (PDB: 1YA0) was superimposed onto the C. elegans SMG7 (PDB: 3ZHE) using the MatchMaker command in Chimera version 1.13¹¹³, to generate a hsSMG7-ceSMG5 hybrid model. ChimeraX version 1.0¹¹⁴ was used to visualize the modeled structure.

The cells were harvested in peqGOLD TriFast reagent (VWR) and total RNA extraction was performed as described above. 2–4 μg of total RNA were resolved on a 1% agarose/0.4 M formaldehyde gel using the tricine/triethanolamine buffer system¹¹⁵ followed by a transfer on a nylon membrane (Roth) in 10× SSC. The blots were incubated overnight at 65 °C in Church buffer containing α-32P-GTP [800 Ci/mmol, 10 mCi/ml] body-labeled RNA probes for detection of the reporter mRNA⁸⁰.

Endogenous 7SL RNA was detected by a 5′-32P-labeled oligonucleotide (5′-TGCTCCGTTTCCGACCTGGGCCGGTTCACCCCTCCTT-3′) for which γ-32P-ATP [800 Ci/mmol, 10 mCi/ml] was used for labeling. For NOP56 northern blots, the ex8b riboprobe sequence (5′-GAAACUUGGUCCCUUUGCUGGGCCCUGGGAAUCACUCAGACACCAGGACUGGCCAUCACCCCCAUAGCAGAGGCCUGUAUAGGUCAGGGAGCCCUGGUCAGCCAUCACCGUGAUCCCCCAACAAGCAGUGGGCACCAGAAGUGGCACCUGAUU-3′)⁸¹ was cloned into the pSP73 vector, linearized and in vitro transcribed using α-32P-GTP [3000 Ci/mmol, 10 mCi/ml]. Ethidium bromide-stained 28S and 18S rRNA served as loading controls. RNA signal from at least three distinct samples was detected with the Typhoon FLA 7000 (GE Healthcare) and was quantified in a semi-automated manner using the ImageQuant TL 1D software (version 8.1) with a rolling-ball background correction. EtBr-stained rRNA bands were quantified with the Image Lab 6.0.1 software (Bio-Rad). Signal intensities were normalized to the internal control (7SL or rRNA) before the calculation of mean values. The control condition was set to unity (TPI WT for reporter assays), quantification results are shown as data points and mean.

The RIP protocol was adapted from Lee et al.⁴⁸. Two days after seeding and reverse transfecting cells with siRNA, the cells were washed with 2 ml PBS, harvested in 1 ml PBS, collected for 10 min at 100 × g, and resuspended in 300 µl RIP lysis buffer (10 mM Tris pH 7.5, 150 mM NaCl, 2 mM EDTA, 0.1% Triton X-100) supplemented with 1 tablet of PhosSTOP (Roche), 100 µl EDTA-free HALT Protease and Phosphatase Inhibitor (ThermoFisher) and 20 µl RNasin (Promega) per 10 ml buffer. Protein concentration was measured, adjusted to 1 mg total protein in 450 µl RIP buffer, 20 µl of the sample was added to 500 µl peqGOLD TriFast and saved as input sample. The remaining sample was divided into 2× 200 µl and combined with 30 µl pre-washed Dynabeads Protein G beads (ThermoFisher), which were pre-incubated with 5 µg of either purified goat IgG (control; Bethyl, P50-100) or UPF1 antibody (Bethyl, A300-036A). The samples were incubated in an overhead rotator for 1 h at 4 °C, washed 5× with 1 ml RIP Wash Buffer (5 mM Tris pH 7.5, 150 mM NaCl, 0.1% Triton X-100), and co-immunoprecipitated RNA was recovered by incubating the beads with 500 µl RNA-solv. for 10 min at room temperature. Both input and IP samples were subjected to RNA extraction with the addition of 1 µl Precipitation Carrier. Ten microliters of resuspended RNA were used for reverse transcription and 2% of cDNA was used for quantitative PCR.

FLAG-tagged proteins were expressed in stable cell lines (2.5–3.0 × 10⁶ cells per 10 cm dish) induced for 48–72 h. The samples were lysed in 600 µl buffer E (20 mM HEPES-KOH (pH 7.9), 100 mM KCl, 10% glycerol, 1 mM DTT, Protease Inhibitor, 1 µg ml⁻¹ RNase A) and sonicated using the Bandelin Sonopuls mini20 with 15 pulses (2.5 mm tip, 1 s pulse, 50% amplitude). Concentration-adjusted lysates were subjected to immunoprecipitation for 2 h at 4 °C with overhead shaking using Anti-FLAG M2 Magnetic Beads (Sigma-Aldrich), the beads were washed three times for 5 min with buffer E, mild wash buffer (20 mM HEPES-KOH (pH 7.9), 137 mM NaCl, 2 mM MgCl₂, 0.2% Triton X-100, 0.1% NP-40) or medium wash buffer (20 mM HEPES-KOH (pH 7.9), 200 mM NaCl, 2 mM MgCl₂, 0.2% Triton X-100, 0.1% NP-40, 0.05% Na-deoxycholate). Co-immunoprecipitated proteins were eluted with SDS-sample buffer, separated by SDS-PAGE, and analyzed by immunoblotting.

FLAG-tagged SMG5 proteins or FLAG control were expressed in stable cell lines (2.5–3.0 × 10⁶ cells per 10 cm dish) induced with 1 µg ml⁻¹ doxycycline for 72 h. Samples were lysed and immunoprecipitated as described above, using mild wash buffer for washing steps and eluted in 42.5 µl of a 200 mg ml⁻¹ dilution of FLAG peptides (Sigma) in 1× TBS. 1 volume of 10% SDS was added and the samples were reduced with DTT and alkylated with CAA (final concentrations 5 mM and 55 mM, respectively). Tryptic protein digestion was performed using a modified version of the single pot solid phase-enhanced sample preparation (SP3)¹¹⁶. In brief, reduced and alkylated proteins were supplemented with paramagnetic Sera-Mag speed beads (Thermo Fisher Scientific) and mixed in a 1:1-ratio with 100% acetonitrile (ACN). After 8 min incubation protein-beads-complexes were captured using an in-house build magnetic rack and two times washed with 70% EtOH. Afterward, samples were washed once with 100% ACN, air-dried and reconstituted in 5 µl 50 mM Triethylammonium bicarbonate supplemented with 0.5 µg trypsin and 0.5 µg LysC and incubated overnight at 37 °C. On the next day, the beads were resuspended and mixed with 200 µl ACN, incubated for 8 min, and again placed on the magnetic rack. Tryptic peptides were washed once with 100% ACN, airdried, dissolved in 4% DMSO, and transferred into 96-well PCR tubes. After acidification with 1 µl of 10% formic acid, the samples were ready for LC-MS/MS analysis.

Proteomics analysis was performed by data-dependent acquisition using an Easy nLC1000 ultra-high-performance liquid chromatography (UHPLC) system coupled via nanoelectrospray ionization to a Q Exactive Plus instrument (all Thermo Scientific). Tryptic peptides were separated based on their hydrophobicity using a chromatographic gradient of 45 min (affinity enrichments samples) or 60 min (TurboID samples) with a binary system of buffer A (0.1% formic acid) and buffer B (80% ACN, 0.1% formic acid). In-house-made analytical columns (length: 50 cm, inner diameter: 75 µm) filled with 1.9 µm C18-AQ Reprosil Pur beads (Dr. Maisch) were used for separation. Using the 45 min chromatographic gradient, buffer B was linearly increased from 9 to 30% over 30 min followed by a steeper increase to 47% within 6 min. Finally, buffer B was increased to 95% within 4 min and stayed at 95% for 5 min to wash the analytical column. For the 60 min chromatographic gradient, buffer B was linearly increased from 3 to 30% over 40 min followed by an increase to 50% within 8 min. Finally, buffer B was increased to 95% within 1 min and the column washed for 10 at 95%. Full MS spectra (300–1750 m/z) were acquired with a resolution of 70,000, a maximum injection time of 20 ms, and an AGC target of 3e6. The top 10 most abundant peptide ions of each full MS spectrum were selected for HCD fragmentation (NCE: 26) with an isolation width of 1.8 m/z and a dynamic exclusion of 20 s. MS/MS spectra were measured with a resolution of 17,000, a maximum injection time of 60 ms and an AGC target of 5e5 for the 45 min gradient, and a resolution of 35,000, a max. injection time of 110 ms and an AGC target of 5e5 using the 60 min gradient.

MS RAW files were analyzed using the standard settings of the MaxQuant suite (version 1.6.17.0)¹¹⁷. Peptides were identified by matching against the human UniProt database using the Andromeda scoring algorithm¹¹⁸. Carbamidomethylation of cysteine was set as a fixed modification, methionine oxidation, and N-terminal acetylation as variable modification. Trypsin/P was selected as the digestion protein. A false discovery rate (FDR) <0.01 was used for the identification of peptide-spectrum matches and protein quantification. Intensities were calculated using the LFQ algorithm implemented in MaxQuant with the standard parameters. Data processing and statistical analysis were done in the Perseus software (version 1.6.5.0)¹¹⁹. LFQ-normalized protein intensities were Log2-transformed for normal distribution and filtered for at least three valid values in at least one sample group. Missing values were imputed by drawing random values from a 1.8 standard deviations downshifted, 0.3 standard deviations broad normal distribution. Statistical testing was performed using a two-sided Welch’s t-test with permutation-based FDR correction (FDR = 0.05, S0 = 0.1). Significantly different proteins were identified using the following cut-off: q-value <0.05, absolute log2 fold change >1. Visualization was performed with InstantClue¹²⁰, the R (version 4.0.4) package ggplot2 (version 3.3.3) or ComplexHeatmap (version 2.6.2)¹⁰⁹.

Stable WT or SMG7 KO cell lines were seeded (5 × 10⁶ cells per 10 cm dish) and reverse transfected using 6.25 µl Lipofectamine RNAiMAX and 200 pmol siRNA (control or SMG5). The expression of FLAG-TurboID-tagged UPF1 or control proteins was induced on the next day with 30 µg ml⁻¹ cumate. In this step the medium was also changed to high-glucose, GlutaMAX DMEM (Gibco) supplemented with 9% dialyzed fetal bovine serum (Gibco; A3382001; to suppress background biotinylation)¹²¹ and 1× Penicillin Streptomycin (Gibco). Proximity labeling by biotinylation was performed on the next day by the addition of 50 µM biotin for 15 min. Afterward, the cells were washed twice with PBS on ice, scraped in 1 ml ice-cold PBS, collected for 5 min at 100 × g and 4 °C, and finally resuspended in 200 µl phospho-RIPA buffer (50 mM Tris pH 8.0, 150 mM NaCl, 1% IGEPAL CA-630, 0.5% deoxycholate, 0.1% SDS, 1 µg ml⁻¹ RNase A) supplemented with 1 tablet of PhosSTOP (Roche), 100 µl EDTA-free HALT Protease and Phosphatase Inhibitor (ThermoFisher) and per 10 ml buffer. Samples were sonicated using the Bandelin Sonopuls mini20 with 10 pulses (2.5 mm tip, 1 s pulse, 50% amplitude). Fifty microliter input aliquots containing 100 µg of total protein were prepared and mixed with SDS-sample buffer. Concentration-adjusted lysates containing 1 mg total protein in 500 µl buffer were concentrated to approximately 100 µl in 0.5 ml Amicon Ultra centrifugal filter devices (3K cutoff) for 45 min at 4 °C and 14.000 × g, to minimize excess biotin in the sample. The concentrated sample was combined with 200 µl RIPA buffer (wash of centrifugal filter), mixed with 25 µl pre-washed Pierce Streptavidin Magnetic Beads (ThermoFisher), and incubated for 2 h at 4 °C with overhead shaking. The beads were washed four times for 5 min with 800 µl RIPA buffer, followed by one wash with 800 µl mild wash buffer (20 mM HEPES-KOH (pH 7.9), 137 mM NaCl, 2 mM MgCl₂, 0.2% Triton X-100, 0.1% NP-40). Biotinylated proteins were eluted with 50 µl 1× SDS-sample buffer, supplemented with 20 mM DTT and 3 mM biotin, for 15 min at 96 °C, followed by another elution with 25 µl and both eluates were combined. Aliquots (10 µl input, 12.5 µl eluate) of the samples were resolved on 10% polyacrylamide gels containing 25 µl TCE (2,2,2-trichloroethanol) to allow fluorescent visible detection of proteins¹²² and subsequently used for western blotting. Tryptic protein digestion and proteomics analysis were performed as described above. Of note, no SMG5 Label-Free Quantification (LFQ) intensities were obtained, although SMG5 peptides were only detected in the WT samples and not in the SMG7 KO.

Quantifications and calculations for other experiments were performed—if not indicated otherwise—with Microsoft Excel (version 1808) or R (version 4.0.4) and all plots were generated using IGV (version 2.8.12), GraphPad Prism 5, ggplot2 (version 3.3.3), or ComplexHeatmap (version 2.6.2). Boxplots were generated using the geom_boxplot() function of ggplot2 with the centerline representing the 50th percentile (median), whereas the lower and upper box limits correspond to the 25th and 75th percentiles. The whiskers extend from the box limits to the smallest or largest value no further than 1.5 * inter-quartile range. Data beyond the end of the whiskers are plotted individually.

Further information on research design is available in thelinked to this article. Nature Research Reporting Summary