Readthrough of stop codons under limiting ABCE1 concentration involves frameshifting and inhibits nonsense-mediated mRNA decay

Abstract

To gain insight into the mechanistic link between translation termination and nonsense-mediated mRNA decay (NMD), we depleted the ribosome recycling factor ABCE1 in human cells, resulting in an upregulation of NMD-sensitive mRNAs. Suppression of NMD on these mRNAs occurs prior to their SMG6-mediated endonucleolytic cleavage. ABCE1 depletion caused ribosome stalling at termination codons (TCs) and increased ribosome occupancy in 3' UTRs, implying enhanced TC readthrough. ABCE1 knockdown indeed increased the rate of readthrough and continuation of translation in different reading frames, providing a possible explanation for the observed NMD inhibition, since enhanced readthrough displaces NMD activating proteins from the 3' UTR. Our results indicate that stalling at TCs triggers ribosome collisions and activates ribosome quality control. Collectively, we show that improper translation termination can lead to readthrough of the TC, presumably due to ribosome collisions pushing the stalled ribosomes into the 3' UTR, where it can resume translation in-frame as well as out-of-frame.

Figure 1.

Depletion of ABCE1 inhibits NMD on many NMD-sensitive transcripts. (A) Western blot analysis to assess efficacy of ABCE1 knockdown in the samples used for RNA-seq. 100% and 10% of the control knockdown sample (CTRL) and 100% of the ABCE1 KD sample were loaded. CPSF-100 served as loading control. A representative blot of three independently performed experiments is shown. (B) Volcano plot showing RNA level changes upon ABCE1 KD from the RNA-seq data along the x-axis and P-values for each differential expression value along the y-axis. For each transcript, the average log2 fold change of three replicates for ABCE1 KD compared to control KD is shown. Red dots denote previously identified high confidence NMD targets identified (5), dark and light grey dots show transcripts with and without significant changes (P-value ≥ 0,05), respectively. Green and blue dots depict upregulated and not upregulated known NMD targets that were validated by RT-qPCR in (C). The dotted and the dashed lines indicate log2 fold changes of ±1 and ±2, respectively. (C) Relative mRNA levels, normalized to β-actin (ACTB), of indicated transcripts were measured by RT-qPCR from total RNA samples of cells with control KD, ABCE1 KD or UPF1 KD. Mean values and standard deviations (SD) of four replicates for ABCE1 KD and 2 replicates for UPF1 KD are shown. P-values > 0.05 are indicated as ns (not significant), *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001 and ****P ≤ 0.0001. Of HNRNPL, TRA2B, SRSF11 and TMEM208, an alternatively spliced NMD-sensitive transcript (NMD) and the NMD-resistant protein coding mRNA (PROT) were measured using splice variant-specific primers. GADD45B and SMG5 are NMD-sensitive mRNAs.

Figure 2.

Depletion of ABCE1 increases mRNA levels of NMD-sensitive reporters before SMG6-mediated endonucleolytic cleavage. (A, B) Top: Schematic representation of TCRβ and miniμ reporter mRNAs. Red dots indicate the position of premature termination codons (ter68 in TCRβ and ter310 in miniμ), green dots represent the normal termination codon. The position of the probes used for northern blot analysis are indicated as black lines below the constructs. Bottom: Western blot analysis monitoring the efficacy of the ABCE1 KD in HeLa cells stably expressing TCRβ WT or ter68, and miniμ WT or ter310, respectively. CPSF-100 served as loading control. Representative blots of three independently performed experiments are shown. (C, D) Relative levels of TCRβ or miniμ mRNA, normalized to β-actin (ACTB) mRNA, were measured by RT-qPCR. Mean values and SD of 4 replicates and P-values are shown as in Figure 1. (E, F) Northern blot analysis of total RNA isolated from HeLa cells expressing TCRβ (WT or ter68) or miniμ (WT or ter310) cells with and without ABCE1 KD. 18S rRNA levels detected from the agarose gel stained with ethidium bromide served as loading control.

Figure 3.

NMD-sensitive mRNAs are associated with polysomes in ABCE1 depleted cells. (A) Overall translation activity in cells with and without ABCE1 KD was measured using the SUnSET technique that incorporates puromycin in newly synthesized proteins (59). 100%, 50% and 25% of each protein extract were loaded on an SDS-PAGE, transferred to a nitrocellulose membrane and probed with antibodies against puromycin, ABCE1 and CPSF100 to monitor translation activity, ABCE1 knockdown efficacy and equal loading, respectively. The blot is representative of 5 independently performed experiments, of which the incorporated puromycin in ABCE1 KD samples relative to Control KD samples was quantified and plotted (below). (B) Top: Western blot of HeLa lysates to assess efficacy of the ABCE1 knockdown in the samples used for the polysome profiling experiments. Tyrosine Tubulin (TyrTub) served as loading control. The blot is representative of three independently performed experiments. Bottom: Polysome profiles of lysates from cells without (CTRL) or with ABCE1 KD. Similar results were obtained in three independently performed experiments. The fractions corresponding to monosomes (M) and early polysomes (P) (highlighted in grey) were used for RNA isolation and RT-qPCR analysis (C). (C) Relative levels of the NMD-sensitive (NMD) and the NMD-insensitive (PROT) isoforms of HNRNPL and TMEM208 RNAs were measured by RT-qPCR in the monosome (M) and polysome (P) fraction and ratios between the polysome and monosome fractions of each condition is depicted. Mean values and standard deviations of three independent experiments are shown.

Figure 4.

Enrichment of ribosome occupancy at termination codon and in the 3′ UTR in ABCE1 depleted cells. (A) Top: Metagene analysis of ribosome-protected footprints from ABCE1 KD (left) or control KD (right). 12 401 transcripts were aligned to the stop codon and mapped reads from 300 nucleotides upstream to 100 nucleotides downstream of the stop codon are shown. Bottom: Heatmap of ribosome-derived reads of all the transcripts considered for the metagene analysis, ordered according to their ribosome occupancy at the stop codon under ABCE1 KD. The 2000 transcripts with the highest ribosome occupancy at the stop codon (green line) and the 2000 transcripts with the lowest ribosome coverage at stop codon (red line) are indicated. Previously identified NMD-sensitive transcripts (5) are depicted with a bright line between the heatmap panels. Each pixel corresponds to 10 transcripts. (B) Top: Metagene analysis of ribosome coverage as in A. Bottom: Heatmap of ribosome-derived reads as in A, this time ordered according to their ribosome occupancy in the 3′ UTR under ABCE1 KD. The 2000 transcripts with the highest ribosome occupancy in the 3′ UTR are marked with a yellow line. The position of the NMD-sensitive transcripts in the heatmap is indicated as in A. (C) Analysis of mean ribosome occupancy in the 3′ UTR relative to mean ribosome occupancy in CDS. For each replicate, the ratio between the RPFs in the 3′ UTR and in the CDS was determined, then averaged and log transformed. P-values were calculated using Anova statistic test (P ≤ 1e–16). (D) Analysis of mean ribosome occupancy in the 3′ UTR as in C, but performed separately for NMD-sensitive transcripts and all other transcripts (Anova statistic test for ‘NMD targets’ and for ‘Others’; P ≤ 1e–16).

Figure 5.

ABCE1 KD promotes frameshift-prone TC readthrough. (A) Schematic representation of the dual luciferase readthrough reporter construct. (B) Western blot analysis to assess efficacy of ABCE1 knockdown in cells expressing the different readthrough reporter constructs. CPSF-100 served as loading control. A representative blot of three independently performed experiments is shown. (C) The ratios between Firefly and Renilla luciferase activities were determined for each of the three the stop codon-containing reporters (UAA, UAG, UGA) in cells with a control or ABCE1 KD and set relative to the ratio measured with the no-stop reporter (UCC), representing the percentage of readthrough at the three different stop codons. Mean values and SD from three independent experiments and P-values are shown as in Figure 1. (D) Schematic representation of the modified dual luciferase readthrough reporter constructs, in which the Firefly luciferase reading frame was shifted by 1 nucleotide relative to the Renilla luciferase reading frame. (E) Assessment of ABCE1 KD efficacy in cells expressing the frameshifted dual luciferase reporter constructs as in (B), except that that tyrosine tubulin served as loading control. (F) The frameshifted dual luciferase readthrough reporter constructs were expressed and the ratios between Firefly and Renilla luciferase activities were determined as in (C), depicting the Firefly/Renilla ratios of the frameshifted constructs as a percentage of the ratio measured with the no-stop reporter. (G) Ig-μ protein detection by western blotting of miniμ-expressing HeLa cells depleted of ABCE1 or after G418 treatment using a polyclonal antibody against the constant region of Ig-μ (C1vC4). Bands corresponding to truncated (translation termination at ter310) and full-length Ig-μ protein are indicated. Based on their size range, the bands denoted with an asterisk most likely represent truncated Ig-μ chains resulting from out-of-frame readthrough events of ter310. CPSF-100 served as loading control and the efficacy of ABCE1 depletion was also assessed. The blot is representative of three independently performed experiments with similar results. (H) Detection of the out-of-frame readthrough product eGFP+S (eGFP with a C-terminal SPOT tag in the +1 frame) by western blotting of protein immunoprecipitated with anti-SPOT nanobody (a-SPOT IP) from cell lysates with control or ABCE1 KD and detected with anti-GFP antibody (a-GFP). eGFP* depicts unspecifically co-precipitated eGFP, which is highly abundant in the lysate (see input, a-GFP panel). Efficacy of the ABCE1 KD was monitored in the input samples.

Figure 6.

ABCE1 KD promotes ribosome collision and RQC activation. (A) Metagene analysis of ribosome-protected footprints from control and ABCE1 KD. Transcripts were aligned to the stop codon and mapped reads from 300 nucleotides upstream to 100 nucleotides downstream of the stop codon are shown. The Top 3’000 transcripts with the highest ribosome occupancy in the 3′ UTR in the ABCE1 KD were included in the analysis (see heatmap in Figure 4B). (B) Polysome profiles of lysates from cells with control or ABCE1 KD. Fractions corresponding to monosomes (M) or disomes (D) (highlighted in light grey) were used for protein isolation and mass spectrometric analysis. (C) The mass spectrometry (MS) data for a subgroup of proteins involved in RQC is depicted. The average of the ratio of the normalized spectral abundance factor (NSAF) for each protein between ABCE1 KD and control KD is shown, averaged from two independent experiments. Monosome and disome fractions were analyzed separately. For the analysis of ubiquitinated proteins, the ubiquitin modification was detected by MS, an average ubiquitin-modification score was calculated for each protein and the ratio of this score between ABCE1 KD and control KD is shown. A value of 0.0 indicates that no ubiquitination was detected. (D) Validation of MS results by western blotting of proteins extracted from the monosome fractions obtained in the polysome profiling experiments (B). Antibodies against RPS3 and RPS27A detect both non-ubiquitinated as well as mono- and di-ubiquitinated forms of the respective protein. RPL4 served as loading control.

Figure 7.

Model for two different modes of stop codon readthrough under limiting ABCE1 concentrations. During normal translation termination, disassembly and recycling of the ribosomal subunits is promoted by ABCE1 after stop codon recognition and release of the nascent peptide chain. In the absence of ABCE1, ribosomes fail to terminate properly and stall at the stop codon for prolonged time. Prolonged stalling of a ribosome at the stop codon on the one hand increases the chance for the binding of a near-cognate tRNA leading to canonical readthrough. On the other hand, this also increases the probability of the following ribosome to run into stalled ribosome and collectively our data suggests that this leads to a ribosome collision-induced push of the terminating ribosome into the 3′ UTR, where it can resume translation in any of the three frames. Start and stop codons are depicted in green and red, the CDS and the encoded protein chain are shown in dark blue, while the 3′ UTR and any peptide sequence encoded therein are depicted in light blue.