Identification of SMG6 cleavage sites and a preferred RNA cleavage motif by global analysis of endogenous NMD targets in human cells

Abstract

In metazoans, cleavage by the endoribonuclease SMG6 is often the first degradative event in non-sense-mediated mRNA decay (NMD). However, the exact sites of SMG6 cleavage have yet to be determined for any endogenous targets, and most evidence as to the identity of SMG6 substrates is indirect. Here, we use Parallel Analysis of RNA Ends to specifically identify the 5' termini of decay intermediates whose production is dependent on SMG6 and the universal NMD factor UPF1. In this manner, the SMG6 cleavage sites in hundreds of endogenous NMD targets in human cells have been mapped at high resolution. In addition, a preferred sequence motif spanning most SMG6 cleavage sites has been discovered and validated by mutational analysis. For many SMG6 substrates, depletion of SMG6 resulted in the accumulation of decapped transcripts, an effect indicative of competition between SMG6-dependent and SMG6-independent NMD pathways. These findings provide key insights into the mechanisms by which mRNAs targeted by NMD are degraded.

Figure 1.

Effect of SMG6 or UPF1 depletion on the concentration of the full-length TCRβ68 transcript and its 3′-terminal SMG6 cleavage product. HeLa cells that expressed a TCRβ68 transcript harboring a PTC at codon 68 (TCRβ68) or a wild-type TCRβ transcript lacking a PTC (TCRβwt) were transfected with siRNAs directed against XRN1, SMG6 or UPF1, or with siGL2 (negative control). Total RNA extracted from these cells was used to prepare PARE libraries. (A) Detection of both a full-length transcript (TCRβ68 FL) and a 3′-terminal decay intermediate (TCRβ68 3′) by northern blot analysis. As a control, some cells were transfected with an siRNA-resistant SMG6 gene (SMG6^R) or a catalytically inactive variant thereof (SMG6^R-mut). In each case, the concentration of the full-length transcript relative to that in mock-transfected cells was calculated after normalization to GAPDH mRNA (internal standard). (B) D-plots for the TCRβ68 reporter identifying monophosphorylated 5′ ends detected by PARE in cells transfected with siXRN1 and either siGL2, siSMG6 or siUPF1. A map of the TCRβ68 transcript is shown above the D-plots. Alternating gray and white zones indicate exons. AUG, translation initiation codon; PTC, premature termination codon; TC, natural termination codon. Because the 5′-terminal portion of the reporter was derived from the human β-actin gene, TCRβ68-derived PARE sequences there cannot be distinguished from those originating from the endogenous β-actin transcript. Transcript positions represent the distance from the first nucleotide unique to the TCRβ68 reporter. (C) High-resolution D-plot identifying 5′ ends detected by PARE in RNA from cells transfected with siGL2 and siXRN1. Gray rectangle, PTC.

Figure 2.

Schematic diagrams for PARE, C-PARE and SPARE library construction. (A) PARE, (B) C-PARE and (C) SPARE libraries were constructed from poly(A)⁺ RNA in multiple steps (red, green or blue arrows). The procedures used to construct all of the libraries are shown in bold lettering, while those used to construct specific libraries are shown in regular lettering. 5′ ^m7Gppp, capped 5′ end; 5′ P monophosphorylated 5′ end; 5′ OH, hydroxylated 5′ end; (A)_n, polyadenylated 3′ end.

Figure 3.

Computational pipeline for identifying SMG6 targets. Input and final output are highlighted in gray. The data were filtered in five steps described in detail in Supplementary Methods. Site refers to a unique PARE MaxSeq in a gene. The pipeline did not include any step intended to favor the presence of a TC near the PARE MaxSeq.

Figure 4.

Representative D-plots of endogenous SMG6 targets identified by PARE. A transcript map is shown above each set of D-plots. Alternating gray and white zones represent exons. AUG, translation initiation codon; TC, natural termination codon; uAUG, predicted translation initiation codon of an upstream ORF; uTC, predicted termination codon of an upstream ORF; SelCys, selenocysteine codon. (A) RAE1–002, a transcript with a long 3′ UTR. (B) IFRD1–001, a transcript with a predicted uORF. (C) GPX1–001, a transcript with a selenocysteine codon. (D) ENY2–009, a transcript with an exon junction downstream of the TC. All of the D-plots were drawn with data from Biorep 1; similar effects were observed for Biorep 2.

Figure 5.

Selected human transcripts targeted by SMG6. Transcripts whose PARE MaxSeq decreased by at least 50% and whose mRNA abundance increased significantly (Cuffdiff q < 0.05) in both the siSMG6+siXRN1 and siUPF1+siXRN1 libraries as compared to the siGL2+siXRN1 control are shown. In each case, the magnitude of the effect is represented by a horizontal bar, and the sequence surrounding the principal site of SMG6 cleavage (arrow) is shown. Sequences conforming to the (U/A)-(G/A)-(A/C)-N-(C/U) motif are indicated in red, while sequences with one mismatch are indicated in blue.

Figure 6.

Degenerate pentameric motif common at SMG6 cleavage sites. (A) Sequence logo representation of the 10-nt region surrounding the PARE MaxSeq in endogenous SMG6 targets and siGL2+siXRN1 pipeline controls. (B) Frequency of the 5-nt motif at each position in SMG6 targets and control transcripts. The red bar indicates where the motif would be expected to start in SMG6 targets, 2 nt upstream from the cleavage site. (C) Mismatch frequency at various positions within PARE sites that conform to the SMG6 motif at all but one position. (D) Frequency of SMG6 targets and pipeline controls whose MaxSeq is surrounded by a pentamer that matches the SMG6 motif at 0–4 of the defined positions, versus the frequency that such matches would occur by chance.

Figure 7.

SPARE analysis of the pentameric motif. (A) The degenerate sequence of the motif. (B) PTC-proximal sequences of the SPARE reporters. The SMG6 cleavage site under investigation is marked by an arrow. The PTC is highlighted in gray. The altered motif sequences are highlighted in yellow. Red or blue letters identify the 5′ end of the expected cleavage products for sequence pentamers that matched the motif at every position (at 4 out of 5 positions). (C) SPARE D-plots. The bar heights in each D-plot were normalized to that of the abundant 5′ end 12 nucleotides downstream of the PTC. The red bar identifies the 5′ end resulting from cleavage at the PARE MaxSeq for the original TCRβ68 reporter. Red stars indicate the expected 5′-end generated by cleavage in pentamers that matched the motif at every position.

Figure 8.

Accumulation of decapped transcripts when SMG6 but not UPF1 is depleted. Arrows in the D-plots of representative transcripts (BEGAIN-003 and DPM1–004) in Biorep 1 identify 3′ cleavage products whose accumulation is SMG6- and UPF1-dependent. Circles indicate the principal location of the 5′ cap, as determined by C-PARE (bottom panel). Similar effects were observed in Biorep 2.