YTHDF2 destabilizes m(6)A-containing RNA through direct recruitment of the CCR4-NOT deadenylase complex

Abstract

Methylation at the N6 position of adenosine (m(6)A) is the most abundant RNA modification within protein-coding and long noncoding RNAs in eukaryotes and is a reversible process with important biological functions. YT521-B homology domain family (YTHDF) proteins are the readers of m(6)A, the binding of which results in the alteration of the translation efficiency and stability of m(6)A-containing RNAs. However, the mechanism by which YTHDF proteins cause the degradation of m(6)A-containing RNAs is poorly understood. Here we report that m(6)A-containing RNAs exhibit accelerated deadenylation that is mediated by the CCR4-NOT deadenylase complex. We further show that YTHDF2 recruits the CCR4-NOT complex through a direct interaction between the YTHDF2 N-terminal region and the SH domain of the CNOT1 subunit, and that this recruitment is essential for the deadenylation of m(6)A-containing RNAs by CAF1 and CCR4. Therefore, we have uncovered the mechanism of YTHDF2-mediated degradation of m(6)A-containing RNAs in mammalian cells.

Figure 1. m⁶A modification promotes deadenylation of RNAs.

(a) The reporter constructs of BG-PLAC2 and BG-PLAC2-mut. A 135-nt fragment from lncRNA PLAC2 was inserted into the 3′-UTR of the BG reporter (BG-PLAC2). Blank box indicate the open reading frame (ORF). Striped box indicate the inserted DNA fragment. BG-PLAC2-mut is identical to BG-PLAC2, except that the two adenosines (underlined) within the m⁶A motifs were mutated to thymidines. (b) The m⁶A-IP enrichment ratio of BG-PLAC2 relative to BG-PLAC2-mut. (c) The YTHDF2-RIP enrichment ratio of BG-PLAC2 relative to BG-PLAC2-mut. The RIP assay was performed using anti-FLAG M2 antibody with HeLa-tTA cells that stably express FLAG-tagged YTHDF2. (d) Deadenylation assay of BG-PLAC2 and BG-PLAC2-mut. The brief removal of tetracycline (tet) from the culture medium generated a homogenous population of BG mRNAs that underwent synchronous decay. RNA samples were collected at the indicated time intervals and subjected to site-specific cleavage by RNase H to produce 3′- and 5′-BG mRNA fragments. RNA samples were then separated by electrophoresis and detected by northern blotting. (e) Retention of the 5′-cap on BG-PLAC2 and BG-PLAC2-mut undergoing deadenylation. The BG-PLAC2 and BG-PLAC2-mut RNA samples collected at 0 and 7.5 h from d were treated or not treated with the 5′-phosphate-dependent exonuclease XRN-1. 18S rRNA, which lacks a 5′-cap, served as a positive control for XRN-1 activity. (f–j) The same experiments as in a–e for BG-SON and BG-SON-mut. Error bars, mean±s.d., n=3, biological replicates.

Figure 2. YTHDF2 destabilizes mRNA by hastening deadenylation as an initial step.

(a) Construct of BG-1boxB, which can bind to proteins fused with a λN peptide through the boxB sequence. (b) Decay of BG-1boxB mRNA in the presence of λN-FLAG-YTHDF2 or FLAG-YTHDF2. AG-GAPDH, a constitutively transcribed internal standard was co-transfected. (c) Graphs of the concentration of BG mRNA described in b as a function of time. The abundance of BG-1boxB mRNA remaining at each time point was normalized to that of AG-GAPDH. Error bars, mean±s.d., n=3, biological replicates. (d) Deadenylation assay of BG-1boxB mRNA in the presence of λN-FLAG-YTHDF2 or FLAG-YTHDF2. (e) Retention of the 5′-cap on BG-1boxB mRNA in the presence of λN-FLAG-YTHDF2 or FLAG-YTHDF2. (f) Deadenylation assay of BG-1boxB mRNA tethered with an N-terminal P/Q/N-rich region (YTHDF2-N) or a C-terminal YTH domain (YTHDF2-C). (g) Construct of BG-2boxB-ORF, which has two boxB sequences inserted in-frame to the ORF of the BG reporter. (h) Deadenylation assay of BG-2boxB-ORF mRNA in the presence of λN-FLAG-FL, λN-FLAG-YTHDF2 or FLAG-YTHDF2.

Figure 3. The CCR4–NOT complex is responsible for YTHDF2-mediated RNA deadenylation.

(a–c) Interaction between YTHDF2 and deadenylases or decapping enzymes. HEK 293 cells were co-transfected with a plasmid encoding FLAG-tagged YTHDF2 and either a V5-tagged CAF1, CCR4A, PAN2, PAN3, PARN, DCP1A, DCP2, FL or an HA-tagged CNOT1 or enhanced green fluorescent protein (EGFP). Lysates were subjected to immunoprecipitation by using anti-FLAG affinity gel. Input and co-purified proteins were blotted by probing with corresponding antibodies. Nonspecific bands were indicated by the asterisk. (d) Interaction between YTHDF2 and CAF1 or CAF1-141/147, a mutant version of CAF1 with M141K and L147K double substitution. (e) Interaction between stably expressed FLAG-tagged YTHDF2 and endogenous CNOT1 in HeLa-tTA cells. (f) Strategy of screening of functional deadenylases that are responsible for YTHDF2-mediated deadenylation. BG-1boxB reporter was co-transfected with λN-FLAG-YTHDF2 and either a wild-type or catalytically inactive deadenylase. (g) Deadenylation assay of BG-1boxB tethered by YTHDF2 with either a co-transfected wild-type or catalytically inactive deadenylase.

Figure 4. The SH domain of CNOT1 directly interacts with YTHDF2.

(a) Schematic diagram of human CNOT1 and the fragments used in b and c. (b,c) Interaction between YTHDF2 and CNOT1 fragments. HEK 293 cells were co-transfected with a plasmid encoding FLAG-tagged YTHDF2 and an HA-tagged CNOT1 fragment or a negative control (EGFP). Immunoprecipitation and immunoblotting were performed as in Fig. 3a–c. (d) Schematic diagram of human YTHDF2 and the fragments used in e and f. (e,f) Determination of a direct interaction between CNOT1-SH and YTHDF2 fragments by GST pull-down assays. Recombinant His-tagged YTHDF2 or its fragments and GST-tagged CNOT1-SH were purified, and equal amounts of both proteins were subjected to GST pull-down assay followed by SDS–PAGE separation and Coomassie Blue staining.

Figure 5. Recruitment of CCR4–NOT through CNOT1 is essential for the degradation of endogenous m⁶A-containing RNAs.

(a) Deadenylation assay of BG-PLAC2 and BG-PLAC2-mut on knocking down of endogenous METTL3 or CNOT1 by siRNA. siNC served as a negative control. (b) Deadenylation assay of endogenous ACTB mRNA on knocking down of endogenous METTL3, CNOT1 or CAF1 by siRNA. (c) Deadenylation assay of BG-1boxB mRNA in the presence of λN-FLAG-YTHDF2 on knocking down of endogenous CNOT1 by siRNA. (d–f) Schematic diagrams of YTHDF2-mediated deadenylation of m⁶A-containing RNAs and the dominant-negative effect of YTHDF2-C or CNOT1-SH. m⁶A is recognized by YTHDF2, which further recruits the CCR4–NOT complex by interacting with CNOT1 (d). Overexpressed YTHDF2-C occupies m⁶A sites but is incapable of recruiting the CNOT1 subunit of the CCR4–NOT complex; thus, it impairs the deadenylation and decay of m⁶A-containing RNAs (e). Overexpressed CNOT1-SH binds to endogenous YTHDF2 but is incapable of recruiting the catalytic subunits CAF1 and CCR4A/B; thus, it impairs the deadenylation and decay of m⁶A-containing RNAs (f). (g) Expression level of endogenous target mRNAs measured by quantitative reverse transcriptase–PCR on overexpression of YTHDF2-C, CNOT1-SH or the negative control green fluorescent protein (EGFP). All values were normalized to HPRT1 mRNA, a housekeeping gene previously reported to contain no m⁶A modifications and is not bound by YTHDF2. Error bars, mean±s.d., n=3, biological replicates. **P<0.01, *P<0.05, t-test. (h–j) Half-life of FBXL19, ZBTB7B and HPRT1 mRNA on overexpression of YTHDF2-C, CNOT1-SH or the negative control EGFP after transcription inhibition (TI). All values were normalized to 18S rRNA. Error bars, mean±s.d., n=3, biological replicates.