PCBP1/2 and TDP43 Function as NAT10 Adaptors to Mediate mRNA ac4C Formation in Mammalian Cells

Abstract

Massive numbers of modified bases in mRNAs sculpt the epitranscriptome and play vital roles in RNA metabolism. The only known acetylated RNA modification, N-4-acetylcytidine (ac⁴C), is highly conserved across cell types and among species. Although the GCN5-related acetyltransferase 10 (NAT10) functions as an ac⁴C writer, the mechanism underlying the acetylation process is largely unknown. In this study, the NAT10/PCBP/TDP43 complex mediated mRNA ac⁴C formation in mammalian cells is identified. RNA-binding proteins (RBPs) are identified, affiliated with two different families, poly(rC)-binding protein 1/2 (PCBP1/2) and TAR DNA binding protein 43 (TDP43), as NAT10 adaptors for mRNA tethering and substrate selection. Knockdown of the adaptors resulted in decreased mRNA acetylation abundance in HEK293T cells and ablated cytidine-rich ac⁴C motifs. The adaptors also affect the ac⁴C sites by recruiting NAT10 to their binding sequences. The presence of the NAT10/PCBP/TDP43 complex in mouse testes highlights its potential physiological functions in vivo. These findings reveal the composition of the mRNA ac⁴C writer complex in mammalian cells and expand the knowledge of mRNA acetylation and ac⁴C site preferences.

Keywords: N‐4‐acetylcytidine; RNA binding protein; mRNA modification; posttranscriptional regulation; testis; transcriptome.

Figure 1

Identification and confirmation of TDP43 and PCBP1/2 as NAT10 adaptors. A) Scheme of N‐4‐acetylcytidine formation in mammalian RNAs. THUMPD1 and box C/D snoRNAs facilitate NAT10 in tRNA and 18S rRNA ac⁴C production, respectively. In mRNA ac⁴C formation, the presumed adaptor remains unknown. B) Comparison of NAT10 interactomes in HEK293T cells and mouse testis. HEK293T cells expressing N‐terminus FLAG‐tagged NAT10 and wild‐type (WT) mouse testis were subjected to affinity purification with anti‐FLAG and anti‐NAT10 antibody, respectively, and mass spectrometry analysis. RNA‐binding proteins (RBPs) including PCBP1, PCBP2, and TDP43 presented affinity toward NAT10.^{[ 15 ]} C) Co‐immunoprecipitation (co‐IP) results showing the interactions between FLAG‐tagged NAT10 and HA‐tagged TDP43, PCBP1, and PCBP2, which were resistant to RNase A digestion. D) Endogenous immunoprecipitation (endo‐IP) results showing spontaneous coordination of TDP43, NAT10, and PCBP1/2 in 293T cells. Anti‐TDP43 antibody was applied in endo‐IP. Arrows indicated the beads band. E‐F) Diagram of N‐terminal truncated PCBP1/2 constructs and validation of the interaction between NAT10 and WT or mutant PCBPs. Co‐IP results indicated that the hnRNP K homology (KH) 1 domain in PCBP1/2 was responsible for their conjugation with NAT10.

Figure 2

Depletion of PCBP1/2 and TDP43 resulted in decreased mRNA ac⁴C abundance in HEK293T cells. A) Comparison of the mRNA expression levels of PCBP1/2, TDP43, and NAT10 between the control group (siNC) and upon PCBP1/2 or TDP43 knockdown (siPCBP1/2 and siTDP43, respectively) by RT‐qPCR. Gene expression levels were normalized to GAPDH. Mean ± SEM, n = 3. P‐values were calculated using the two‐tailed Student's t‐test between the knockdown groups and siNC group. ***P < 0.001. n.s., not significant. B) Comparison of the protein expression levels of PCBP1/2, TDP43, and NAT10 between the control group (siNC) and PCBP1/2 or TDP43 knockdown groups. DDB1 was blotted as a loading control. C‐E) LC‐MS/MS detection of ac⁴C abundance (ac⁴C/C) in total RNA, poly(A) RNA, and non‐poly(A) RNA in the control group and in PCBP1/2, TDP43, or NAT10 knockdown groups. The ac⁴C abundance was normalized to the siNC group in each RNA type. Mean ± SEM. *P < 0.05, **P < 0.01. F‐H) LC‐MS/MS detection of m⁶A, m⁵C, and f⁵C abundance (m⁶A/A, m⁵C/C, and f⁵C/C, respectively) in poly(A) RNA in the control group and upon PCBP1/2, TDP43, or NAT10 knockdown. The indicated modification abundance was normalized to the control group. Mean ± SEM.

Figure 3

Transcriptome‐wide mapping of ac⁴C in HEK293T mRNAs. A) Schematic illustration of ac⁴C RNA‐immunoprecipitation tandem sequencing analysis (acRIP‐seq). Using oligo (dT)₁₅ beads, poly(A) RNAs were enriched from total RNA extracted from HEK293T cells and subjected to affinity purification with anti‐ac⁴C antibody, cDNA pool construction, and further sequencing analysis. B‐C) ac⁴C(+) poly(A) RNAs were defined based on transcript enrichment levels in the ac⁴C‐enriched groups relative to the inputs and the IgG‐enriched groups in the WT (B) and the siTDP43 (C) groups. Only transcripts that passed two‐tailed Student's t‐tests both in comparison to the inputs and the IgG groups were selected as pooled ac⁴C targets (P < 0.05). FC, fold change. D) Heatmap indicating enrichment levels of WT ac⁴C targets in WT and siTDP43 mRNAs. The color key from red to blue indicated relative enrichment extents from high to low. E) IGV browser views of transcript reads in RRBP1 (highly acetylated), ZFP36L2 (moderately acetylated), and EEF1A1 (not‐acetylated) transcripts mapped to the human reference genome (hg19). Reads in the acRIP, IgG‐enriched, and input groups are presented. The intron/exon (line/box) genomic structure is shown in dark blue. F) Enriched sequence motif analysis of ac⁴C peak clusters identified by acRIP‐seq. Up, ac⁴C motif in WT 293T mRNAs (P = 1.5E‐70). Down, ac⁴C motif in siTDP43 mRNAs (P = 3.1E‐35). Binding motifs were analyzed by the MEME motif. G) Overlap between ac⁴C‐modified transcripts and TDP43‐bound mRNAs in 293T cells.^{[ 18a ]} H) Enriched sequence motif of ac⁴C transcripts bound by TDP43. The top 2 sequences were presented.

Figure 4

Depletion of PCBP1/2 and TDP43 resulted in decreased ac⁴C abundance or loss of acetylation in WT ac4C mRNAs. A) Schematic of acRIP‐qPCR analysis. RNA samples extracted from the control and PCBP1/2, TDP43, or NAT10 knockdown groups were incorporated with the in vitro transcribed ac⁴C‐containing mouse β‐globin probe and ac⁴C‐null Egfp probe and enriched by anti‐ac⁴C antibody‐conjugated beads. The enriched RNAs by affinity purification were further reverse transcribed with oligo(dT)₃₀ primer. B‐C) Validation of RT‐qPCR results of 18S and 5S rRNA recovered from acRIP. 18S rRNA served as a known ac⁴C target control. Mean ± SEM. ***P < 0.001. n.s., not significant. D‐F) RT‐qPCR results showing that low acetylated mRNAs (GAPDH and EEF1A1), highly acetylated mRNAs (RRBP1, RBBP6, and UPF3B), and moderately acetylated mRNAs (FUS, ZFP36L2, and LAMP1) varied in acRIP enrichment. Mean ± SEM. *P < 0.05. G‐H) IGV browser views of transcript reads of ac⁴C targets that either contained or did not contain 5′‐UTR acetylation peaks (UPF3B in G and LAMP1 in H) in the ac⁴C‐, IgG‐enriched and the input in the control group. Enlarged views of the 5′‐UTR and selected exons within the CDS are presented. UTR, untranslated region. CDS, coding sequence. I‐J) RT‐qPCR results showing diverted enrichment levels of the 5′‐UTR of UPF3B (I) and LAMP1 (J) in the indicated groups. Mean ± SEM.

Figure 5

PCBP1/2 and TDP43 facilitated the binding of NAT10 to ac⁴C‐preferred mRNAs. A) Western blots of immunoprecipitated PCBP1, PCBP2, and TDP43 in endogenous RNA‐immunoprecipitation (RIP). Rabbit isotype IgG was applied as the control. B‐D) RT‐qPCR results showing that mRNAs that served as non‐preferred (GAPDH and EEF1A1), highly preferred (RRBP1, RBBP6, and UPF3B), and moderately preferred (FUS, ZFP36L2 and LAMP1) ac⁴C substrates were tethered by PCBP1/2 and TDP43 in varying degrees. Mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001. n.s. not significant. E‐F) RT‐qPCR results showing that acetylation‐preferred mRNA 5′‐UTRs were diversely tethered by PCBP1/2 and TDP43. Mean ± SEM. G) Western blots of immunoprecipitated NAT10, TDP43, PCBP1, and PCBP2 in RIP by anti‐NAT10 antibody in the control (siNC) and the adaptor‐depleted groups (siPCBP1/2 and siTDP43, respectively). The interaction between TDP43 and NAT10 was maintained when PCBP1/2 was depleted. Vice versa. H) Schematic illustration of NAT10/PCBP/TDP43 complex in mRNA acetylation. Under normal conditions, NAT10/PCBP1/TDP43 formed a stable heterogeneous tetramer, with PCBPs binding to NAT10 through their KH1 domain. By tethering mRNAs, PCBP1/2 and TDP43 recruited NAT10 to preferred mRNAs and NAT10 transferred acetyl groups to cytidines at the indicated sites. Loss of PCBP1/2 or TDP43 incurred instability in the complex, resulting in unspecific binding to non‐acetylation mRNA substrate or even failure in mRNA binding, consequently leading to aberrant ac⁴C formation. I‐K) RT‐qPCR results showing the varied connection between NAT10 and mRNAs of the indicated acetylation‐preferred groups upon PCBP1/2 and TDP43 knockdown. Mean ± SEM. L‐M) RT‐qPCR results showing the diverted connection between NAT10 and mRNA 5′‐UTRs of the UPF3B (L) and LAMP1 (M) upon PCBP1/2 and TDP43 knockdown. Mean ± SEM.

Figure 6

Affinity between PCBP1 and mRNA affected ac⁴C site preference. A) IGV browser shots of transcript reads of RRBP1 exon 14 – 16 mapping in acRIP‐seq mapping to the human reference genome. Nucleotides 3299–3313 within RRBP1 ^exon15 (CGCCAGCUCCCGCGG) were predicted to harbor ac4C according to PACES, further designated as RRBP1 ^site. B) RT‐qPCR results confirming RRBP1 ^exon15 mRNA acetylation in the control group (siNC) but lost upon PCBP1/2, TDP43, or NAT10 depletion. Mean ± SEM. *P < 0.05, ***P < 0.001. C) RT‐qPCR results showing PCBP1 affinity in interaction with RRBP1 ^exon15 mRNA in endogenous RIP. Mean ± SEM. n.s. not significant. D) RT‐qPCR results showing the affinity of NAT10 toward RRBP1 ^exon15 mRNA decreased or lost upon TDP43 and PCBP1/2 knockdown, respectively. Mean ± SEM. E) Illustration of RNA probe designed for RNA pull‐down. RRBP1 ^site was cloned from HEK293T cDNA and in vitro transcribed with biotinylated‐uridine incorporation directly or after neutral mutation in which most cytidines within RRBP1 ^site were substituted by other nucleosides (C3299U, C3302G, C3305U, C3307G, C3308U, C3309A, and C3311G, indicated as RRBP1 ^site1–WT and RRBP1 ^site1–mut, respectively). F) RNA pull‐down results indicated PCBP1/2 presented specific affinity toward WT RRBP1 ^site. G) RNA pull‐down results indicated diverted affinity between RRBP1 ^exon15 and NAT10/PCBP/TDP43 complex subunits in the control group and upon PCBP1/2, TDP43, and NAT10 knockdown. H) Schematic of RNA pull‐down analysis with cytidine and acetylated cytidine incorporated probes. WT and mutated RRBP1 exon 15 were in vitro transcribed with cytidines and acetylated cytidines as indicated. Biotinylated uridines were also incorporated for further enrichment. The RNA probes were mixed with the indicated cell lysates and subjected to affinity pull‐down. I) RNA pull‐down results showing the affinity between the PCBP/TDP43/NAT10 complex and the indicated RNA probes. PCBP1 specifically interacted with RRBP1 ^exon15, discarding the incorporation of ac⁴C, while PCBP2 merely contacted with ac⁴C(–) RNAs. TDP43 presented no difference in binding with ac⁴C‐containing or ac⁴C‐lacking mRNAs. The gray values of the bands were quantified by ImageJ and normalized to the control group pulled down through probes containing WT sequences. #N/A indicated that no band detected.

Figure 7

NAT10/PCBP/TDP43 complex functioned in mRNA acetylation in mouse testes. A) Endo‐IP results showing the spontaneous coordination of TDP43, NAT10, and PCBP1/2 in mouse testes. Anti‐TDP43 antibody was applied in endo‐IP. Arrows indicated the beads band. B) Western blot results showing TDP43 expression in spermatocytes isolated from WT mouse testes using flow cytometry sorting (FACS) (LZ, leptotene and zygotene, PD, pachytene and diplotene, MII, metaphase II and RS, round spermatids). α‐tubulin was blotted as a loading control. C) Acetylated mRNAs in mouse testes were defined through enrichment abundance between the acRIP and the inputs and IgG‐enriched groups. FC, fold change. D) Heatmap indicating enrichment levels of WT mouse testes ac⁴C targets. The color key from red to blue indicates relative enrichment extents from high to low. E) Enriched sequence motif analysis of ac4C peak clusters identified by acRIP‐seq in mouse testis. A CCHCAGSHC (H = C/U/A, S = C/G, P = 1.5E‐7) motif was detected by MEME analysis. F) IGV browser views of highly acetylated (Enho 3′‐UTR), moderately acetylated (Hoxd9), and not‐acetylated (Eef1a1) transcripts mapping to the mouse reference genome (mm10). Reads in the acRIP, IgG, and input groups are presented. The intron/exon (line/box) genomic structure is shown in dark blue. G) RT‐qPCR results showing diverted acetylation abundance of highly acetylated (3′‐UTR of Enho), moderately acetylated (Hoxd9), and low acetylated mRNAs (Eef1a1). Mean ± SEM. *P < 0.05, ***P < 0.001. H) RT‐qPCR results showing TDP43 affinity toward preferred ac⁴C‐targeted mRNAs in mouse testes. Mean ± SEM. **P < 0.01.