Nonsense-mediated decay machinery in Plasmodium falciparum is inefficient and non-essential

Abstract

Nonsense-mediated decay (NMD) is a conserved mRNA quality control process that eliminates transcripts bearing a premature termination codon. In addition to its role in removing erroneous transcripts, NMD is involved in post-transcriptional regulation of gene expression via programmed intron retention in metazoans. The apicomplexan parasite Plasmodium falciparum shows relatively high levels of intron retention, but it is unclear whether these variant transcripts are functional targets of NMD. In this study, we use CRISPR-Cas9 to disrupt and epitope-tag the P. falciparum orthologs of two core NMD components: PfUPF1 (PF3D7_1005500) and PfUPF2 (PF3D7_0925800). We localize both PfUPF1 and PfUPF2 to puncta within the parasite cytoplasm and show that these proteins interact with each other and other mRNA-binding proteins. Using RNA-seq, we find that although these core NMD orthologs are expressed and interact in P. falciparum, they are not required for degradation of nonsense transcripts. Furthermore, our work suggests that the majority of intron retention in P. falciparum has no functional role and that NMD is not required for parasite growth ex vivo. IMPORTANCE In many organisms, the process of destroying nonsense transcripts is dependent on a small set of highly conserved proteins. We show that in the malaria parasite, these proteins do not impact the abundance of nonsense transcripts. Furthermore, we demonstrate efficient CRISPR-Cas9 editing of the malaria parasite using commercial Cas9 nuclease and synthetic guide RNA, streamlining genomic modifications in this genetically intractable organism.

Keywords: RNA-seq; intron retention; mRNA degradation; malaria; nonsense-mediated decay.

Fig 1

(A)Model of EJC-dependent NMD in humans. After ribosome stalling at a PTC, UPF2 and UPF3b associated with the downstream (>50–55nt) EJC will bind to UPF1. SMG1 phosphorylates and activates UPF1, resulting in the recruitment of SMG5 and SMG7 (involved in the further recruitment of mRNA decay factors) and the endonuclease SMG6. Drawn using biorender.com. (B)Presence of NMD-associated proteins in selected species (H. sapiens, D. melanogaster, C. elegans, A. thaliana, S. cerevisiae, T. thermophila, T. gondii, and P. falciparum). Proteins are grouped based on their role in NMD as determined in mammals. The NMD core is required for the initiation of NMD. EJC core proteins are deposited upstream of exon-exon boundaries after splicing and are required for EJC-dependent NMD. SMG1 is a phosphatidylinositol 3-kinase-related kinase (PIKK) which phosphorylates UPF1. SMG6 is an endonuclease. Both SMG5 and SMG7 are involved in the recruitment of UPF1 to cytoplasmic RNA degradation sites such as the exosome. This table was compiled using genes identified in reference (4). (C)Schematic protocol for disruption of the PfUPF1 and PfUPF2 genes using the Alt-R CRISPR Cas9 system (Integrated DNA Technologies). The gene-specific crRNA is annealed to the Cas9-binding tracrRNA to form gRNA (1) which is then complexed with recombinant Cas9 to form the gRNA:Cas9 RNP (2). The gRNA:Cas9 RNP is then mixed with a linearized DNA repair template encoding two homology regions (HRs) flanking a human dihydrofolate reductase (hDHFR) expression cassette. Finally, the DNA + gRNA:Cas9 mixture is electroporated into ring-stage parasites (3). (D)Growth analysis of WT, ΔPfUPF1, and ΔPfUPF2 asexual parasites. Parasite cultures were initiated at 1% parasitemia as determined by flow cytometry. Parasitemia was then measured by flow cytometry every 48h. Error bars represent SEM (n = 3). (E)Giemsa-stained WT, ΔPfUPF1, and ΔPfUPF2 gametocytes imaged at stages III, IV, and V of development.

Fig 2

(A and B) Mean-difference plots showing differential gene expression in ΔPfUPF1 and ΔPfUPF2 parasites. Highlighted genes are considered significantly upregulated (red) or downregulated (blue) with an adjusted P-value < 0.05 (Benjamini-Hochberg) as calculated using the limma “treat” method (log-fold-change >1) with voom normalization. (C)Schematic depicting the calculation of the global proportion of intron retention (PIR). An example gene with two exons (dark gray) and one intron (light gray) is shown. J3 reads span the spliced junction, E1I reads span the 5′ exon-intron boundary, and IE2 reads span the 3′ intron-exon boundary. For a given expression level bin, reads were summed according to their designation as E1I, J3, or IE2, and the global PIR was then calculated using the formula shown. Figure adapted from the ASpli reference manual (April 27, 2020 release). (D)Introns (n = 7523) were grouped into 10 equal-sized bins based on expression level (WT FPKM), and the mean PIRs and confidence levels were computed using the R package emmeans within each bin for WT, ΔPfUPF1, and ΔPfUPF2. Error bars represent 95% CI. (E)Introns were classified into two groups: non-3n if the intron length is not a multiple of 3 (i.e., retention of the intron causes a frameshift) and 3n if the intron length is a multiple of 3. Introns were divided equally into 100 bins based on length (bp) before plotting. (F)Introns were classified into three groups based on the effect of retention: those that induce neither a frameshift nor a PTC (3n introns = 143), introns that induce a frameshift but no PTC (non-3n introns = 34), and PTC-inducing introns (6,725 introns). Intron retention was then calculated globally within each bin. For plotting, the intron retention rates and confidence intervals were multiplied by the number of introns in each of the three groups. Error bars represent 95% CI.

Fig 3

(A)Expression of PfUPF1 in WT and ΔPfUPF1 parasites. The coverage plot shows Illumina RNA-seq reads mapped to the PF3D7_1005500 (PfUPF1) locus in WT and ΔPfUPF1 parasites (genomic coordinates Pf3D7_10_v3: 237506–242245, negative strand). Each plot represents reads from three biological replicates. (B)Phylogram created with annotated UPF1 sequences (teal box) and sequences most similar to the PF3D7_0703500 protein identified by BLASTp search (pink box) from the apicomplexans Eimeria tenella, Neospora caninum, and T. gondii. The phylogenetic tree was constructed using the maximum likelihood method and 1,000 bootstrap replicates. Bootstrap values <50 are not displayed. Scale bar for the branch length represents the number of substitutions per site. Accession numbers and full alignments are available at https://gitlab.com/e.mchugh/nmd-paper. Conserved UPF1 domains are displayed on top of the sequence alignment. Three longer unique P. falciparum sequences are highlighted in red.

Fig 4

(A)Strategy for introducing epitope tags to PfUPF1 and PfUPF2. Homology regions (HRs) targeting the C-terminal of the CDS (excluding the stop codon) and a region in the 3′ UTR were cloned on either side of a 3X-HA and hDHFR drug resistance cassette in the plasmid pHAX. (B)Immunofluorescence assay with PfUPF1-HA and PfUPF2-HA parasites. Infected RBCs were fixed with paraformaldehyde/glutaraldehyde, permeabilized with Triton X-100, and probed with rat anti-HA (1:300), followed by AlexaFluor anti-rat 568 (1:600). Images are maximum projections of wide-field deconvoluted z-stacks. Nuclei were visualized with DAPI (cyan) and HA signal is presented in magenta. BF, brightfield; scale bars = 3 μM. (C)Immunoprecipitation and Western blot of WT, PfUPF1-HA, and PfUPF2-HA. For each parasite line, the input (3% of the Triton X-100-soluble fraction) was loaded beside the immunoprecipitated eluate (100% of the IP). The membrane was probed with anti-HA (1:1,000). (D)Alignment of UPF3b sequences. The residues in human UPF3b that are important for EJC-binding are denoted with black dots [as determined in reference (38)]. Accession numbers and full alignments are available at https://gitlab.com/e.mchugh/nmd-paper.

Fig 5

(A)Map of PfUPF1-HA and PfUPF2-HA protein-protein interactions detected by co-IP and LC- MS/MS. The dashed line represents the interaction detected in a co-immunoprecipitation experiment with PyALBA4-GFP in P. yoelii in reference (46). Full protein names are listed in Tables 1 and 2.