Unveiling the role of PUS7-mediated pseudouridylation in host protein interactions specific for the SARS-CoV-2 RNA genome

Abstract

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), a positive single-stranded RNA virus, engages in complex interactions with host cell proteins throughout its life cycle. While these interactions enable the host to recognize and inhibit viral replication, they also facilitate essential viral processes such as transcription, translation, and replication. Many aspects of these virus-host interactions remain poorly understood. Here, we employed the catRAPID algorithm and utilized the RNA-protein interaction detection coupled with mass spectrometry technology to predict and validate the host proteins that specifically bind to the highly structured 5' and 3' terminal regions of the SARS-CoV-2 RNA. Among the interactions identified, we prioritized pseudouridine synthase PUS7, which binds to both ends of the viral RNA. Using nanopore direct RNA sequencing, we discovered that the viral RNA undergoes extensive post-transcriptional modifications. Modified consensus regions for PUS7 were identified at both terminal regions of the SARS-CoV-2 RNA, including one in the viral transcription regulatory sequence leader. Collectively, our findings offer insights into host protein interactions with the SARS-CoV-2 UTRs and highlight the likely significance of pseudouridine synthases and other post-transcriptional modifications in the viral life cycle. This new knowledge enhances our understanding of virus-host dynamics and could inform the development of targeted therapeutic strategies.

Keywords: MT: RNA/DNA editing; PUS7; RNA-binding proteins; SARS-CoV-2; catRAPID; nanocompore; nanopore direct-RNA sequencing; protein-RNA interactions; proximity biotinylation; pseudouridine.

Graphical abstract

Figure 1

Experimental approach to investigate SARS-CoV-2 interactome (A) Schematic representation of the SARS-CoV-2 RNA fragments selected to be studied with the RaPID-MS strategy. The scheme shows the fragments names, positions within the SARS-CoV-2 genome and degree of overlap between them. (B) catRAPID predictions of the 10 selected RNA fragments with the catalog of human RBPs. Fragments belonging to the first and last 1.5 Kbp of SARS-CoV-2 genome are colored in different shades of pink and green, respectively. Error bars for each fragment correspond to the average value ± SE. The gray dashed line indicates the trend of the catRAPID score in the chart. (C) Description of the technique RaPID-MS, in which the RNA fragment of interest is expressed in cells flanked by BoxB stem loops. BoxB is specifically recognized by the co-transfected λN peptide fused to the biotin ligase BASU. Upon biotin addition to the growth medium, BASU biotinylates the host protein interactors attracted by the RNA of interest.

Figure 2

Defining SARS-CoV-2 interaction network with the human proteome (A) Number of specific interactors identified for each SARS-CoV-2 RNA fragment (labeled with the letter “F” follow by a number). (B) Enrichment distribution of the specific interactors for all considered fragments compared with the control “Scramble.” Data are normalized for protein length and abundance, as specified in the materials and methods. (C) The network of proteins identified by RaPID-MS. Only proteins significantly enriched over the control “Scramble” are displayed. Data are derived from the analysis of three independent biological replicates. In pink are displayed the RNA fragments, while in green the retrieved interactors. Proteins circled in green are RBPs, according to the RBPome database (https://rbpbase.shiny.embl.de/).

Figure 3

catRAPID performances on the RaPID-RBP datasets (A) catRAPID performances on the RaPID-RBP dataset. The predictive power of the method is calculated for different percentages of the dataset. catRAPID performance is calculated on LFQ experimental value, normalized by taking into account the abundance of the proteins in PAXdb and protein length in Uniprot. (B) catRAPID performance on the RaPID-RBP dataset, focusing on the single RNA fragments. For each fragment, the area under the curve at the different percentage of the dataset is shown. The performances are evaluated as in (A) and the fragments are ordered according to the respective genomic position. (C) (Left) Scatter chart of the analyzed RBPs ranked according to their relative log₂catRAPID score. The dashed lines indicate the 85th percentile. (Right) Boxplot representation of the whole distribution divided in the indicated percentiles. (D) Scatter chart of the analyzed RBPs ranked according to their relative log2 catRAPID score, displaying only proteins present in the 85th percentile of the distribution.

Figure 4

Nanocompore analysis identified significant k-mers in the analyzed cell lines (A) UCSC Genome Browser (http://genome.ucsc.edu) annotation of the modified k-mers containing a uridine found in SARS-CoV-2 infected CaCo-2, Calu-3, and Vero E6 cells and distributed over 14 different reference sgRNAs. From the top to the bottom, IVT and RaPID fragments, SARS-CoV-2 NRCeq assembly and RefSeq SARS-CoV-2 ORFs are present as a reference. (B) Graphical representation of nucleotide ranges shared between at least two cell lines (red circles). Here, sites have been grouped per ORFs encoded by each reference sgRNA of the assembly and the sequence of those carrying an UGUAR motif is displayed.

Figure 5

PUS7 modifies the SL2 of SARS-CoV-2 TRS-L favoring the binding of NSP1 (A) The secondary structure of SARS-CoV-2 TRS-L with the Nanocompore p value (NS6 sgRNA vs. IVT, GMM-logit test) overlaid as a color scale. The color represents the lowest p value among those of the five overlapping k-mers. Only k-mers with a p value ≤0.01 are colored according to the relative scale. (B) (Left) UV-RIP of HA-NSP1 co-expressed in HEK293 cells together with BoxB-Frag.1 expressing vector and either APEX2-FLAG, APEX2-FLAG-PUS7, or APEX2-FLAG-PUS7 D294A. The RNA bound to HA-NSP1 was eluted, retrotranscribed into cDNA and analyzed by RT-qPCR analysis using specific primers for RPLP0 and BoxB-Frag.1. The data shown are the average of four biological replicates (n = 4) and are expressed as relative enrichment over each respective input. Statistical analysis was performed using non-parametric, two-tailed t-test. ∗p value < 0.05. (Right) Representative image of the four WB analysis conducted on an aliquot of the UV-RIP input material (10%), showing the expression levels of the exogenous protein transiently co-expressed in HEK293 cells.