Modulation of UPF1 catalytic activity upon interaction of SARS-CoV-2 Nucleocapsid protein with factors involved in nonsense mediated-mRNA decay

Abstract

The RNA genome of the SARS-CoV-2 virus encodes for four structural proteins, 16 non-structural proteins and nine putative accessory factors. A high throughput analysis of interactions between human and SARS-CoV-2 proteins identified multiple interactions of the structural Nucleocapsid (N) protein with RNA processing factors. The N-protein, which is responsible for packaging of the viral genomic RNA was found to interact with two RNA helicases, UPF1 and MOV10 that are involved in nonsense-mediated mRNA decay (NMD). Using a combination of biochemical and biophysical methods, we investigated the interaction of the SARS-CoV-2 N-protein with NMD factors at a molecular level. Our studies led us to identify the core NMD factor, UPF2, as an interactor of N. The viral N-protein engages UPF2 in multipartite interactions and can negate the stimulatory effect of UPF2 on UPF1 catalytic activity. N also inhibits UPF1 ATPase and unwinding activities by competing in binding to the RNA substrate. We further investigate the functional implications of inhibition of UPF1 catalytic activity by N in mammalian cells. The interplay of SARS-CoV-2 N with human UPF1 and UPF2 does not affect decay of host cell NMD targets but might play a role in stabilizing the viral RNA genome.

Graphical Abstract

Figure 1.

The SARS-CoV-2 Nucleocapsid (N) protein directly interacts with the core NMD factor UPF2. (A) Schematic representation of the domain organization of human UPF1, UPF2 and SARS-CoV-2 N, and the constructs used in this study. Structured domains are shown as filled rectangles while intrinsically disordered regions (IDRs) are indicated by lines. (B) GST-pulldown assay of UPF1, UPF2 and MOV10_fus (a fusion of the N-terminus of MOV10 with the helicase core of UPF1) with GST-N as a bait. GST was used as a negative control in all such assays. The asterisk (*) indicates a contaminant. The top and bottom panels depict the inputs and precipitates, respectively, in this and all other GST-pulldown experiments. GST-N binds UPF2 but not UPF1 or MOV10_fus. (C) GST-pulldown assay to determine the domain of UPF2 that interacts with N. The MIF43G domain is the primary binding site for N, with additional weak interactions mediated by the UPF1-binding domain, U1BD. (D) GST-pulldown assay to identify the UPF2-binding region of N. Two IDRs (IDR1 and the inter-domain linker) and the dimerization domain (DD) of N make up a composite binding site for UPF2. No single site on N can mediate a strong interaction with UPF2 but a combination of any two binding sites restores binding comparable to that of full-length N (see also Supplementary Figure S1C). A complete gel including negative controls with GST is shown in Supplementary Figure S1D.

Figure 2.

The MIF4G3 domain of UPF2 stably associates with an N variant comprising two of the three identified binding motifs. (A) Isothermal titration calorimetry (ITC) experiments of UPF2_S with N_IDR1-core (left panel) and N_core (right panel). The dissociation constant (K_D) calculated from the binding isotherm is shown wherever applicable. Deletion of IDR1 in addition to the inter-domain linker abrogates binding of N to UPF2. (B) ITC experiments of N_IDR1-core with the MIF4G3 domain (left panel) and the U1BD (right panel) of UPF2. Removal of the U1BD does not impact binding of UPF2 to N (compare with left panel of Figure 2A), consistent with the observation that the U1BD shows no appreciable affinity for N. All N variants are considered as a dimer in this and subsequent experiments (see Supplementary figure S3A).

Figure 3.

Indirect association of N with UPF1 inhibits its catalytic activity. (A) Analytical size-exclusion chromatography (SEC) depicting formation of a ternary complex of UPF1, UPF2_S and N_IDR1-core (top panel). The terms Abs and V_r refer to absorbance (at 280 or 260 nm, as indicated in the figure) and retention volume, respectively, in this and all other figures. The exclusion volume of the column is 0.8 ml. SDS-PAGE analysis of the peak fractions visualized by Coomassie-staining, and a quantitative comparison of the relative amounts of UPF1 (normalized with respect to UPF1 in the UPF1-UPF2 complex) are shown in the middle and bottom panels, respectively. Interaction of N with the UPF1-UPF2 complex is sub-stoichiometric and leads to release of a small amount of UPF1 from the complex. (B) Fluorescence-based nucleic acid-unwinding activity measurements of a mixture of UPF1 and UPF2_S in presence of increasing concentrations of N (0.25- to 4-fold excess over UPF1). The data and the associated error bars represent the mean and standard deviation of three independent experiments. Technical duplicates were performed for each experiment. Controls without any protein and with N alone were included to monitor the stability of the RNA:DNA hybrid over the course of the experiment. The unwinding activity of UPF1 alone is shown for comparison. A low amount of N is sufficient to initiate inhibition of UPF1 unwinding activity, which is completely blocked in presence of high amounts of N. (C) RNA-dependent ATPase activity of a mixture of UPF1 and UPF2_S in the presence of low (0.25-fold of UPF1) and high (2-fold excess over UPF1) concentrations of N. An enzyme-coupled phosphate detection assay was used to measure ATPase activity. The data and the associated error bars represent the mean and standard deviation of three independent experiments. Technical duplicates were performed for each experiment. As with the unwinding activity measurements, a low amount of N slightly inhibits UPF1 ATPase activity while a higher amount shows more robust inhibition. (D) RNA-dependent ATPase activity of UPF1 in presence of different variants of UPF2 (added in 2-fold excess over UPF1). Although a UPF2 variant comprising the MIF4G3 and the U1BD domains strongly stimulates the ATPase activity of UPF1 (UPF2_S, dark blue trace), these domains are not capable of activating UPF1 on their own. Experimental setup and data presentation are as described above in (C).

Figure 4.

High concentrations of N impede UPF1 catalytic activity by displacing it from RNA. (A) Nucleic acid-unwinding activity of UPF1 in presence of increasing concentrations of N (0.25- to 4-fold excess over UPF1). The experiment, including controls, was conducted as described in Figure 3B. The unwinding activity of a mixture of UPF1 and UPF2_S is shown for comparison. In absence of UPF2, N inhibits UPF1 activity only at high concentrations. (B) RNA-dependent ATPase activity of UPF1 in presence of low (0.25-fold) and high (2-fold excess) concentrations of N. As observed with unwinding activity measurements, inhibition is only achieved upon addition of high amounts of N. (C) Analytical SEC experiments of UPF1 and U₄₅ RNA in the presence of 2-fold excess of N_ΔL. An overlay of the chromatogram of the U₄₅-UPF1-N mixture (black) with those of U₄₅-N (green) and U₄₅-UPF1 (yellow) are shown in the top panel. The bottom panels show the corresponding SDS- and urea–PAGE analyses of peak fractions of the U₄₅-UPF1-N chromatogram. Proteins were visualized by Coomassie staining and RNA was detected by radiolabeling with ³²P followed by phosphorimaging. Addition of an excess of N to a UPF1–RNA mixture leads to higher occupancy of N on RNA (peaks 1 and 2) and a concomitant release of UPF1 from RNA (peak 3). A quantitative comparison of the relative amounts of UPF1 released upon addition of equimolar and 2-fold excess of N is shown in Supplementary figure S4C. (D) Effect of addition of sub-stoichiometric (0.25×) and equimolar (1×) amounts of N_L-DD on the nucleic acid-unwinding activity of UPF1 in complex with UPF2_S. Reactions carried out in the presence of Nfl are shown for comparison. N_L-DD does not bind RNA but interacts with UPF2 and inhibits UPF1 in the UPF1–UPF2 complex. No inhibition is observed in the absence of UPF2 (see Supplementary Figure S4D).

Figure 5.

NMD is not substantially inhibited by N overexpression in human cell lines. (A) Experimental overview including schematic representations of the overexpressed FLAG-tagged proteins and beta-globin reporters. Expression of proteins and reporter was induced for 48 h with cumate, whereas treatment with different concentrations of SMG1i was performed for 24 h. (B) Probe-based quantitative RT-PCR analysis of GAS5 and ZFAS1 and SYBR-green-based analysis of globin expression levels in the respective overexpression and treatment condition normalized to the B2M reference. FLAG-mGold serves as negative control for protein expression and WT-globin as control reporter. Data points and means are plotted as log₂ fold change (log₂FC) (n = 3). Statistical analysis via Tukey's HSD was performed with adjusted p-value significance cutpoints 0.05 (*), 0.01 (**), 0.001 (***) and 0.0001 (****). (C) Overview of N overexpression RNA-Seq experiment, percentage of reads mapping to N, clustering of samples by PCA and differential expression analysis on gene- and transcript-level. (D) Overlap of significantly upregulated genes (cut-offs are indicated) in RNA-Seq data from four different CoV-2-N overexpression datasets and the union of four NMD-factor knockout/knockdown combinations. (E) Heatmap of DESeq2-derived log₂ fold changes of five selected known NMD target genes in various RNA-Seq datasets. The size of points corresponds to the statistical significance. The CoV-2-N-2xStrep dataset is obtained from: https://www.medrxiv.org/content/10.1101/2021.05.06.21256706v2. (F) Boxplots of differential transcript expression determined by Swish and stratified by GENCODE biotype (protein-coding and nonsense-mediated decay).

Figure 6.

A mechanistic model for inhibition of UPF1 catalytic activity on SARS-CoV-2 RNA by N-protein. The amount of N in host cells continuously increases with progression of infection. Despite robust inhibition of UPF1 catalytic activity by SARS-CoV-2 N, host cell NMD targets remained unaffected by expression of the N-protein. The NMD inhibitory effect of MHV N on viral RNA allows us to speculate a similar role for SARS-CoV-2 N in protecting its own genomic and sg-mRNA.