Author response: Ribosome profiling of porcine reproductive and respiratory syndrome virus reveals novel features of viral gene expression

2022

Georgia M. Cook, Katherine A. Brown, Pengcheng Shang, Yànhuá Lǐ, Lior Soday, Adam M. Dinan, Charlotte Tumescheit, A. Mockett, Yīng Fāng, Andrew E. Firth, Ian Brierley

Abstract

Article Figures and data Abstract Editor's evaluation eLife digest Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract The arterivirus porcine reproductive and respiratory syndrome virus (PRRSV) causes significant economic losses to the swine industry worldwide. Here we apply ribosome profiling (RiboSeq) and parallel RNA sequencing (RNASeq) to characterise the transcriptome and translatome of both species of PRRSV and to analyse the host response to infection. We calculated programmed ribosomal frameshift (PRF) efficiency at both sites on the viral genome. This revealed the nsp2 PRF site as the second known example where temporally regulated frameshifting occurs, with increasing −2 PRF efficiency likely facilitated by accumulation of the PRF-stimulatory viral protein, nsp1β. Surprisingly, we find that PRF efficiency at the canonical ORF1ab frameshift site also increases over time, in contradiction of the common assumption that RNA structure-directed frameshift sites operate at a fixed efficiency. This has potential implications for the numerous other viruses with canonical PRF sites. Furthermore, we discovered several highly translated additional viral ORFs, the translation of which may be facilitated by multiple novel viral transcripts. For example, we found a highly expressed 125-codon ORF overlapping nsp12, which is likely translated from novel subgenomic RNA transcripts that overlap the 3′ end of ORF1b. Similar transcripts were discovered for both PRRSV-1 and PRRSV-2, suggesting a potential conserved mechanism for temporally regulating expression of the 3′-proximal region of ORF1b. We also identified a highly translated, short upstream ORF in the 5′ UTR, the presence of which is highly conserved amongst PRRSV-2 isolates. These findings reveal hidden complexity in the gene expression programmes of these important nidoviruses. Editor's evaluation The article presents a first example of a detailed quantitative study of host and PRRSV gene expression over the time course of infection. The study not only identifies multiple non-canonical mechanisms of PRRSV gene expression regulation, but also shows that the frameshifting efficiency at the canonical ORF1ab frameshifting site changes with time. This finding provides new insights into the viral gene expression and into the regulation of programmed ribosome frameshifting, which has important implications for understanding viral biology and for developing antiviral drugs. https://doi.org/10.7554/eLife.75668.sa0 Decision letter Reviews on Sciety eLife's review process eLife digest Viruses have tiny genomes. Rather than carry all the genetic information they need, they rely on the cells they infect. This makes the few genes they do have all the more important. Many viruses store their genes not in DNA, but in a related molecule called RNA. When the virus infects cells, it uses the cells’ ribosomes – the machines in the cells that make proteins – to build its own proteins. One of the central ideas in biology is that one molecule of RNA carries the instructions for just one type of protein. But many viruses break this rule. The ribosomes in cells read RNA instructions in blocks of three: three RNA letters correspond to one protein building block. But certain sequences in the RNA of viruses act as hidden signals that affect how ribosomes read these molecules. These signals make the ribosomes skip backward by one or two letters on the viral RNA, restarting part way through a three-letter block. Scientists call this a ‘frameshift’, and it is a bit like changing the positions of the spaces in a sentence. The virus causes these frameshifts using proteins or by folding its RNA into a knot-like structure. The frameshifts result in the production of different viral proteins over time. The porcine reproductive and respiratory syndrome virus (PRRSV) uses frameshifts to cause devastating disease in pigs. Besides the sequences in its RNA that allow the ribosomes to skip backwards, the viral enzyme that copies the RNA can also skip forward. This results in shortened copies of its genes, which also changes the proteins they produce. To find out exactly how PRRSV uses these frameshifting techniques, Cook et al. examined infected cells in the laboratory. They monitored the RNA made by the virus and looked closely at the way the cells read it using a technique called ribosome profiling. This revealed that frameshifting increases over the course of an infection. This is partly because the viral protein that causes frameshifts builds up as infection progresses, but it also happened with frameshifts caused by RNA knots. The reason for this is less clear. Cook et al. also discovered several new RNAs made later in infection, which could also change the proteins the virus makes. RNA viruses cause disease in humans as well as pigs. Examples include coronaviruses and HIV. Many of these also have frameshift sites in their genomes. A better understanding of how frameshifts change during infection may aid drug development. Future work could help researchers to understand which proteins viruses make at which stage of infection. This could lead to new treatments for viruses like PRRSV. Introduction Porcine reproductive and respiratory syndrome virus (PRRSV) is an enveloped, positive-sense, single-stranded RNA virus in the family Arteriviridae (order: Nidovirales) (Meulenberg et al., 1994; Cavanagh, 1997), and the aetiological agent of the disease from which it takes its name: porcine reproductive and respiratory syndrome (PRRS). Attempts to control PRRS by vaccination have had limited success (Nan et al., 2017), and it remains one of the most economically devastating diseases of swine, causing reproductive failure in adult sows and respiratory failure in young pigs, at an estimated cost of $664 million a year in the US alone (Holtkamp et al., 2013; Kappes and Faaberg, 2015). The two lineages of PRRSV, formerly known as ‘European’ (Type 1) and ‘North American’ (Type 2) PRRSV, share just ~60% pairwise nucleotide similarity and were recently reclassified as two separate species, Betaarterivirus suid 1 and 2 (viruses named PRRSV-1 and PRRSV-2) (Kappes and Faaberg, 2015; Collins et al., 1992; Wensvoort et al., 1991). For ease of reference, PRRSV-1 is herein referred to as EU (European) and PRRSV-2 as NA (North American) PRRSV, although both lineages are observed worldwide (Guo et al., 2018a). Despite the substantial genetic and antigenic diversity between the two species, the overall clinical symptoms are similar, although there is also considerable (~20%) genetic diversity within each species, rendering this isolate-dependent (Nan et al., 2017; Kappes and Faaberg, 2015). This is largely due to PRRSV’s rapid mutation rate, which leads to relatively frequent emergence of highly pathogenic strains capable of escaping existing immunity, particularly within the NA PRRSV species (Nan et al., 2017; Kappes and Faaberg, 2015). The PRRSV genome (14.9–15.5 kb; Figure 1A) is 5′-capped, 3′-polyadenylated, and directly translated following release into the cytoplasm (Snijder et al., 2013). Like most members of the order Nidovirales, PRRSV replication includes the production of a nested set of subgenomic (sg) RNAs by discontinuous transcription, where the viral RNA-dependent RNA polymerase (RdRp) jumps between similar sequences in the 3′-proximal region of the genome and the 5′ UTR, known as body and leader transcription regulatory sequences (TRSs), respectively (Kappes and Faaberg, 2015; Posthuma et al., 2017). These sgRNAs are 5′- and 3′-co-terminal and are translated to express the structural proteins encoded towards the 3′ end of the genome (Kappes and Faaberg, 2015; Posthuma et al., 2017). The 5′-proximal two-thirds of the genome contains two long ORFs, ORF1a and ORF1b, with a −1 programmed ribosomal frameshift (PRF) site present at the overlap of the two ORFs (Meulenberg et al., 1993; Nelsen et al., 1999). Ribosomes that frameshift at this site synthesise polyprotein (pp)1ab, while the remainder synthesise pp1a, both of which are cleaved by viral proteases into several non-structural proteins (nsps) (Kappes and Faaberg, 2015; Snijder et al., 1994). The proteins encoded by ORF1b include the RdRp and the helicase, and frameshifting at this site is thought to set the stoichiometry of these proteins relative to those encoded by ORF1a, a prevalent expression strategy in the Nidovirales order (Gorbalenya et al., 2006). Figure 1 with 6 supplements see all Download asset Open asset An overview of the experimental set-up and the quality of the datasets. (A) Genome map of the North American porcine reproductive and respiratory syndrome virus (NA PRRSV) isolate used in this study (SD95-21, GenBank accession KC469618.1). ORFs are coloured and offset on the y-axis according to their frame relative to ORF1a (0: purple, no offset; +1/–2: blue, above axis; +2/–1: yellow, below axis). Subgenomic (sg) RNAs are shown beneath the full-length genomic RNA, with the region of 5′ UTR that is identical to the genomic 5′ UTR shown in grey (known as the ‘leader’). ‘FS’, frameshift. ORFs translated from each sgRNA are depicted as coloured boxes and named to the right. The nucleotide sequence at the non-structural protein (nsp)2 programmed ribosomal frameshift (PRF) site of the NA PRRSVs used in this study is shown (boxed), with mutations made to disrupt PRF and/or expression of nsp2TF in the KO2 mutant virus highlighted in blue. All mutations are synonymous with respect to the ORF1a amino acid sequence. (B) Key experimental steps in preparation of RiboSeq libraries (left) and parallel RNASeq libraries (right). Schematics of ribosomes protecting classical length ribosome-protected fragments (RPFs) (A site occupied) and short RPFs (A site unoccupied) are shown to the left, with numbers within the decoding centre indicating nucleotide positions within codons. (C) Length distribution of positive-sense RiboSeq (left) and RNASeq (right) reads mapping within host (grey) or viral (green, mock excluded) coding sequences (CDSs) in each library. For 9 hr post-infection (hpi) replicate 1 samples (RiboSeq and RNASeq), fragments of 25–34 nt were size-selected during the library preparation; for all other samples, the minimum length selected was 19 nt for RiboSeq and ~45 nt for RNASeq. Note that the RiboSeq library 9 hpi mock replicate 3 was discarded due to poor quality. Further quality control analyses can be found in Figure 1—figure supplement 1–Figure 1—figure supplement 6. Canonical −1 PRF signals are characterised by two main features, a heptanucleotide ‘slippery’ sequence (SS) which permits re-pairing of the codon:anticodon duplex in the new reading frame, separated by a 5–9 nucleotide (nt) spacer from a downstream RNA structure, often a pseudoknot. This is thought to present a ‘roadblock’ which impedes ribosome processivity over the slippery sequence and stimulates frameshifting (Rodnina et al., 2020; Atkins et al., 2016; Firth and Brierley, 2012; Plant and Dinman, 2005; Namy et al., 2006; Caliskan et al., 2014). In the PRRSV genome, the ORF1ab frameshift signal comprises a U_UUA_AAC slippery sequence (where underscores delineate codons in the 0 frame) and a pseudoknot beginning 5 nt downstream (Meulenberg et al., 1993; Nelsen et al., 1999). The efficiency of −1 PRF at the PRRSV ORF1ab site has not been measured in the context of infection, but is thought to be around 15–20% based on assays using reporter constructs (den Boon et al., 1991; Bekaert and Rousset, 2005). Recently, the region of the PRRSV genome encoding nsp2 was found to contain a second PRF signal (Figure 1A, inset, WT), conserved in all known arteriviruses except equine arteritis virus (EAV) and wobbly possum disease virus (WPDV) (Fang et al., 2012; Li et al., 2014; Napthine et al., 2016; Li et al., 2019). This PRF signal is unusual in that it stimulates both −1 and −2 PRF, enabling production of three variants of nsp2 and rendering it the first example of efficient −2 PRF in a eukaryotic system (Fang et al., 2012; Li et al., 2014). These three proteins share the N-terminal two-thirds of nsp2 (the 0-frame product), which encodes a papain-like protease (PLP) 2 domain – an ovarian tumour domain (OTU) superfamily protease with deubiquitinase (DUB) and deISGylase activity (Han et al., 2009; van Kasteren et al., 2012; Frias-Staheli et al., 2007; Sun et al., 2010; Sun et al., 2012; Li et al., 2018). This has an immune antagonistic effect, and interferon (IFN)-β signalling inhibition has been demonstrated for all three variants of nsp2, most strongly for the frameshift products (Li et al., 2018). After the PRF site, nsp2 contains a multi-spanning transmembrane (TM) domain, thought to promote formation of double-membrane vesicles (DMVs) in the peri-nuclear region and anchor nsp2 to these membranes (Kappes et al., 2015; Snijder et al., 2001; Knoops et al., 2012). Ribosomes which undergo −2 PRF at this site translate 169 codons in the –2 frame to produce nsp2TF. This contains an alternative putative multi-spanning TM domain, thought to be responsible for targeting nsp2TF to the exocytic pathway, where it deubiquitinates the PRRSV structural proteins GP5 and M, preventing their degradation (Fang et al., 2012; Guo et al., 2021). Nsp2N, the product of −1 PRF, is a truncated form of nsp2, which is generated following termination of translation at a −1-frame stop codon immediately downstream of the slippery sequence, and is predicted to be cytosolic (Fang et al., 2012; Li et al., 2014). A second unique feature of the nsp2 PRF site is its non-canonical nature. Rather than an RNA secondary structure, the stimulatory element is a complex of a cellular protein, poly(rC) binding protein (PCBP), and the viral protein nsp1β, bound at a C-rich motif (CCCANCUCC) located 10 nt downstream of the slippery sequence (G_GUU_UUU) (Fang et al., 2012; Li et al., 2014; Napthine et al., 2016; Li et al., 2019). How binding of this motif by the protein complex stimulates PRF is uncertain, but it may act as a roadblock analogous to the RNA structures of canonical PRF (Li et al., 2014; Napthine et al., 2016; Li et al., 2019; Patel et al., 2020). In contrast to RNA structure-directed PRF sites, which are commonly assumed to operate at a fixed efficiency, the trans-acting mechanism of PRF stimulation at the nsp2 site presents a potential mechanism for temporal regulation, as observed for cardioviruses – the only other known example of protein-stimulated PRF (Loughran et al., 2011; Finch et al., 2015; Napthine et al., 2017; Napthine et al., 2019; Hill et al., 2021b; Hill et al., 2021a). Frameshift efficiency in EU PRRSV-infected MARC-145 cells at 24 hpi was calculated as 20% for −2 PRF and 7% for −1 PRF (Fang et al., 2012); however, this has not been measured over a timecourse of infection. In recent years, both low- and high-throughput studies of nidoviruses have highlighted considerably greater complexity in both the transcriptome and translatome than is captured solely by the canonical transcripts and ORFs (Nelsen et al., 1999; Yuan et al., 2000; Di et al., 2017; Kim et al., 2020; Wang et al., 2021; Stewart et al., 2018; Irigoyen et al., 2016; Dinan et al., 2019; Finkel et al., 2021b; Zhang et al., 2021b). Here, we use ribosome profiling (RiboSeq), a deep-sequencing-based technique which generates a global snapshot of ongoing translation (Ingolia et al., 2009), in parallel with RNASeq, to probe viral and host gene expression over a timecourse of PRRSV infection. Host differential gene expression analysis revealed that many of the transcriptional changes upon infection were counteracted by reductions in translation efficiency (TE), indicating a dampened host response, and highlighting the importance of looking beyond transcription when analysing gene expression. On the viral genome, our studies reveal, for the first time, a significant increase in frameshift efficiency over the course of infection at the nsp2 −2 PRF site, highlighting arteriviruses as the second example of temporally regulated frameshifting during infection. In addition, we identify several novel viral ORFs, including a highly expressed upstream ORF (uORF), the presence of which is conserved amongst NA PRRSV isolates. In both species of PRRSV, related non-canonical sgRNAs overlapping ORF1b were identified and characterised. These likely facilitate the expression of several of the novel ORFs which overlap ORF1b, and the observation of increased ribosome density in the 3′-proximal region of ORF1b suggests they may also function to temporally regulate expression of the C-terminal region of ORF1b itself. This first application of RiboSeq to an arterivirus uncovers hidden layers of complexity in PRRSV gene expression that have implications for other important viruses. Results Experimental set-up PRRSV gene expression was investigated using three viruses: an EU PRRSV isolate based on the Porcilis vaccine strain (MSD Animal Health; GenBank accession OK635576.1), NA PRRSV SD95-21 (GenBank accession KC469618.1), and a previously characterised mutant variant (NA PRRSV SD95-21 KO2) which bears silent mutations in the nsp2 PRF site slippery sequence and C-rich motif rendering it unable to bind PCBP, induce −1 or −2 PRF, or produce nsp2N or nsp2TF (Figure 1A, inset) (Fang et al., 2012; Li et al., 2014; Napthine et al., 2016; Li et al., 2018). MA-104 cells (Chlorocebus sabaeus) were infected with EU PRRSV at a multiplicity of infection (MOI) of ~1–3 and harvested at 8 hr post-infection (hpi) following a 2 min pre-treatment with the translation elongation inhibitor, cycloheximide (CHX). MARC-145 cells (a cell line derived from MA-104) were infected with NA PRRSV (WT or KO2 mutant) at MOI 5 or mock-infected and harvested at 3, 6, 9, or 12 hpi by flash-freezing without CHX pre-treatment. Cell lysates were used for ribosome profiling, in which RNase I was added to digest unprotected regions of RNA and ribosomes were purified to isolate ribosome-protected fragments (RPFs) of RNA (Figure 1B), which indicate the positions of ribosomes at the time of harvesting. In parallel, aliquots of the same lysates were subjected to alkaline hydrolysis to generate fragments of RNA for RNASeq. Amplicons were prepared, deep sequenced, and reads aligned to host (C. sabaeus) and viral genomes (Supplementary file 1) to characterise the transcriptome and translatome of infected cells. Data quality analysis Quality control analyses were performed as described previously (Irigoyen et al., 2016; Figure 1C, Figure 1—figure supplements 1–6), revealing that the overall data quality is The length distribution of coding sequence RPFs is observed to at nt (where fragments of this length were and at with RPFs of these thought to ribosomes with an A site or an A site by (Figure and Figure 1—figure supplement et al., 2009; and et al., 2019; et al., 2014). the of RPFs is in the NA PRRSV-infected libraries than mock libraries at and 12 hpi see Materials and In this has been to of eukaryotic elongation to inhibition of translation elongation et al., which suggests a similar regulatory response may be by the of PRRSV infection. The between the 5′ end of an and the site of the ribosome is 12 nt in these (Figure 1—figure supplements 1 and in RiboSeq reads known as with the of 5′ mapping to the first within the known as 0 (Figure 1—figure supplement Figure 1—figure supplement Figure 1—figure supplement and with the observed length distribution (Figure 1C, Figure 1—figure supplement this that a of these reads are In the length and 5′ end of RNASeq reads is by alkaline hydrolysis and to a length distribution (Figure 1C, Figure 1—figure supplement and of a (Figure 1—figure supplement Figure 1—figure supplement Figure 1—figure supplement and reads a similar to host reads (Figure 1C, Figure 1—figure supplement Figure 1—figure supplement and with the of 3 hpi NA PRRSV RiboSeq in which the of in the to be relative to the of likely due to the of viral translation at this These libraries are from all analyses except those in Figure and Figure supplement where they an The of the length distribution and of reads to reads in NA PRRSV RiboSeq libraries at (Figure 1C, Figure 1—figure supplement 2) suggests that a of viral reads from as from RNase I by viral complex This of the library referred to as although it could from several is reads mapping to the ORF1b region of the viral genome (Figure 1—figure supplements 3 and where the read from translation is RiboSeq read for which a of reads map to 0 were to be likely to have a of (Figure 1—figure supplement and were selected for all NA PRRSV RiboSeq analyses is not a for RNASeq libraries proteins are RNA and it not affect the EU PRRSV RiboSeq RPFs mapping to the host transcriptome (Figure 1—figure supplements 5 and we that these have a of RiboSeq reads and where is in translated regions of the viral genome its likely be by of read Figure 2 with 3 supplements see all Download asset Open asset An overview of viral transcription and translation over a timecourse of infection. (A) Genome map of North American porcine reproductive and respiratory syndrome virus (NA from Figure (B) RNASeq read in reads million reads on the viral genome, application of a from cells harvested over a timecourse of hr post-infection reads are in the in the axis). The libraries with the RiboSeq quality control results were selected for this hpi replicate 6 hpi replicate 9 hpi replicate 12 hpi replicate with and KO2 libraries shown in Figure supplement (C) RiboSeq read on the viral genome from the libraries to were separated according to (0: and application of a Further and KO2 libraries are shown in Figure supplement of reads to host positive-sense reads read in the were into the following positive-sense RNASeq RNASeq and positive-sense RiboSeq of RiboSeq reads can be found in Figure supplement The line the for each for NA PRRSV (WT and KO2 with for and KO2 offset to the and EU PRRSV are as by in the The RiboSeq 3 hpi is to the of the NA PRRSV at this likely by the relatively of reads in these Data from on a Here, data from and KO2 are and are as of the density of subgenomic reads to All read were and were calculated as reads million reads of reads from each in RiboSeq 3 hpi libraries were and RNASeq was from the at 3 hpi due to the of reads for of the and as in with a grey line indicating a of Figure 3 Download asset Open asset and translation of the porcine reproductive and respiratory syndrome virus PRRSV) genome and of non-structural protein (A) Genome map of the EU PRRSV strain used in this study (GenBank accession Genome map as in Figure 1A, with subgenomic RNAs for (B) RNASeq read on the EU PRRSV genome. as in Figure (C) RiboSeq read on the EU PRRSV genome. as in Figure except for the of read to include – in this read were selected for in Figure 1—figure supplement of lysates used for North American PRRSV ribosome profiling 1 with to viral protein and cellular protein as a M, of the from to the of relative to of expression in MA-104 cells infected with EU PRRSV, harvested over a hr Figure data 1 of in Figure Download Figure data 2 of in Figure with Download Figure data 3 of in in Figure and results in Figure Download Figure data of in Figure Download Figure data 5 of in Figure with Download transcription and translation over a timecourse of infection data we on to analyse virus replication over the timecourse by RNASeq and RiboSeq read at each on the viral genome 2 and 3, Figure supplements 1 and RNASeq revealed a of PRRSV replication and transcription, with read at 3 likely to for genome replication at 6 with the of and of at later (Figure Figure supplement The observed of virus translation was also with (Figure Figure supplement 3 hpi not a of RPFs were observed indicating that translation of the NA PRRSV genome is just beginning to the by RiboSeq these 6 translation of ORF1ab is and comprises the of viral translation (Figure Figure supplement density with the of significant sgRNA production at this 9 translation of sgRNAs the and viral translation a of ongoing translation in the cell (Figure Figure supplement with viral expression at 9 hpi is by (Figure and and other studies have shown expression of viral proteins and viral RNA replication at this (Li et al., 2012; and to between 9 and 12 although accumulation of the to a at both production and translation of sgRNAs are highly over (Figure Figure

Abstract

Related Papers