Learn more: PMC Disclaimer | PMC Copyright Notice
Metatranscriptomics analysis reveals a novel transcriptional and translational landscape during Middle East respiratory syndrome coronavirus infection
Summary
Among all RNA viruses, coronavirus RNA transcription is the most complex and involves a process termed “discontinuous transcription” that results in the production of a set of 3′-nested, co-terminal genomic and subgenomic RNAs during infection. While the expression of the classic canonical set of subgenomic RNAs depends on the recognition of a 6- to 7-nt transcription regulatory core sequence (TRS), here, we use deep sequence and metagenomics analysis strategies and show that the coronavirus transcriptome is even more vast and more complex than previously appreciated and involves the production of leader-containing transcripts that have canonical and noncanonical leader-body junctions. Moreover, by ribosome protection and proteomics analyses, we show that both positive- and negative-sense transcripts are translationally active. The data support the hypothesis that the coronavirus proteome is much vaster than previously noted in the literature.
Graphical abstract
Highlights
- •MERS-CoV produces a complicated network of noncanonical sgmRNAs during replication
- •Minus-strand translation activity evidence of the MERS-CoV genome and transcripts
- •Identification of a novel open reading frame within the nucleocapsid coding region
Medical microbiology; Virology; Transcriptomics
Introduction
Middle East Respiratory Coronavirus (MERS-CoV) infection is associated with high morbidity and mortality in the Middle East and Sub-Saharan Africa.1 The outbreak is ongoing and is associated with both symptomatic and asymptomatic infections in humans.2 As seen with the emergence of SARS-CoV-2 in late 2019 and the subsequent pandemic, coronaviruses (CoVs) pose a significant threat to global health. However, the molecular mechanisms governing emerging coronavirus pathogenesis are unknown but likely involve novel genetic factors that contribute to efficient replication and pathogenesis. During their replication cycles, CoVs reproduce their genomes and generate 3′-nested sets of subgenomic RNAs (sgmRNAs) that serve as mRNAs for the translation of downstream open reading frames (ORFs). The production of these various RNA species is mediated by a complex discontinuous transcription program that involves negative-sense, double-stranded intermediate RNAs and the acquisition of a common ∼70-nt leader sequence from the 5′ end of the input RNA genome.3,4,5,6,7,8 The characterization and quantification of these RNA species, many of which are present in comparatively low abundance in infected cells, have historically been challenging, as positive-sense RNAs typically overwhelm traditional detection methodologies.6 This limitation in data analysis has plagued researchers in many positive-sense RNA virus fields.
However, the advent of deep sequencing and other metagenomics technologies has vastly increased our capacity to detect low-abundance RNA species and minute changes in RNA production at the cellular level, which has enhanced studies analyzing viral RNA species in the context of infection. Such studies were previously limited to biochemical and radioactive detection methodologies, which were often insufficient for detecting the presence of low-abundance viral RNAs and minor changes in transcription that might occur in studies of viral mutants with similar replication phenotypes. These limitations have been magnified in studies of viruses with complex replication cycles, such as CoVs and other nidoviruses.3,6,9,10,11
The replication and transcription of CoV RNA species involves a complicated process termed “discontinuous transcription”3,4,10,12,13 (Figure S1). In this process, RNA is transcribed by the viral RNA-dependent RNA polymerase (RdRp) from the input genome from the 3′ end, during which it encounters conserved regulatory sites known as transcription regulatory sequences (TRSs), each of which contains 5′ and 3′ flanking sequences surrounding a 6- to 7-nt core sequence (CS) that serves as the RNA signal for the polymerase to slow down and then either read through to the next TRS or to discontinuously acquire the 5’ ∼70-nt leader sequence through a mechanism that likely involves a combination of RNA looping via secondary structure interactions and protein-protein and protein-RNA interactions.7,14,15,16,17,18,19 This process produces a series of negative-sense, minus-strand RNAs that are both genome-length and sgmRNA-length, and transcription then continues to produce positive-sense, plus-strand genome RNA and sgmRNAs that are 3′ nested and, by virtue of the discontinuous acquisition of the 5′ leader, also 5′ co-terminal.4,5,18,20
Conventional nidovirus dogma holds that the strict conservation of the TRS CS limits the production of sgmRNAs to those with perfect or near-perfect sequence matches at the leader-body junction. However, sequence analysis of several CoV species indicates the presence of multiple additional, near-match TRS CSs. Furthermore, our laboratory has shown that the core TRSs of severe acute respiratory syndrome CoV (SARS-CoV) can be both conservatively and nonconservatively replaced without destroying viral infectivity, indicating that the virus’s replication machinery is very tolerant to both minute and global changes in the TRS CS than previously thought.21,22 Additionally, nonconservative replacement of the network of TRSs (i.e., the transcription regulatory network—TRN) in SARS-CoV resulted in the detection of multiple additional RNA species in addition to the canonical species via Northern blot, demonstrating that the viral RdRp and the replication machinery were capable of recognizing alternate TRS CSs.22 In parallel, Bentley et al. demonstrated that the gammacoronavirus infectious bronchitis virus utilizes a noncanonical TRS to transcribe an RNA that serves as the template for a novel accessory protein,23,24 while Stewart et al. used RNAseq and ribosome profiling to identify previously “hidden” ORFs in equine torovirus.25 Together, these studies hypothesize that the CoV transcriptional machinery may be programmed to produce multiple additional “cryptic” sgmRNA species that utilize noncanonical, near-match TRS CSs, but that their relative low abundance in comparison to the canonical sgmRNA set has rendered them historically difficult to detect. This hypothesis was supported by data from multiple groups using a combination of direct RNA sequencing and DNA nanoball approaches, which defined numerous noncanonical subgenomic RNA species in both SARS-CoV-2 and HCoV-229E.11,26,27,28
Deep sequencing technology, with its read output in the hundreds of thousands to tens of millions and its capacity to filter reads based on different, user-customizable parameters, presents opportunities to sequence large virus genomes and to detect multiple types of viral RNA species within a single sample set, even low-abundance RNAs. Applications of this technology have allowed researchers to identify, characterize, and quantify the RNA species produced during viral infections, furthering our knowledge of how plus-strand RNA viruses maximize the use of their comparatively limited genome sizes.29 Moreover, proteomics, coupled with ribosome scanning techniques, provides a means for identifying low-abundance proteins during virus infection. In this study, we used deep sequencing approaches and MERS-CoV as a model virus to detect and characterize the full contingent of leader-containing RNA species produced during CoV infection (i.e., the viral transcriptome). We also began to explore whether low-abundance RNAs produced during CoV infection could be used as templates for the translation of previously uncharacterized viral proteins using mass spectrometry-based proteomics. Our data support the hypothesis that MERS-CoV, along with other CoVs, encodes a considerably more complex genome organization and expression network than represented in the literature, which has clear implications in viral emergence and disease potential.
Results
RNAseq analysis of MERS-CoV-infected cells indicates the production of canonical and noncanonical transcripts
To preliminarily evaluate the cumulative viral transcriptome produced during MERS-CoV infection, Vero cells were inoculated with MERS-CoV at an MOI of 5, and RNA was harvested at 16 h p.i. Following poly-A purification, RNA was then submitted for library preparation and Illumina RNAseq analysis. Reads were initially aligned to the full-length MERS-CoV genome reference sequence. As expected, poly-A-containing viral RNA reads mapped across the genome, with proportionally high concentrations of reads at the extreme 5′ end, reflecting full-length RNA, and in overall increasing relative concentrations starting from the location of the ORF2 TRS, with peaks at each additional canonical TRS (Figure 1A). However, the read mapping analysis also indicated the presence of viral RNA peak signals at locations outside of canonical transcript start sites, both within ORF1 and between the downstream canonical ORFs, suggesting the production of transcripts originating from noncanonical transcript start locations.
Illumina and non-strand-specific Nanopore sequence analysis of MERS-CoV-infected cells
Top: The MERS-CoV genome is depicted as a black line, with annotated proteins shown as blue bars. Transcription regulatory sequences (TRSs) are indicated with dashed lines in the genome schematic and in all panels.
(A) Total Illumina reads were aligned to the MERS-CoV genome reference sequence using the RNA aligner in Geneious Prime and were output as a BAM file. Reads were then sorted using Samtools Sort and arranged by coverage using Bedtools GenomeCoverage. Reads are shown plotted according to number of reads (y axis) versus genome position (x axis).
(B) Total Illumina reads were first aligned to a minimal leader sequence (MERS-CoV nts 40–60), and reads that aligned to the minimal sequence were then aligned to the MERS-CoV genome reference sequence. Reads were then sorted and plotted as in A.
(C) Leader-containing transcripts were amplified from RNA harvested from MERS-CoV-infected cells using the genome-spanning primer walking strategy shown in Figure S2A, and the transcripts were then sequenced using Oxford Nanopore amplicon sequencing chemistry. Transcripts were aligned to the genome using the RNA aligner in Geneious, with intron discovery turned on. Reads were output and graphed as described for A and B, and junctions are described in detail in Table S1. All RNAseq studies were performed in duplicate, with the most representative datasets shown.
To aid in the identification of potential noncanonical start sites, the complete library of RNAseq reads was first mapped to a minimal sequence upstream of the MERS-CoV leader (nts 40–65). Following this preliminary alignment, reads that mapped to this sequence were then re-mapped to the full-length MERS-CoV genome (Figure 1B). Interestingly, this alignment revealed not only high enrichment of canonical leader-containing reads at the 5′ end of the genome and at the 5′ ends of ORFs 2–8, but also the greatly enriched presence of additional leader-containing reads in the first ∼4000 nt of the genome and within the downstream ORFs, particularly ORFs 2, 4, 5, 7, and 8, as well as at the 3′ end of the genome (Table S1). These observations suggest the production of high levels of noncanonical leader-containing viral RNA during viral replication.
Leader-primed sequencing of MERS-CoV-infected cells identifies leader- and antileader-containing transcripts on the plus and minus strands
Our initial observations of the production of high levels of leader-containing transcripts at noncanonical locations during MERS-CoV infection were in the context of poly-A purification, which likely greatly biases the analysis for positive-sense RNA reads. Comparatively little is known about negative-sense CoV replication, especially in the context of the utilization of the TRS during discontinuous negative-sense transcription. Moreover, the CoV transcription strategy would predict that paired plus- and minus-strand sgmRNAs are indicative of active transcription complexes engaged in canonical and noncanonical sgmRNA synthesis. Therefore, to better explore the leader-containing CoV transcriptome, we designed an experiment in which leader-containing species across the genome would be produced for sequence analysis using Oxford Nanopore MinION sequencing (Figure S2A). In our initial experiment, first-strand cDNAs were generated using random primers, which would non-selectively produce cDNAs from both the plus and minus strands. The amplicon library was pooled and sequenced, and the results are shown in Figure 1C. Similar to the results of the Illumina sequencing analysis, reads sharply aligned with canonical TRS locations; additionally, reads also aligned with several regions showing Illumina-identified noncanonical TRS usage, suggesting that these noncanonical locations participated in leader-containing discontinuous transcription. Interestingly, and in much higher proportionate levels to what was seen in the Illumina data, there was evidence of several areas of leader-containing discontinuous transcription spanning the length of ORF1, similar to what has been reported for SARS-CoV-2 and 229E.26,28 However, because our amplification strategy utilized random primers in the first-strand cDNA step, it was not clear whether these regions were indicative of plus- or minus-strand discontinuous transcription.
Thus, to discriminate between plus- and minus-strand transcription events, we designed a strand-specific amplicon generation strategy using first-strand primers that would specifically prime either positive- or negative-sense first-strand cDNA (Figure S2B). Amplicons were then generated from these strand-specific first-strand cDNAs using the genome-spanning strategy described above, and products were sequenced via Oxford Nanopore MinION sequencing (Figures 2A and 2B). Interestingly, while both positive- and negative-sense strategies identified canonical TRS usage, the proportions and locations of noncanonical usage differed by sense, with higher proportions of negative-sense TRS loci identified past nt 3000 in ORF1, with particularly distinct peaks at nts 11338, 12362, and 15974.
Strand-specific leader-containing transcripts and leader-body junctions produced during MERS-CoV infection
Top: The MERS-CoV genome is depicted as a black line, with annotated proteins shown as blue bars. Transcription regulatory sequences (TRSs) are indicated with dashed lines in the genome schematic and in all panels.
(A) Strand-specific leader-containing transcripts were amplified from RNA harvested from MERS-CoV-infected cells using the first-strand cDNA production and genome-spanning primer walking strategy shown in Figure S2B, and the transcripts were then sequenced using Oxford Nanopore amplicon sequencing chemistry. Transcripts were aligned to the genome using the RNA aligner in Geneious, with intron discovery turned on. Reads were output and graphed as described for Figure 1C. A. Positive-sense reads.
(B) Negative-sense reads.
(C and D) Leader-body junction usage for plus and minus strands is shown graphed as strength of junction usage (y axis, expressed as percentages of total junctions identified from the novel junction finder algorithm in Geneious) versus genome position (x axis).
(C) Positive-sense junctions. Red: Canonical leader-body junction positions. Yellow: Near-canonical leader-body junction positions (within the TRS CS but not exactly at the canonical leader-body junction). Blue: Noncanonical leader-body junction positions. These junctions are described in detail in Table S2.
(D) Negative-sense junctions. Green: Canonical leader-body junction positions. Yellow: Near-canonical leader-body junction positions (within the TRS CS but not exactly at the canonical leader-body junction). Blue: Noncanonical leader-body junction positions. These junctions are described in detail in Table S3. All RNAseq studies were performed in duplicate, with the most representative datasets shown.
The use of canonical TRS loci has historically been validated by Sanger sequencing and Northern blot analysis. However, as indicated by the Nanopore sequencing data, the virus appears to avail itself of far more variations of noncanonical than canonical TRSs across the genome, albeit at comparatively low abundance levels. Deep sequence analysis adds the possibility of the use of highly sensitive sequence junction-finding algorithms, which, while originally developed to identify eukaryotic splice junctions, can also be employed to find junctions formed as a result of the discontinuous transcription strategy used by CoVs. Therefore, we analyzed the plus- and minus-strand read alignments from MERS-CoV infection for the use of novel junctions. As expected based on the Nanopore analysis data, both the plus- and minus-strand alignments yielded multiple supported novel junctions in addition to the canonical junctions (Figures 2C, 2D, and and33 and Tables S2 and S3). These findings clearly indicate that CoV discontinuous transcription yields multiple low-abundance RNA species across the genome on both the plus and minus strands in addition to the canonical leader-body subgenomic RNAs. Moreover, these novel plus- and minus-strand transcripts can be grouped into pools in which the minus strand appears to drive the transcription of the plus strand (i.e., junctions appear at corresponding locations: 71/121 minus-strand transcripts and 72/160 plus-strand transcripts), minus strands that have no plus-strand counterpart (i.e., unique junctions in the minus strand data: 50/121 minus-strand transcripts), and plus strands that appear to have been generated by a subsequent discontinuous step (i.e., plus strand junctions with no clear minus-strand counterpart 88/160 plus-strand transcripts), indicating that discontinuous transcription is not limited to transcription from genome-length RNA.
Chord diagrams of plus- and minus-sense leader-body junctions produced during MERS-CoV infection
Junctions shown in Figures 2C and 2D and detailed in Tables S2 and S3 are depicted in chord diagrams.
(A) Positive sense.
(B) Negative sense. The MERS-CoV genome is depicted as a multicolored ring. Strand-specific Nanopore reads are shown outside of the ring. Chords indicating the source and destination of leader-body junctions are shown within the ring, with widths of the chord lines corresponding to proportion of usage. The strongest junctions (see Figures 2C and 2D and Tables S2 and S3) are enumerated between the ring and the Nanopore reads. All features are color-coded according to the ring colors, which are colored according to genome regions: bright red: leader; burgundy: ORF1; purple: ORF2; navy blue: ORF3, dark teal: ORF4; dark green: ORF5; light green: ORF6; yellow: ORF7; orange: ORF8. Chords are color-coded based on the junction origin region.
This junction analysis informed our search for ORFs that could be translated from novel transcripts. Using an in silico approach, putative plus-sense RNAs with noncanonical leader-body junctions (with abundance levels > 0.05%) were constructed, and ORF analyses were performed (Figure 4). All RNAs yielded at least one ORF, some of which spanned the noncanonical leader-body junction. These results provided further support for the likelihood that low-abundance, novel proteins might be produced during the course of MERS-CoV infection.
Putative noncanonical ORFs in MERS-CoV infection
Junctions identified in the analyses described in Figures 2A and 2C and Table S2 were used to manually construct alternate leader-body transcripts, and an ORF-finding algorithm was then run on the novel transcripts. One ORF for each novel transcript is shown; most ORFs are the 5′-most ORFs, unless a significantly larger ORF was identified on the novel transcript. The ORFs are depicted by genome position in green; those ORFs that span the leader-body junction are shown with corresponding green dashed lines. The MERS-CoV genome schematic with TRSs is shown above the graph for positional reference, and the positive-sense strand-specific Nanopore reads are shown in maroon within the plot for comparison.
Ribosome profiling analysis of MERS-CoV-infected cells identifies time-dependent translation activities on the plus and minus strands and translation corresponding to noncanonical loci
Conventional dogma posits that viral RNAs transcribed during the course of infection should function as transcripts for the translation of viral proteins. To explore the possibility that the plus- and minus-strand leader-containing RNAs identified functioned as transcripts for viral protein translation, we performed ribosome profiling analysis at early and late times post-MERS-CoV infection. Ribosome-protected RNAs were then subjected to strand-specific Illumina RNAseq analysis (Figures 5 and and6;6; corresponding plus- or minus-strand Nanopore sequencing data are provided for comparison in Figures 5A and and6A,6A, respectively). Analysis of the positive-sense ribosome-protected reads indicated the following: (a) ORF1-dependent translation occurred only at early times post-infection (compare Figures 5B and 5C); (b) translation at early times post-infection from genome locations corresponding to ORF2 through the 3′ UTR peaked predominantly at sites with canonical TRSs but also included locations separate from canonical TRS loci that highly corresponded with ORFs encoded on atypical canonical and noncanonical leader-containing transcript sites identified in the Nanopore sequencing analysis (Figures 5B and 5C); and (c) translation from ORF1 had ceased by late times post-infection, though not translation downstream of ORF1 corresponding to canonical and noncanonical TRS loci (Figure 5C). Furthermore, analysis of the negative-strand ribosome-protected reads indicated the following: (a) translation genome-wide peaked predominantly at sites with canonical TRSs, but also included locations separate from canonical TRS loci that highly corresponded with leader-containing transcript sites identified in Nanopore sequencing analysis at sites both similar to and distinct from sites identified in the positive-sense analysis (Figure 6B); (b) ribosome-protected peaks corresponding to the reverse complement TRS junction ORF1 loci identified above (loci 11338, 12362, and 15974—see Figures 2C and and6A6A and Table S3) were detectable at early times post-infection (Figure 6B); and (c) several ribosome protected peaks within ORF1 at both early and late times post-infection did not correspond closely with Nanopore-identified leader-containing loci (Figures 6B and 6C), suggesting that a mechanism in addition to canonical and noncanonical TRS usage may also be contributing to CoV translation from the minus strand. Also of interest, both positive- and negative-sense analyses indicated comparatively high amounts of translation activity at locations within ORF8 corresponding with Nanopore leader-containing transcript data.
Ribosome profiling of the positive-sense MERS-CoV transcriptome
Top: The MERS-CoV genome is depicted as a black line, with annotated proteins shown as blue bars. Transcription regulatory sequences (TRSs) are indicated with dashed lines in the genome schematic and in all panels. MERS-CoV-infected cells were lysed at 6 and 16 h p.i. and submitted to a ribosome profiling assay.
(A) The positive-sense strand-specific Nanopore reads from Figure 2A are shown for reference.
(B and C) Positive-sense ribosome-protected loci at 6 (B) and 16 (C) h p.i. Y axes indicate the ratio of ribosome-protected to total reads. X axis indicates genome position.
Ribosome profiling of the negative-sense MERS-CoV transcriptome
Top: The MERS-CoV genome is depicted as a black line, with annotated proteins shown as blue bars. Transcription regulatory sequences (TRSs) are indicated with dashed lines in the genome schematic and in all panels. MERs-CoV-infected cells were lysed at 6 and 16 h p.i. and submitted to a ribosome profiling assay.
(A) The negative-sense strand-specific Nanopore reads from Figure 2B are shown for reference.
(B and C) Negative-sense ribosome-protected loci at 6 (B) and 16 (C) h p.i. Y axes indicate proportion of coverage. X axis indicates genome position.
Proteomics analysis of MERS-CoV-infected cells reveals the presence of a novel MERS-CoV accessory protein
The high amounts of noncanonical transcription and translation occurring across the CoV genome, particularly within ORF8, prompted us to ask whether as-yet-unidentified protein products were produced from any of these loci. To that end, we surveyed an existing proteomics dataset captured from MERS-CoV-infected Calu3-2B4 cells from 0 to 24 h p.i. for peptides corresponding to novel proteins by searching the data against a novel ORF database containing all possible ORFs equal to or larger than 8 aa on both the positive and negative strand of the MERS-CoV genome. We identified a high-confidence, low-abundance peptide, IPTTISGWSFLSK, produced from a hypothetical ORF encoded within the ORF8 region at multiple time points, particularly late times p.i., in the dataset (Table S4) (the locus of the hypothetical protein is designated ORF8c for this manuscript). ORF8c, encoded from nts 29416–29784 on the plus strand, encodes a putative 123-aa protein with no predicted transmembrane regions. This area corresponded with increased activity on plus-strand ribosome profiling data (Figure 7). BLAST searches in the UniProt and nr databases revealed no high sequence identities to any other known viral, bacterial, or eukaryotic proteins. Junction analysis in this area (see Figures 2C and and3A)3A) revealed the presence of only a low-confidence minus-strand junction near this region (with a junction destination at nt 29506), detected in less than 0.05% of junctions mapped. Thus, while ORF8c might be directly translated from an extremely low-abundance noncanonical leader-containing transcript, the data suggest that translation more likely occurs as a downstream product of the sgmRNA8 transcript.
Novel MERS-CoV proteins identified via untargeted proteomics analysis
MERS-CoV-infected lysates were analyzed by mass spectrometry using a library of noncanonical protein candidates as a reference for the identification of novel proteins. A high-confidence hit (ORF8c, nts 29428–29874) was identified at the 3′ end of ORF8 (see Table S4). The ORF8c protein is depicted as a black box above a graph of positive-sense Nanopore reads (red) and 16 h ribosome profiling reads (blue). Canonical TRSs are shown as dashed black lines. The start positions of ORF8a (Nucleocapsid – N), ORF8b, and ORF8c are shown as dashed green lines. X axis: genome position. Left Y axis: Number of Nanopore reads. Right Y axis: Ratio of ribosome-protected to total reads.
Our proteomics analysis also identified two medium-confidence hits corresponding to peptides within the coding region of ORF1a/b and the negative sense of ORF5 (Table S4). QTFATSSSFGIMNVAIFMVLLLLVYVSTLILMLIQL (representing a translation region encompassing nts 19803–19910 on the plus-strand) is present as the +1 translation frame relative to the nsp15 coding sequence within ORF1a/b. This peptide is associated with a start codon-containing ORF that would begin at an ATG codon in position 19746 and an area with increased ribosome protection within the nsp15 coding region. IMVWCVLQGPYHEKK (representing a translation region encompassing nts 27296–27252 on the minus strand) lies within the coding region of ORF5 and accounts for the majority of a potential novel ORF beginning with the N-terminal isoleucine on the negative strand from 27296–27246. There are low levels of ribosome protection on the negative strand covering these nts; however, other regions on the negative strand have much higher relative coverage. We do not find evidence for strong junction usage in either of the regions upstream of the medium-confidence peptides or ORFs in question and are therefore uncertain how they may be expressed. Verification of these peptides and identification of further peptides from the nsp15 out-of-frame ORF are underway using proteomics approaches to specifically enrich proteins produced late in infection.
To confirm our high-confidence peptide hit we collected lysates at late times p.i. from infectious clone MERS-CoV and mouse-adapted MERS-CoV infection, and performed liquid chromatography–tandem mass spectrometry (LC-MS/MS)-based targeted proteomics analysis to confirm the presence of the previously identified ORF8c peptide as well as additional ORF8c peptides. As a positive control, we made a replicon containing ORF8c up to the 5′ most in-frame ATG codon and infected cells to collect lysates. Using the targeted proteomics method, we were able to identify not only the original high-confidence peptide, IPTTISGWSFLSK, but also two other peptides from ORF8c in all infected samples (Figure 8), confirming ORF8c expression. There was an interest in comparing all three ORFs that may be made from the N sgmRNA, so we also obtained data from ORF8a (N) and ORF8b (Table S5). When comparing the average abundance of all identified ORF8a-c peptides ORF8a is unsurprisingly the most abundant, with decreasing abundance for ORF8b, and ORF8c being the least abundant of the three proteins (Figure 8). Part of our current work is focused on characterizing both translation initiation and the role of ORF8c in the context of MERS-CoV infection.
Confirmation of ORF8C expression and comparison of sgmRNA8 protein abundances, using targeted proteomics analysis
(A–C) MS/MS spectra of the three ORF8c tryptic peptides identified; IPTTISGWSFLSK (A), VASLSALAPVQVLVQ (B), and ILMPTKPSLR (C). Quantification of the Log2 area under the curve for peptides shown in panels (A–C) from icMERS-CoV in closed circles, maMERS-CoV in open circles, and VRP-ORF8c in gray triangles at 24 and 36 hpi (D). Targeted proteomic coverage in red and two potential versions of ORF8c in black (E). Quantification of the Log2 area under the curve for all ORF8a-c peptides in both icMERS-CoV and maMERS-CoV samples at 24 and 36 hpi (F). Average and standard area of the mean calculated and plotted in Prism 9 (D and F).
Discussion
Nidoviruses, the TRS, and discontinuous transcription
The TRSs of nidoviruses, with their three-component structure of 5′ and 3′ flanking sequences surrounding a conserved, 6- to 7-nt CS, mediate a recombination program that conceptually resembles the splice-donor/splice-acceptor convention of eukaryotic mRNA splicing, in which short motif sequences regulate the looping out of introns to form splice isoforms. In fact, preliminary RNA structure analysis indicates that both the leader and body TRS CSs are presented on exposed regions of hairpin loops, suggesting that these sequences mediate a copy-choice selection during TRS CS recognition and either read through a TRS or discontinuously acquire leader sequence.5,15,30,31 The fact that both RNA structure elements and protein interactions have been implicated in TRS recognition indicates that there are sequence features that are pivotal for the efficient production of the various viral RNA species,14,15,32,33 though our studies in SARS-CoV have indicated that the viral RdRp is highly tolerant of dramatic CS positional shifts when duplicating the leader and body TRS domains21,22 (see Tables S1–S3). The studies described herein have identified a much more complex transcription network than has been previously conceived for MERS-CoV (see Figures 3 and and4)4) and parallels findings in SARS-CoV-2.26 In addition to the classic highly conserved TRN directed by highly canonical CS sites, we have described a rich noncanonical network of low- to moderate-abundance paired plus- and minus-strand sgmRNAs, which has not been previously described for CoVs. Using ribosome profiling and proteomics datasets, we have also shown that at least two of these noncanonical transcripts encode novel proteins or peptides that can be identified in infected cells. The use of ribosome profiling in SARS-CoV-2 has additionally highlighted the importance of these datasets to reveal noncanonical ORFs and translation initiation.34 Although speculative, a newly predicted genome organization emerges that potentially encodes a large number of novel ORFs (see Figure 4), C-terminally truncated, out-of-frame ORFs and/or noncoding RNAs that may function in MERS-CoV replication and pathogenesis. As has been described for a different Nidovirus, the arterivirus simian hemorrhagic fever virus18 and for SARS-CoV-2,27 these data provide a resource for targeted molecular and genetic studies designed to identify and characterize the proteome and transcriptome of an emerging coronavirus.
The widely accepted model of coronavirus discontinuous transcription involves a three-step process of leader TRS CS-body TRS CS junction base pairing, formation of an RNA complex that depends on cis interactions among sequences 5′ and 3′ of the junction, and polymerase jumping, with the polymerase including the RdRp and additional replicase/transcriptase proteins in complex with RNA primary sequence and secondary structures.5,35 The leader TRS CS-body TRS CS recognition and pairing process described in the model, with its rapid kinetics and exponential increases in dynamics as the transcription/replication program becomes established in the infected cell, supports the hypothesis that the CoV RdRp can tolerate near-match “cryptic” TRS recognition, as leader-body TRS pairing is likely temporally finite and could be further influenced by the availability of both cis and trans factors, which can be different at different times post-infection. At its core, discontinuous transcription essentially needs a leader-body sequence pairing, which may require only a few nts.4,36,37 Thus, as strongly demonstrated by our data, the actual viral transcriptome has the capacity to include not only sgmRNA species with canonical leader-body junctions but also lower-abundance noncanonical species. Differences in noncanonical sgmRNA production have been illustrated with SARS-CoV-2, wherein more noncanonical sgmRNA species occur at later time points in infection.11 While our methodology most benefitted from long-read sequencing, a recent method, JUMPER, has been developed to assemble discontinuous transcripts from archived short-read coronavirus sequencing data.38 Of the 14 MERS-CoV noncanonical sgmRNAs described in Sashittal et al., four shared identical junctions were described in this study, while some of the remaining transcripts had junctions in similar neighboring areas, highlighting the reproducibility of the detection of noncanonical sgmRNAs across datasets. Notably, while JUMPER does provide the ability to assemble noncanonical transcripts, it cannot differentiate between plus- and minus-strand sgmRNAs unless the experiment was designed to specifically amplify strands prior to short-read sequencing. It is not yet fully clear whether all of these RNA species are translationally active (or if only a subset of them are), or if they serve roles during the course of infection as RNA mediators of replication, immune evasion, or pathogenesis. Our detection of the low-abundance peptides produced from ORF8c and the possible slippage of ORF2 and their correlation with noncanonical leader-containing transcripts indicate that at least some of these additional RNAs are translationally active.
Our analysis of the untargeted proteomics dataset was inherently limited to proteins expressed at relatively high abundance levels, as in the absence of an enrichment step, the majority of proteins identified in mass spectrometry analyses of infected cells are host cell proteins. Because many of the RNAs identified in the Nanopore sequencing data are produced from very low-abundance junctions, we anticipate that any proteins translated from them will also be produced at correspondingly low levels. Therefore, further experiments are underway to enrich viral proteins in infected cell lysates for proteomics analysis along with employing the targeted proteomic methodology used for ORF8c confirmation, which we anticipate will aid in the identification of additional novel low-abundance viral ORFs that are translated during the course of infection.
Data from SARS-CoV-2 studies strengthen the hypothesis that noncanonical RNA may encode additional isoforms or novel CoV proteins.27 In the past 15 years, at least six novel small ORFs in influenza viruses that play novel roles in virus replication and pathogenesis have been identified.39 Other novel small ORFs have been described in arteriviruses, alphaviruses and CoVs.23,40,41 Additionally, there is evidence that small RNAs contribute to the pathogenesis of SARS-CoV,42 and other low-abundance positive- and negative-sense RNA species may play similar roles in MERS-CoV and SARS-CoV-2. There may also be differences in sgmRNA species and noncanonical ORFs produced between interferon (IFN)-competent Calu3 cells and IFN-deficient Vero cells, as IFN-stimulated genes restrict viral replication, transcription, and translation.43 Our untargeted proteomics data was produced in Calu3 cells, while our transcriptomics and targeted proteomics data were produced in Vero cells, which may have led to differences in ORF8c protein abundance and unrestricted transcription of noncanonical sgmRNA species. We are continuing to investigate noncanonical sgmRNA production and to mine CoV mass spectrometry data for other low-abundance CoV proteins produced during infection of more relevant replication models, such as ex vivo cultures.44
Coronavirus minus-strand transcription
The production of viral genome RNAs and subgenomic RNAs, while mechanistically identical, is by convention conceptualized as separate events (replication and transcription, respectively). Therefore, viral minus-strand RNAs, which serve as templates for the production of plus-strand genome RNAs and sgmRNAs, exist in replication/transcription intermediates (RIs and TIs, respectively) and serve as templates to multiple plus-strand molecules over time. The 5′ ends of newly synthesized viral plus-strand RNAs are thought to disassociate soon after they are produced, possibly mediated by interactions with other viral proteins or by being physically dislocated by a component of the viral RdRp complex as it continues to produce more viral plus-strand RNA behind the previously synthesized molecule. This disassociation leaves the 5′ ends of newly synthesized molecules free to serve as translation initiation sites. This replication/transcription mechanism explains both the exponential nature of CoV translation as the infection progresses and the far lower abundance of viral minus-strand RNA. However, our data indicate that both plus and minus strands are translationally active and potentially ambisense, as has been shown for arenavirus, phlebovirus, tospovirus, and tenuivirus45 and for influenza A virus.46 While the minus strand is far less translationally active than the plus strand, its involvement in translation appears to increase as infection progresses (see Figures 6B and 6C), possibly as RNA species accumulate and the viral replication program shunts from active replication to assembly and as molecular requirements change from the production of replicase/transcriptase proteins to proteins that might aid in the successful assembly and egress of infectious particles.16 While ribosome profiling data have been shown for the positive sense RNA of SARS-CoV-2, the translational activity of the negative sense strand has not been investigated thus far.34,47 The identities of these potential proteins encoded by the negative strand of CoVs are not yet known, as their extremely low abundance levels will be exceedingly difficult to detect, even via mass spectrometry. However, by pairing leader-containing transcript data with ORF prediction analyses, we will be able to focus on the most likely produced targets—for example, the minus-strand transcript loci 11338 (featuring multiple ORFs, including a 52-aa ORF), 12362 (featuring an 85-aa ORF), and 15974 (featuring a 143-aa ORF)—and may be able to detect these potentially exceedingly rare protein products using a combination of ribosome profiling and deep sequencing approaches.
The strengths of deep sequencing analysis in the detection of rare transcripts produced during coronavirus replication
Historically, the nature of RNA and sequence detection technologies meant that our knowledge of CoV replication biology was limited to the most abundant species, e.g., those readily detected using radiolabeling, Northern blot, or a variety of molecular biology techniques.6,21,22,48,49,50,51,52,53 This limitation has been especially true for minus-strand RNA, which is produced at levels that are magnitudes below those of plus-strand RNA. The depth of sequencing possible with RNAseq-based strategies such as Illumina and Oxford Nanopore provides opportunities to characterize the complicated replication/transcription program of CoVs more fully than was ever previously attainable. Clearly, our deep sequencing analyses have revealed that the MERS-CoV transcriptome is far vaster than was previously known. This study matches positive-strand data from SARS-CoV-2 experiments and provides key insights into the negative strand transcriptional and translational profiles of CoVs. The depth of data derived from these types of analyses and the possibility of achieving highly sensitive detection of specific types of RNAs based on customizable parameters have greatly opened up viral replication biology, both within the nidoviruses18 and for other viral genera, as has been shown with Kaposi’s sarcoma herpesvirus and Epstein-Barr virus.54,55 These studies, when paired with RNA structure and protein function analyses, will be able to better inform researchers’ investigations of RNA and protein targets for viral inhibition, which will eventually lead to vast improvements in the rational design of CoV vaccines and therapeutics for precise control of emerging viruses with pandemic potential, like SARS-CoV-2.
Limitations of study
Our sequencing and junction analyses were performed on a single isolate of MERS-CoV grown in Vero cells, so junctions may differ when grown in other cell types that have intact interferon signaling or other innate immunity pathways that restrict viral replication. Future studies intend to identify how the interplay between host innate immune effectors and CoV replication may dynamically change sgmRNA networks and TRS biases. Our untargeted proteomic analyses were performed using a single isolate of MERS-CoV in Calu3 cells; however, we validated protein expression using two different infectious clone-derived viral strains in a different cell type. Both our sequencing and proteomics approaches are limited via confidence in read and peptide mapping, wherein more abundant RNA and protein species give the best signal, and our data may bias these species.
STAR★Methods
Key resources table
| REAGENT or RESOURCE | SOURCE | IDENTIFIER |
|---|---|---|
| Bacterial and virus strains | ||
| icMERS-CoV EMC | Scobey et al., 2013 | N/A |
| MERS-CoV Jordan-N3/2012 | NIH, Bethesda, MD | GenBank: KC776174.1 |
| maMERS-CoV M35c4 | Douglas et al., 2018 | N/A |
| Chemicals, peptides, and recombinant proteins | ||
| TRIzol | Thermo Fisher | 15596-026 |
| GlycoBlue | Thermo Fisher | AM9515 |
| Superscript III | Invitrogen | 18080044 |
| Superscript IV | Invitrogen | 18090010 |
| Superase-In Ribonuclease Inhibitor | Thermo Fisher | AM2696 |
| Polynucleotide kinase | NEB | M0201S |
| Apa1 | NEB | R0114S |
| Asc1 | NEB | R0558S |
| 8M Urea | Millipore Sigma | U4883-6X25ML |
| LysC | Wako | 125-05063 |
| Trypsin | Promega | V5111 |
| Zeba spin desalting columns | Thermo Fisher | |
| Critical commercial assays | ||
| Phase Lock Heavy Gel Tube | VWR | 10847-802 |
| Qiagen Oligotex mRNA minikit | Qiagen | 70022 |
| Zymo RNA Direct-Zol kit | Genesee Scientific | 11-331 |
| Phusion PCR enzyme kit | NEB | M0530S |
| Zymo DNA Clean and Concentrator kit | Genesee Scientific | 11-303C |
| 1D Ligation Sequencing Kit | Oxford Nanopore Technologies | SQK-LSK108 SQK-LSK109 |
| Illumina TruSeq Ribo Profile kit | Illumina | ASLPA1212 |
| Ribo-Zero rRNA Removal Kit | Illumina | MRZH11124 |
| NEBNext Small RNA Library Prep Set 1 | NEB | E7300 |
| PureLink PCR Micro kit | Invitrogen | K310050 |
| Pierce Quantitative Colorimetric Peptide Assay | Thermo Fisher | 23275 |
| Deposited data | ||
| Illumina Sequencing Data | NCBI Sequence Read Archive | SRA: SRX952143 |
| Oxford Nanopore Sequencing Data | NCBI Sequence Read Archive | SRA: PRJNA895061 |
| Ribosome Profiling Sequencing Data | NCBI Sequence Read Archive | SRA: PRJNA895061 |
| Untargeted Proteomics Data | MassIVE | MSV000080025 |
| Experimental models: Cell lines | ||
| Vero-81 | ATCC | CCL-81 |
| BHK-21 | ATCC | CCL-10 |
| Oligonucleotides | ||
| 5’- GGCACTGTTCACTTGCAATC-3’ | IDT | N/A |
| 5’-GAATAGCTTGGCTATCTCAC-3’ | IDT | N/A |
| Genome Spanning PCR Primer Set: See Table S3 | IDT | N/A |
| VRP-ORF8c G-Block: CAGCTACCTGAGAGGGGCCCCTATAACTCTCTACGGCTAACC TGAATGGACTACGACATAGTCTAGTCCGCCAAGCGCCACCATGATGATCATGGCA ACCCTGTGTACTTCCTTCGGTACAGTGGAGCCATTAAACTTGACCCAAAGAATCCC AACTACAATAAGTGGTTGGAGCTTCTTGAGCAAAATATTGATGCCTACAAAACCTTC CCTAAGAAGGAAAAGAAACAAAAGGCACCAAAAGAAGAATCAACAGACCAAATGT CTGAACCTCCAAAGGAGCAGCGTGTGCAAGGTAGCATCACTCAGCGCACTCGCAC CCGTCCAAGTGTTCAGCCTGGTCCAATGAGGCGCGCCGTTTAAACGGCCGGCCTT AATTAAGTAACCGATACAGC | IDT | N/A |
| Recombinant DNA | ||
| pvr21-ORF8C | this paper | N/A |
| pvr21-Capsid 3526 | Agnihothram et al., 2018 | N/A |
| pvr21-Glycoprotein 3526 Δ18nt E3 | Agnihothram et al., 2018 | N/A |
| pvr21-GFP | Agnihothram et al., 2018 | N/A |
| Software and algorithms | ||
| Illumina aligner in Geneious Prime | BioMatters | |
| GraphPad Prism 8 and greater | GraphPad | |
| MinKNOW interface v19.05 and greater | ||
| Guppy basecaller v3.1.5 and greater | ||
| Oxford Nanopore aligner in Geneious Prime | BioMatters | |
| Circa | OMGenomics | |
| Proteome Discoverer 2.1. | ThermoFisher | |
| Sequest | Proteome Discoverer | |
| getorf | EMBOSS | |
| Skyline v20.2.0.286 | MacLean et al., 2010 | |
| IPSA | Brademan et al., 2019 | |
| Other | ||
| HiSeq 2500 platform | Illumina | |
| Oxford Nanopore Technologies MinION sequencer | Oxford Nanopore Technologies | |
| Ultimate 3000 nLC | Thermo Scientific | |
| Orbitrap Exploris480 mass spectrometer | Thermo Scientific | |
| Aurora C18 column | IonOpticks | |
Resource availability
Lead contact
Further information and requests for resources and reagents should be directed to an will be fulfilled by the lead contact, Rachel Graham (rlgraham@ad.unc.edu).
Materials availability
pvr21-VRP-ORF8C plasmid is available upon request.
Data and code availability
- •Illumina, Oxford nanopore, and ribosome profiling sequencing data have been deposited at NCBI Sequence Read Archive. Untargeted proteomics data analyzed existing, publicly available data. All data is publicly available as of the date of publication, and accessional numbers are listed in the key resources table.
- •This paper does not report original code.
- •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.
Experimental model and subject details
Cells
Vero-81 (ATCC; CCL-81) cells were cultured in MEM (Gibco, Thermo Fisher Cat. 11-090-099), 10% FetalClone II (HyClone, Thermo Fisher Cat. SH3006603) and 1X Anti/Anti (Gibco, Thermo Fisher Cat. 15240-062). Prior to infection, medium was aspirated from cells and replaced with Opti-MEM (Gibco, Thermo Fisher Cat. 31-985-070) with 4% FetalClone II and 1X Anti/Anti. Cells were maintained in this medium for the course of infection with MERS-CoV. In all circumstances, cells were cultured at 37°C with 5% CO2.
Virus
The MERS-CoV-EMC-based infectious cDNA was generated previously in our laboratory.48 MERS-CoV strain Jordan-N3/2012 (GenBank: KC776174.1) was kindly provided by Kanta Subbarao (National Institutes of Health, Bethesda, MD). MERS-CoV mouse adapted strain was previously generated in our laboratory.56 Virus stocks were made in Vero-81 cells cultured in Opti-MEM + 4% FetalClone II. All infections involving MERS-CoV were performed at Biosafety Level 3 (BSL-3) as previously described,48 and all samples were inactivated according to our established protocols and were verified to lack infectious activity prior to their removal from the BSL-3 for downstream analyses.
Method details
Virus infection and RNA harvest
Vero-CC81 cells were cultured to ∼80% confluence in T75 flasks. Immediately prior to infection, the culture medium was aspirated and replaced with Opti-MEM with 4% FetalClone II. Cells were infected at a multiplicity of infection (MOI) of 5 with MERS-CoV and were incubated at 37°C for 1 h. After 1 h, cells were aspirated, washed 1X with phosphate-buffered saline (PBS), and supplemented with fresh, pre-warmed Opti-MEM with 4% FetalClone II. Cells were then incubated at 37°C for an additional 17 h (total infection time: 18 h). After the infection was complete, the cell supernatant was aspirated and discarded, and 3 mL of TRIzol (Life Technologies, Thermo Fisher Cat. 15596-026) was added to the cell monolayer. The TRIzol was left on the flasks for 5 min to dissolve cell membranes and was then collected in 1-mL aliquots in screw-cap tubes for removal from BSL-3 to BSL-2 for further processing.
RNA isolation for illumina sequencing
RNA for Illumina sequencing was purified from TRIzol suspensions as per the manufacturer’s directions with the following modifications. Once the upper aqueous phase was separated from the phenol phase, the aqueous phase was added to the top of a pre-spun Phase Lock Heavy Gel Tube (5PRIME, VWR Cat. 10847-802). The aqueous phase was supplemented with 500 μL of phenyl/chloroform/isoamyl alcohol (pH 4.3, Thermo Fisher Cat. BP1754I-100), and the tube was closed and shaken for 1 min. Chloroform (500 μL, Thermo Fisher Cat. BP1145-1) was added to the tube, which was then inverted 5 times. The sample was incubated on ice for 5 min and was then centrifuged at 16,000 x g for 10 min at 4°C. After centrifugation, 200 μL of the clear upper phase was transferred to a clean, RNase-free tube, and 1 μL of GlycoBlue (Thermo Fisher Cat. AM9515) and 200 μL of isopropanol (Sigma Cat. I-9516) were added to the tube. The sample was mixed by inverting and was kept at -80°C overnight. The next day, the sample was removed to ice and allowed to thaw completely, after which it was centrifuged at 16,000 x g for 20 min at 4°C. The supernatant was then aspirated, and the RNA sample was allowed to dry at room temperature for 10 min and was then dissolved and resuspended in 50 μL of 0.1X Tris-EDTA (TE) buffer (1X TE, Thermo Fisher Cat. BP2473) diluted with DEPC-treated water (Thermo Fisher Cat. BP561) at 65°C for 10 min.
Enrichment for polyA-containing RNA
PolyA enrichment was accomplished using a Qiagen Oligotex mRNA minikit (Qiagen Cat. 70022) according to the manufacturer’s instructions with the following modification: in the final elution, the hot (70°C) eluate was run through the spin column twice to maximize the RNA yield.
Reverse transcription, library prep, and next-generation sequencing
PolyA-enriched RNA was then submitted to the University of North Carolina Vironomics Facility for reverse transcription, library prep, and next-generation sequencing on the Illumina HiSeq 2500 platform. FASTQ files of the raw data were provided for downstream analysis. Reads were aligned using the Illumina aligner in Geneious Prime (BioMatters). Read coverage (ranging from >200X to >106X depending on genome position) was visualized using GraphPad Prism 8 (GraphPad).
RNA isolation for nanopore sequencing
RNA for Nanopore sequencing was purified from TRIzol suspensions using the Zymo RNA Direct-Zol kit as per the manufacturer’s instructions (Genesee Scientific Cat. 11-331), omitting the DNase digestion step. RNA was eluted from the column in 100 μl of nuclease-free H2O (Life Technologies, Thermo Fisher Cat. AM9937) and was then carried forward to first-strand cDNA production.
Oxford Nanopore sequencing analysis of MERS-CoV amplicon samples
First-strand cDNA was generated from MERS-CoV RNA harvested from infected Vero cells using random hexamers or a strand-specific primer to generate a plus-sense amplicon set (5’- GGCACTGTTCACTTGCAATC-3’) or a minus-sense amplicon set (5’-GAATAGCTTGGCTATCTCAC-3’) with Superscript III (random primers) or Superscript IV (strand-specific primers) (Invitrogen Cats. 18080044 and 18090010). First-strand cDNA was then used to generate genome-spanning amplicons using a genome-spanning PCR primer set (Table S6) using the Phusion PCR enzyme kit (NEB Cat. M0530S). Amplicons were pooled using a Zymo DNA Clean and Concentrator kit (Genesee Scientific Cat. 11-303C). Pooled amplicons were then sequenced using the 1D Ligation Sequencing Kit (Oxford Nanopore Technologies Cat. SQK-LSK108 or SQK-LSK109) on the Oxford Nanopore Technologies MinION sequencer following the manufacturer’s recommended protocols. Sequence reads were basecalled either within the MinKNOW interface (v.19.05 and greater) or using the Guppy basecaller (v.3.1.5 and greater). Reads were then aligned using the Oxford Nanopore aligner in Geneious Prime (Biomatters). Depending on genome position, plus-strand read coverage ranged from 1X to >38000X, and minus-strand coverage ranged from 0X (20 positions) to >34000X. Read coverage and junction locations were visualized using GraphPad Prism 8 (GraphPad). Junction chord diagrams were generated using Circa (OMGenomics).
Noncanonical junction mapping, reads filtering, and ORF discovery
Leader-body junctions were mined from RNAseq data, read mapped using the Geneious for RNA mapping algorithm, with medium sensitivity and "Find fusion genes and introns” and “Trim” turned on. Abundance levels were determined from the number of reads supporting discovery, as measured by Geneious, with percentages calculated based on the total number of junctions mapped. Reads appearing with less than 0.01% representation in discovery analysis were discarded due to low quality or the lack of sufficient support. The ORFs were identified using the “Find ORFs” function in Geneious, with minimum size set to 30 and Start Codons set to ATG, TTG, CTG. All operations were performed using Geneious Prime (Biomatters).
Ribosome profiling analysis of MERS-CoV-infected samples
Vero-81 cells were cultured to ∼80% confluence in 150-mm flasks. Immediately prior to infection, the culture medium was aspirated and replaced with Opti-MEM with 4% FetalClone II. Cells were infected at a multiplicity of infection (MOI) of 5 with MERS-CoV and were incubated at 37°C for 1 h. After 1 h, cells were aspirated, washed 1X with phosphate-buffered saline (PBS), and supplemented with fresh, pre-warmed Opti-MEM with 4% FetalClone II. Cells were then incubated at 37°C for an additional 6 h (for early post-infection analysis) or 16 h (for late post-infection analysis). At the indicated times, cells were lysed in a buffer from the Illumina TruSeq Ribo Profile kit (Illumina Cat. ASLPA1212) containing the following: Illumina TruSeq Mammalian Polysome Buffer, 1% Triton X-100, 10 mM DTT, 10 U DNase I, 100 ng cycloheximide, and nuclease-free H2O. The lysate was clarified by centrifugation for 10 minutes at 20,000 × g and was then carried forward for footprinting analysis. The total RNA (nonfootprinted) sample was treated with 1% SDS, and the footprinted sample was treated with 5 U of TruSeq RiboProfile Nuclease. The samples were incubated for 45 minutes at room temperature, and the nuclease reaction was then stopped with 15 μl of Superase-In Ribonuclease Inhibitor (Thermo Fisher Cat. AM2696). Total and footprinted RNAs were then extracted following the addition of 1 mL of TRIzol (Invitrogen Cat. 15596026) following the manufacturer’s protocol, as described above. The total RNA sample was treated with Ribo-Zero rRNA Removal Kit following the manufacturer’s protocol (Illumina Cat. MRZH11124), and both samples were then treated with polynucleotide kinase (NEB Cat. M0201S). Following gel extraction of the 35-bp ribosome-extracted product from the footprinted sample, the total RNA and footprinted samples were then prepped for RNAseq using the NEBNext Small RNA Library Prep Set 1 (NEB, Cat. E7300) following the manufacturer’s instructions. The libraries were then cleaned up using the PureLink PCR Micro kit (Invitrogen Cat. K310050), and the libraries (155-175 bp) were gel-extracted and submitted for RNAseq analysis. Raw read counts on the plus and minus strands were then normalized based on the ratio of total (nontreated) reads to ribosome-protected reads across iterative spans of 15 nucleotides.
Untargeted proteomics analysis of MERS-CoV-infected lysates
For untargeted, discovery-based proteomics analysis, Thermo .raw proteomics data files were downloaded from the ProteoSAFe server at the University of California at San Diego (MassIVE Accession MSV000080025). To identify previously unannotated peptides/proteins, raw data were imported into Proteome Discoverer 2.1. Using the database searching algorithm Sequest, data were searched against novel MERS-CoV protein databases generated using the getorf algorithm from EMBOSS (https://www.bioinformatics.nl/cgi-bin/emboss/getorf; constraints: 8 aa, ATG/CTG start codons), appended with a reviewed Uniprot human database and a common contaminants database. Tryptic peptides were identified using the following parameters: 10 ppm precursor mass tolerance; 0.6 Da product mass tolerance; up to 2 missed cleavages; minimum peptide length of 6; fixed modification – carbamidomethylation of Cys, variable modification – oxidation of Met. The Percolator mode was used to determine false discovery rates (FDRs), and a peptide FDR of 5% was used to filter all results. Peak areas were then extracted for relative quantification. Scaffold (version 4.7.3, Proteome Software) was used to validate MS/MS-based peptide and protein identifications. Peptide identifications were accepted if they could be established a false discovery rate (FDR) less than 5% by the Scaffold Local FDR algorithm. Further filtering based on Sequest XCorr Score was used to categorize peptides as high confidence (Xcorr 2), medium confidence (2<Xcorr≥1) or low confidence (Xcorr <1).
Generation of ORF8C replicon
A replicon was created using the pVR21 Venezuelan Equine Encephalitis Virus (VEEV) system to express ORF8C up to the furthest in-frame 5’ methionine.57 The pVR21 genome vector was digested using Apa1 and Asc1 (New England Biosciences). A gene block (IDT) was ordered to encode ORF8C with upstream Kozak sequence and flanking matching restriction sites for cloning. The gene block was reconstituted, digested using the Apa1 and Asc1 enzymes, and subsequently used with a T7 DNA ligase kit (New England Biosciences) to introduce the ORF8C gene into the digested pVR21 genome vector. Ligated constructs were transformed into competent DH5-alpha cells for DNA amplification (New England Biosciences). Purified DNA was used to confirm ORF8C insertion and plasmid fidelity via sanger sequencing. Sequence verified pVR21-VRP-ORF8C construct was used along with helper pVR21 Capsid and E glycoprotein constructs as previously described to produce and purify replicons.57 VRP-ORF8C aliquots were frozen at -80 Celsius, safety tested, and titered via immunofluorescent staining of VEEV vector proteins.
Targeted proteomics analysis of novel MERS peptides
Vero-CCL-81 cells were cultured to 100% confluence in 6 well plates. Immediately prior to infection, the culture medium was aspirated cells were washed once with 1X with phosphate-buffered saline (PBS), then cells were infected at a multiplicity of infection (MOI) of 0.3 with icMERS-CoV, maMERS-CoV, or VRP-ORF8C and were incubated at 37°C for 1 h. After 1 h, cells were aspirated, washed 1X with phosphate-buffered saline (PBS), and supplemented with fresh, pre-warmed Opti-MEM with 4% FetalClone II. Cells were then incubated at 37°C for an additional 24 or 36 h. After the infection was complete, the cell supernatant was aspirated and discarded, cells were washed with ice cold 1X PBS and 250 μL of 8M urea in 50 mM Trizma-HCl was added to the cell monolayer. The lysis buffer was left on the flasks for 15 min to dissolve cell membranes and was then collected in screw-cap tubes for removal from BSL-3 to BSL-2 for further processing.
Protein lysates (100 ug; n=3) were reduced with 5 mM DTT at 56°C for 30 min, then alkylated with 15 mM iodoacetamide at room temperature in the dark for 45 min. The samples were diluted to 1M urea and subjected to digestion with LysC (Wako) for 2 h and trypsin (Promega) overnight at 37°C at a 1:50 enzyme:protein ratio. The resulting peptide samples were acidified, desalted using Thermo desalting spin columns, then the eluates were dried via vacuum centrifugation. Peptide concentration was determined using Pierce Quantitative Colorimetric Peptide Assay, then all samples were diluted to 0.3 ug/ul concentration prior to LC-MS/MS analysis.
The peptide samples were analyzed by LC-MS/MS using an Ultimate 3000 nLC coupled to an Orbitrap Exploris480 mass spectrometer (Thermo Scientific). First, a subset of the positive control samples were injected onto an Ion Opticks Aurora C18 column (75 μm id × 15 cm, 1.6 μm particle size), separated over a 120 min method, and analyzed using an untargeted, data dependent acquisition (DDA) method. The gradient for separation consisted of 2–40% mobile phase B at a 250 nl/min flow rate, where mobile phase A was 0.1% formic acid in water and mobile phase B consisted of 0.1% formic acid in 80% ACN. Peptide identification was confirmed for the peptides of interest and a targeted parallel reaction monitoring (PRM) method was developed for these peptides, along with other peptides used as controls (Table S5). For the PRM method, the gradient composition and length were the same as above. The Exploris480 was operated in MS1 and tMS2 mode with a loop cycle time of 1 s. Resolution for the precursor scan (m/z 370–1300) was set to 60,000 with a AGC target set to standard and a maximum injection time set to 60 ms. tMS2 scans (30,000 resolution) consisted of higher collision dissociate (HCD) set to 30; AGC target set to 100%; maximum injection time set to 120 ms; isolation window of 1.5 Da.
For the untargeted, DDA analysis raw data files were processed using Proteome Discoverer version 2.5 (Thermo Scientific). Peak lists were searched against a reviewed Uniprot human database, appended with a common contaminants database, using Sequest. The following parameters were used to identify tryptic peptides for protein identification: 10 ppm precursor ion mass tolerance; 0.02 Da product ion mass tolerance; up to two missed trypsin cleavage sites; (C) carbamidomethylation was set as a fixed modification; (M) oxidation was set as a variable modification. Peptide false discovery rates (FDR) were calculated by the Percolator node using a decoy database search and data were filtered using a 1% FDR cutoff.
For the targeted, PRM analysis, spectral libraries of the targeted peptides were generated from the DDA Proteome Discoverer results. PRM raw data were imported into Skyline (version 20.2.0.286) and peak areas for all product ions were extracted. MS/MS spectrum was annotated using IPSA.58 Average peak area, standard error of the mean, and individual values were plotted in Prism 9.
Accession numbers
The MERS-CoV genome reference sequence is available (GenBank: JX869059). The MERS-CoV Illumina, Oxford Nanopore, and ribosome profiling reads have been deposited on the NCBI Sequence Read Archive (SRA: SRX952143, Illumina Poly-A RNAseq, and SRA: PRJNA895061, Nanopore RNAseq and Ribosomal Profiling RNAseq). The proteomics data were retrieved from the ProteoSAFe server at the University of California at San Diego (MERS-CoV infection MCL002, massive.ucsd.edu, Accession MSV000080025).
Acknowledgments
The authors would like to thank Kevin Chavez for technical support. This work was supported by NIH grants U19 AI107810 to R.S.B., U19 AI106772 to A.C.S., and T32 AI007419 to E.J.F. This research is based in part upon work conducted using the UNC Proteomics Core Facility, which is supported in part by P30 CA016086 Cancer Center Core Support Grant to the UNC Lineberger Comprehensive Cancer Center.
Author contributions
E.J.F.: Investigation, Validation, Formal Analysis, Writing – Review & Editing.
W.S.: Investigation, Validation, Writing – Review & Editing.
A.C.S.: Investigation, Validation, Writing – Review & Editing, Funding Acquisition.
L.E.H.: Investigation, Validation, Formal Analysis, Writing – Review & Editing.
N.K.B.: Investigation, Writing – Review & Editing.
A.A.S.: Investigation, Writing – Review & Editing.
K.K.W.: Formal Analysis, Validation, Writing – Review & Editing.
J.R.T.: Formal Analysis, Writing – Review & Editing.
D.P.D.: Formal Analysis, Writing – Review & Editing.
L.M.G.: Project Administration, Supervision, Writing – Review & Editing.
R.D.S.: Project Administration, Supervision, Writing – Review & Editing.
K.M.W.: Project Administration, Supervision, Writing – Review & Editing.
N.J.M.: Conceptualization, Resources, Project Administration, Supervision, Writing – Review & Editing.
R.S.B.: Conceptualization, Resources, Writing – Original Draft, Writing – Review & Editing, Supervision, Project Administration, Funding Acquisition.
R.L.G.: Conceptualization, Methodology, Validation, Formal Analysis, Investigation, Data Curation, Writing – Original Draft; Writing – Review & Editing, Visualization, Supervision.
Declaration of interests
The authors declare no competing interests.
Notes
Published: May 2, 2023
Footnotes
Supplemental information can be found online at https://doi.org/10.1016/j.isci.2023.106780.
Supplemental information
Peptide list and related identification information for untargeted proteomic analysis of previously deposited data from accession MSV000080025. Tab 1 contains experimental information regarding samples and data processing. Tab 2 contains “high-confidence” peptides identified. Tab 3 contains “medium-confidence” peptides identified. Novel peptides highlighted in yellow.
Peptide list and raw area under the curve values from targeted proteomics. Tab 1 contains the complete list of peptides chosen for targeted proteomic identification of ORF8a-c. Tab 2 is peptides and quantification values corresponding to ORF8c, Tab 3 is peptides and quantification values corresponding to ORF8b, and Tab 4 is peptides and quantification values corresponding to ORF8a (nucleocapsid).