Identification of fidelity-determined residues of Porcine reproductive and respiratory syndrome virus through structural alignment

ABSTRACT

Porcine reproductive and respiratory syndrome virus(PRRSV) is an economically important pathogen for global pork industry. As a positive-strand RNA virus, lacking exonuclease-mediated proofreading, its RNA-dependent RNA polymerase (RdRP) domain within the nonstructural protein 9(nsp9) plays a vital role in maintaining replication accuracy. To identify the residues of PRRSV that regulates replication fidelity, its RdRP structure was predicted by using Alpha Fold 2 and aligned with the solved structure of coxsackievirus B3 (CVB3) RdRP. This comparison identified conserved residues in PRRSV RdRP that are potentially involved in fidelity. Using site-directed mutagenesis, nucleoside analog sensitivity tests, and next-generation sequencing(NGS), it was found that the nsp9 K541R mutation enhances fidelity, as increasing viral resistance to mutagens like ribavirin, 5-Fluorouracil(5-FU), and 5-Azacytidine(5-AZC), as well as generating lower rate of non-contiguous junctions. In contrast, mutations at other positions, including A394G, L396S, and R401A, reduced fidelity and elevated frequency of recombination and mutation accumulation. Structural modeling revealed that the highly conserved residue K336 is spatially adjacent to the key fidelity site K541 but situated on the opposite side of the RNA channel. We found that K336R exhibits a dissociated “resistance-high recombination” phenotype. The findings reveal the importance of specific residues in PRRSV RdRP for replication fidelity and provide insights into the potential for improving the stability and safety of live attenuated vaccines through targeted modifications. Furthermore, the study emphasizes the structural conservation of fidelity determinants across RNA viruses, despite low sequence similarity, which can offer a framework for identifying fidelity key sites in other viral RdRPs.

Introduction

RNA viruses accumulate mutations during replication at a rate of approximately 10⁻⁴ to 10⁻⁶ per site [1–4]. Low fidelity is thought to facilitate viral evolution by enabling adaptation to host and environmental pressures [5,6]. The RdRP, a crucial enzyme for the replication of RNA viruses, is regarded as a critical part to determin the replication fidelity and nucleotide selection [7,8]. Previous studies on coxsackievirus B3(CVB3) and poliovirus(PV) have shown that mutant viruses remain viable only within a fourfold range of fidelity alterations due to RdRP mutations [9]. Generally, changes in replication fidelity reduce a virus’s fitness compared to its wild-type strain [10–12]. Polymerases typically share a “right-hand” structure, consisting of finger, palm, and thumb domains [13]. The finger domain forms a channel allowing template RNA to be correctly aligned and assists in positioning the incoming nucleotides to the active site located in the palm domain [14–16]. Despite structural similarities, the amino acid sequences of polymerases from different RNA viruses show low identity, except for conserved motifs [17–19].

PRRSV is one of the most economically significant pathogens, which have affected the global swine industry more than 30 years, since its discovery in the early 1990s [20–22]. Over the past decades, PRRSV has been noted for its genetic and antigenic diversity, with numerous virulent strains, genetically distinct from early prototypes, causing severe outbreaks worldwide [23–27]. Efforts to control PRRSV are hindered by its high mutation rate and genetic recombination, which drive the emergence of new strains with altered virulence and antigenicity [28–40]. The use of modified live virus (MLV) vaccines is further complicated by the risk of regaining virulence through serial passaging in pigs or recombination with field strains, raising concerns about safety and efficacy [32,41]. Therefore, understanding PRRSV fidelity is crucial for developing effective control strategies, enhancing vaccine safety, and managing the spread of viral infections [29].

PRRSV is a positive-sense, single-stranded RNA virus in the genus Betaarterivirus of the family Arteriviridae in the order Nidovirales, with genomes of approximately 15kb [36,37]. The viral genome’s ORF1ab encodes at least 14 non-structural proteins(nsps), including RNA polymerase, RNA helicase, and other proteins involved in genome replication [38,39]. In Nidovirales viruses, the replication-transcription complex (RTC) undergoes intra-molecular recombination at specific transcription regulatory sequences (TRSs) to produce several subgenomic mRNAs (sgmRNAs), all sharing identical 5’and 3’ untranslated region (UTR) sequences [40,42,43]. An RTC with low fidelity can also generate defective viral genomes (DVGs), characterized by deletions or lethal mutations, while retaining the 5’and 3’ genome regions, similar to full genomes or sgmRNAs. It has been revealed that in the process of viral replication, various non-canonical viral transcript are extensively generated due to the variation of template switching [44,45], remaining a challenge to solve the mechanism of virus recombination. Tools like ViReMa (Viral Recombination Mapper) [46,47], DI-tector [48], DVG-profiler [49], and VODKA2 [50] are instrumental in detecting the junctions indicative of DVGs within RNA-seq short-read data. The role of DVGs in PRRSV replication remains unclear, though DVGs in related viruses can function as pathogen-associated molecular patterns (PAMPs), stimulating the immune system and disrupting viral replication [51,52]. The PRRSV nsp9 contains RdRP domains, a critical component of the RTC, predicted to regulate replication fidelity [53]. Although no crystal structures of PRRSV RdRP have been resolved, the RdRP domain has conserved motifs and can be modeled by Alpha-Fold 2, which are similar to other RNA viruses [54].

Mutagenesis assays in other viruses have identified certain residues that affect fidelity, prompting the question of whether these orthologous residues in PRRSV also regulate fidelity. Consequently, the PRRSV nsp9 model was aligned with the solved polymerase structure of coxsackievirus B3(CVB3) to identify orthologous residues potentially critical for fidelity, based on previous fidelity studies [7,9,55,56]. Substitutions at these residues were introduced into the genome of the PRRSV strain JXwn06 using reverse genetic techniques. The fidelity of these rescued viruses was then assessed through sensitivity tests to mutagenic nucleoside analogs, along with NGS to analyze mutation accumulation and non-contiguous junction frequencies using ViReMa [46]. After K541 was identified as a fidelity-determining residue, another site, K336, located on the opposite side of the RNA channel, was also analyzed using the same method. These findings successfully identified fidelity-determining residues in PRRSV. It’s suggesting that fidelity-related sites may be spatially conserved across RNA viruses, and the lysine on the RNA channel influence viral fidelity in distinct ways. This opens the possibility of enhancing the genomic stability and safety of live vaccines by increasing replication fidelity through precise modifications of the RdRp.

Materials and methods

Virus and cell culture

The representative highly pathogenic PRRSV (HP-PRRSV) strain JXwn06 (GenBank accession number: EF641008) and its infectous clone recovered strain RvJXwn have been documented previously [57]. The MARC-145 cell line(ATCC, CRL-12231), derived from the African monkey kidney MA-104 cell line, was cultured in Dulbecco’s modified Eagle’s medium(DMEM; Gibco, 12,491,015) supplemented with 10% fetal bovine serum(FBS; Gibco, 16,140,071), 50 U/mL penicillin, and 50 μg/mL streptomycin, at 37°C in a humidified 5% CO₂ atmosphere. Primary porcine alveolar macrophages(PAMs) were isolated from the lungs of 3 female one-month-old specific pathogen-free(SPF) Large White pigs (non-immunized), acclimatized for one week in a sterilized environment prior to experimentation. Pigs were euthanized by pentobarbital sodium (200 mg/kg) intravenous injection and then exsanguinated to ensure death. The lungs were removed for the collection of primary PAMs cells, and the remaining bodies were disposed of harmlessly. All procedures were in accordance with the AVMA Guidelines for the Euthanasia of Animals: 2020 Edition. The PAMs were cultured in RPMI 1640 (Gibco, 61,870,044) with 10% FBS, 50 U/mL penicillin, and 50 μg/mL streptomycin, at 37°C in a humidified 5% CO₂ atmosphere.

Sequence analysis and modeling of PRRSV nsp9-RdRp

The PRRSV RdRP domain was modeled by Alpha Fold 2 using nsp9 residues 184–643 [54]. The structure model was aligned to the crystal structures of CVB3 and PV(Protein Data Bank accession numbers 3DDK and 1RA7) with PyMol.

Cloning, recovery, and verification of mutant viruses

The pWSK-JXwn, a CMV promoter-driven full-length infectious clone plasmid of HP-PRRSV strain JXwn06, was utilized to generate nsp9 mutant strains [57,58]. The primers JX-C-Fusion-F, JX-C-Fusion-R, and the site-directed mutation primers are listed in Table S1. The up- and down-stream fragments were, respectively, amplified with primer pairs (JX-C-Fusion-F/mutation primer R) and (mutation primer F/JX-C-Fusion-R). These two overlapping fragments(≥20 bp) were purified and fused using PCR with JX-C-Fusion-F and JX-C-Fusion-R to create the mutated fragment. The mutated fragment was then cloned into the backbone plasmid, digested with restriction enzymes AscI and NheI-HF (NEB, R0558 and R3131), and assembled via homologous recombination using the ClonExpress MultiS one-step cloning kit (Vazyme, C113-02) to generate the full-length infectious clone plasmid. A similar strategy was applied to generate infectious clone plasmid with two mutations via overlapping PCR. All the full-length plasmids of infection clones were verified by sequencing and prepared using the PureYield plasmid midiprep system (Promega, A2492). The sequenced infectious clone plasmids were transfected into MARC-145 cells using Lipofectamine™ LTX Reagent (Invitrogen, 15,338,100), following the manufacturer’s instructions. Virus-induced cytopathic effects (CPE) were observed 5–7 days post-transfection, after which cells in six-well plates were harvested by freezing and thawing and passaged in MARC-145 cells or PAMs up to passage 3. The genomes of the passaged viruses were extracted and verified by sequencing.

Immunofluorescence assay

After 48 h infection, the inoculated MARC-145 cells in six-well plates were fixed with 100% ethanol for 20 min and then washed twice with phosphate-buffered saline (PBS). The fixed cells were incubated with monoclonal antibody (MAb) CGMCC NO.3238, targeting the PRRSV N protein, at 37°C for 2 hours. After washed with PBS twice, the cells were incubated with a goat anti-mouse secondary antibody conjugated to Fluorescein isothiocyanate isomer I (FITC) at 37°C for 1 hour. Nuclear DNA was stained with 4,’6-diamidino-2-phenylindole (DAPI) for 10 min and then washed thrice with PBS. The cells were imaged using a Nikon Eclipse Ti-U Inverted Fluorescence Microscope, and images were merged with NIS-Elements D software (Nikon).

Nucleoside analog sensitivity assay

The ribavirin(Sigma, R9644) and 5-AZC(Sigma, A2385) was prepared as 200 mM and 100 mM solution in water, respectively. And the 5-FU(Sigma, F6627) was prepared as 200 mM solution in dimethyl sulfoxide(DMSO). MARC-145 cells were pretreated for 1 hour with DMEM containing the specified concentrations of ribavirin, 5-FU, or 5-AZC as nucleoside analogs, or with DMEM containing an equivalent volume of water or DMSO as a control. After the supernatant was removed and the cells were washed twice with PBS, the virus was incubated at an multiplicity of infection(MOI) of 0.01 for 1 hour at 37°C as the inoculum. The inoculum was then removed, and medium with or without nucleoside analogs was added to the respective wells. After 24 hours or 48 hours incubation, the infected cells are lysed by freeze-thaw once to harvest. The virus titers were determined by 50% tissue culture infective dose(TCID₅₀) assay [59]. The nucleoside analog sensitivity assay was performed in three independent experiments.

Preparation of full viral genomes for deep sequencing

PAMs or MARC-145 cells were infected with the indicated virus at an MOI of 0.01 for 24 h or 48 h. Viral RNA was extracted from the cells by using TRIzol (Invitrogen, 15596026CN) and reverse transcribed using Fastking RT kit (Tiangen, KR116) with 14 primers (S2 Table) to cover the whole PRRSV genome.

Short-read Illumina RNA-sequencing of viral RNA

For each reverse-transcribed sample, 2 μg of cDNA was used to construct an NGS library, followed by sequencing using the 2 × 150 nucleotide paired-end approach on the Illumina platform. The NGS was carried out in the sequencing company Baihuiyineng.

Illumina RNA-seq processing and alignment

Raw reads were processed by Trimmo-matic to remove the Illumina adapter with settings (command line trimmomatic PE sample.raw.R1.fq.gz sample.raw.R2.fq.gz sample_1_paired.fq.gz sample_1_unpaired.fq.gz sample_2_paired.fq.gz sample_2_unpaired.fq.gz ILLUMINACLIP:TruSeq3-PE.fa:2:30:10 LEADING:3 TRAILING:3 SLIDINGWINDOW:4:15 MINLEN:36) [60]. Reads shorter than 36 bp were discarded, and low-quality bases (Q score < 30) were trimmed from read ends. The fastq files were aligned to the parantal strain’s genomic sequence using the Python2 script ViReMa(version 0.29) with the commandline python ViReMa.py sample.fq.gz reference_index input.fastq sample.sam -BED – MicroInDel_Length 5 –Output_Dir sample_virema [46]. The sequence alignment map(SAM) file was processed using the samtools to calculate nucleotide depth at each position in a sorted binary alignment map (BAM) file (using command line samtools depth -a -m 0 sample_virema.sort.bam > sample_virema.coverage.txt) [61].

Nucleotide mutation and non-contiguous junction analysis

The mutation rate of each virus was determined by comparing the number of mutations with the total number of mapped nucleotides. Mutations in library were filtered out if their proportion was lower than 0.001. The total number of nucleotides at each genomic position was calculated based on the nucleotide depth recorded in the coverage files generated by samtools. The accumulative mutation numbers were reported as the count of mutations with proportions exceeding 1% at each genomic position.

The junction frequency was determined by comparing the number of nucleotides involved in genomic junctions with the total number of mapped nucleotides. Nucleotides involved in junctions were quantified by summing the nucleotide depths of junctions detected by ViReMa in the BED files. The total number of mapped nucleotides was the sum of the nucleotides at each genomic position calculated as described above. Junctions were plotted and analyzed following the methodology described in the previous report [60].

The reported mutation and junctions should be interpreted as relative comparisons between strains rather than absolute measures, due to the lack of error-corrected sequencing.

Statistical analysis

Statistical analysis for the figures was performed using GraphPad Prism 8 (La Jolla, CA). For comparisons between two groups, a two-tailed unpaired Student’s t-test was used. For comparisons among more than two groups, one-way or two-way analysis of variance (ANOVA) was used. Data are shown as mean ± SD of biological triplicates. Details of the statistical analysis are provided in the respective figure legends. For all statistical tests, p values > 0.05 were considered nonsignificant; p values < 0.05 were marked as *, < 0.01 as **, and < 0.001 as ***. GraphPad Prism software was also used to perform the analysis.

Results

Modeling the PRRSV RdRP structure by Alpha Fold 2 predicts putative fidelity-determinant sites

Several mutations in the RdRP domain of RNA viruses have been linked to replication fidelity and are conserved across the RdRP domains of different RNA viruses [8,16,62–65]. However, it remained unclear if these residues were conserved and functional in PRRSV. Thus, it became important to identify structurally orthologous residues in PRRSV to assess their role in replication fidelity. The PRRSV RdRP domain was modeled by using Alpha Fold 2 (Figure 1A, B), based on the amino acid sequence of nsp9 from PRRSV JXwn06. Residues 1–183 of nsp9 were truncated during modeling to focus on the RdRP domain structure (Figure 1A). The PRRSV RdRP model was then aligned with the reference structure of CVB3 RNA polymerase (PDB ID 3DDK) using PyMol (Figure 1C). The residues previously identified as fidelity regulation sites in the RdRP of CVB3 and PV were mapped in CVB3 (Figure 1C), and the corresponding sites in the PRRSV RdRP domain were selected for mutagenesis based on the model alignment with CVB3 RdRP [7,9,56]. The PRRSV RdRP residues aligning to fidelity-determining sites were identified as nsp9-I342, A394, L396, R401, T403, Y428, K541, and K542. Among these, R401/T403 and K541/K542 aligned closely with nsp12-A239 and K360 of CVB3 (Figure 1A, B). Conservation of these residues was further analyzed by aligning the amino acid sequences of different lineages of genotype 2 PRRSV. All selected residues were conserved across 16 representative PRRSV strains from lineages 1, 3, 5, and 8 (Figure 1D) [66,67].

Figure 1.

PRRSV RdRP structure modeled by AlphaFold 2 and alignment with CVB3 RdRP (A) The RdRP domain of PRRSV nsp9 is depicted in a schematic diagram, with selected residues for mutagenesis marked. The active site of RdRP, identified as SDD, is indicated. (B and C) The PRRSV RdRP domain (B) Modeled by Alpha Fold 2 (without nsp9 N terminal domain) was aligned to the CVB3 RdRP structure(C). Fidelity-determinant residues in CVB3 RdRP are displayed in the 3D graphic, with corresponding residues in PRRSV RdRP labeled. (D) Amino acid sequence alignments of representative PRRSV-2 strains demonstrate that all marked residues are conserved.

Resistance of recovered mutant viruses to the nucleoside analog

Candidate residues were replaced with others differing in charge or side chain size. Residues mutations introduced in this study correspond to substitution types previously documented in CVB3 and poliovirus [55]. These substitutions were individually introduced into the infectious clone of PRRSV strain RvJXwn, as detailed in Table 1. The mutant viruses were rescued by transfecting the full-length plasmids into MARC-145 cells via lipofection. After transfection, the mutant viruses were passaged at least three times to assess their viability, with only 4 out of 23 full-length clones being successfully rescued (Table 1). All non-rescuable clones underwent at least five independent rescue attempts, confirming non-viable in our reverse genetics system.

Table 1.

Mutation sites at PRRSV RdRP that matched to the fidelity-determin residues of CVB3.

RdRp Region	CVB3	PRRSV	Engineered Substitution	Successful Recovery
Fingers	I176V	I342	I342V	No
			I342A	No
			I342K	No
	Y268H	Y428	Y428H	No
	Y268W		Y428W	No
			Y428F	No
Palm	I230F	A394	A394L	No
			A394F	No
			A394G	Yes
			A394I	No
			A394Y	No
	F232Y	L396	L396I	No
			L396F	No
			L396K	No
			L396S	Yes
	A239G	R401	R401K	No
			R401D	No
			R401A	Yes
		T403	T403Q	No
			T403N	No
			T403A	No
	K360R	K541	K541R	Yes
		K542	K542R	No

At the third passage, all rescued mutant viruses were verified by sequencing and indirect immunofluorescence assay (IFA) (Figure 2A). The viruses growth kinetics were also tested (Figure 2B). The nsp9 L396S mutant strain exhibited significantly lower viral titers during late-stage replication compared to the RvJXwn, whereas the nsp9 K541R mutant strain demonstrated both faster replication kinetics and higher peak viral titers than the RvJXwn. To check whether the mutation in each mutant strain was remained after passaging, the mutant strains were passed 10 generation and were sequenced (Fig. S1). The nsp9 A394G\L396S\K541R strain, which were modified based on structural alignment, kept the mutation till the 10th generation, while the nsp9 R401A strain could not keep the mutation till the 4th generation (Fig. S1A).

Figure 2.

Resistance of PRRSV nsp9-RdRp mutant viruses to nucleoside analog mutagens. (A) Identification of the nsp9 mutant viruses by IFA in MARC-145 cells at 48 hours post inoculation(hpi). (B) The multistep growth analysis of mutant strains. At the indicated time points, the infected cells were harvested and subjected to titration by TCID₅₀ assay. (C and D) resistance of PRRSV nsp9 mutant viruses to ribavirin(orange), 5-FU(blue) and 5-AZC(purple). MARC-145 cells were infected at MOI of 0.01, and infection proceeded for 24 hours(C) or 48 hours(D). The virus titers were determined by TCID₅₀ assay. (E and F) The virus titer difference between with or without analog treatment at 24 hpi(E) or 48 hpi(F). The nucleoside analog resistance data are from three independent experiments, with error bars representing standard error of the mean (SEM). Statistical significance is indicated as *p < 0.05, **p < 0.01, ***p < 0.001.

The nucleoside analogs ribavirin, 5-FU, and 5-AZC were used to assess the changes in replication fidelity of the rescued viruses, by evaluating their sensitivity to these mutagens [68–70]. The concentrations of all nucleoside analogs used in this study were determined based on previously established fidelity-testing concentrations used in PRRSV and other viral systems [68–72]. MARC-145 cells were infected with the mutant viruses at a MOI of 0.01, either with or without ribavirin, 5-FU, or 5-AZC treatment for 24 hours, or with ribavirin alone for 48 hours. The titers of each mutant or parental virus, with or without nucleoside analog treatment at various concentrations, were determined by using the TCID₅₀ assay and are shown in Figure 2C, D. These titer differences reflect the viruses’ sensitivity to nucleoside analogs. A larger difference typically indicates greater sensitivity, suggesting that nucleoside analogs induce more potent lethal mutations during viral replication. All viruses exhibited the highest sensitivity to 5-AZC and limited sensitivity to 5-FU. Compared to the parental strain RvJXwn, the nsp9 K541R and L396S mutant strains showed less reduction in titer with 200 μM or 400 μM ribavirin treatment at 24 and 48 hours (Figure 2E, F). Only the nsp9 K541R mutant strain exhibited less reduction in titer with 400 μM 5-FU or 50 μM 5-AZC treatment for 24 hours (Figure 2E). These results demonstrate that the nsp9 K541R mutation conferred resistance to nucleoside analogs, indicating an alteration in nucleotide selectivity in the mutant strain. Although the nsp9 L396S mutant strain showed increased resistance to ribavirin compared to the parental strain RvJXwn, its’ titer was reduced to the same degree as the parental virus with 5-FU and 5-AZC treatment.

JXwn06 nsp9 K541R generated lower rate of non-contiguous junctions and mutations

To better understand the self-recombination characteristics of the mutants in comparison to the parental strain, we analyzed non-contiguous junction patterns and quantified junction frequency using NGS. The mutants and parental strains were used to infect MARC-145 cells at MOI of 0.01, proceeding for 48 h. Viral RNA was subsequently extracted and reverse-transcribed using 14 primers that spanned the entire PRRSV genome. Equal amounts of cDNA from each of the 14 reverse transcription reactions were pooled for sequencing on the Illumina platform. Sequencing reads were aligned to the respective reference genomes using ViReMa, designed to detect deletions larger than 5 base pairs and to align sequences around 25 base pairs upstream and downstream of the junction site [46]. The non-contiguous junctions detected by ViReMa can arise from either inter-molecular recombination or intra-molecular deletion events (characteristic of DVGs), both of which are related to the polymerase detach and rebind during viral replication.

The junction patterns for each strain were plotted based on the positions of junctions detected by ViReMa across the entire viral genome, with coverage ranging from 4200 to 4500 for both mutant and parental strains. The nsp9 A394G and R401A strains had more junction types than the RvJXwn, and the nsp9 L396S had the least junction types among all strains (Figure 3A). The nsp9 K541R mutant strain has similar junctions to the RvJXwn, while showed a reduction in the total number of unique junction types, primarily characterized by a relative absence of junctions forming a cluster near the plot’s diagonal, which typically represent small deletions ( < 500 nt). In contrast, a notable increase in junction diversity was observed specifically within the 5’-terminal ~5000 nt region of the mutant’s genome, indicating a localized shift in recombination or deletion patterns. To quantify recombination, junction frequency was determined by dividing the number of nucleotides involved in junctions by the total number of mapped nucleotides in the corresponding library. This approach ensured that junction frequency was not influenced by the number of approximate mapped reads or genome coverage. Junction frequency was scaled by a factor of 10,000 and reported as the number of junctions per 10,000 mapped viral nucleotides. Based on the statistical analysis of junction frequency (Figure 3B), the nsp9 K541R mutant strain had a slightly lower frequency than the RvJXwn. In contrast, the nsp9 A394G, L396S, and R401A mutant strains exhibited higher junction frequencies, although the nsp9 L396S strain had the fewest junction types (Figure 3A). Additionally, we calculated the number of accumulated mutations that exceeded 1% at each position in the viral genome (Figure 3C), to reflect the quasispecies evolution related to the viral mutation frequency. The nsp9 A394G, L396S, and R401A mutant strains accumulated more mutations than the parental strain RvJXwn, whereas the nsp9 K541R mutant strain accumulated fewer mutations. It should be noted that mutation frequencies reported here are based on standard Illumina sequencing and reflect relative trends rather than absolute mutation rates. The > 1% threshold was chosen to minimize technical artifacts, but background errors may still contribute. Therefore, differences in mutation accumulation should be interpreted as indicative of relative fidelity variations.

Figure 3.

Genome-wide recombination and mutation analysis of nsp9 mutant strains. (A) The non-contiguous junction pattern of the nsp9 mutant strain and RvJXwn. Non-contiguous junctions were mapped to their genomic locations and color-coded according to their frequency: the 5’ junction site represents the start position, and the 3’ junction site represents the stop position and each dot represents a unique junction type, with warmer colors indicating less frequent junctions and cooler colors signifying more frequent junctions. (B) Junction frequency was calculated by dividing the number of nucleotides involved in junctions by the total number of nucleotides mapped by ViReMa. The result was scaled by a factor of 10,000, and expressed as the number of junctions per 10,000 mapped nucleotides. (C) The total number of accumulated mutations, with proportions exceeding 1% at each genomic position, is shown for both mutant and parental strains. The NGS data for mutation and recombination frequency analysis were derived from a single, independent infection experiment, designed for deep sampling of each viral population.

To assess the replication fidelity of the nsp9 K541R mutant and parental strains in PAMs, we infected PAMs at an MOI of 0.01 for 24 hours. During replication in PAMs, the nsp9 K541R mutant strain produced fewer junction types compared to the RvJXwn at approximate 850 genomic coverage (Figure 4A). Statistical analysis of junction frequency indicated that the nsp9 K541R mutant strain generated junctions less frequently than the RvJXwn in PAMs (Figure 4B). While, the nsp9 K541R mutant strain accumulated more mutations than the parental strain in PAMs (Figure 4C), suggesting that host cellular environment may influence the manifestation of fidelity-related phenotypes [63,73].

Figure 4.

Recombination and mutation analysis of the nsp9 K541R strain in PAMs. (A) The non-contiguous junction pattern of the nsp9 mutant strain and RvJXwn in PAMs. Non-contiguous junctions were plotted to their corresponding genomic locations and color-coded to reflect their frequency within the junction population: the 5’ junction site is indicated as the start position and the 3’ junction site as the stop position, and each dot represents a unique junction type, with warmer tones signifying less frequent junctions and cooler tones indicating more frequent junctions. (B) The junction frequency for each column was determined by dividing the number of nucleotides involved in junctions by the total number of nucleotides mapped by ViReMa. This ratio was scaled by a factor of 10,000, resulting in a frequency expressed as the number of junctions per 10,000 nucleotides. (C) The total number of accumulative mutations of the nsp9 K541R and RvJXwn strain at each genomic position with a proportion exceeding 1% is presented. The NGS data for mutation and recombination frequency analysis were derived from a single, independent infection experiment, designed for deep sampling of each viral population.

Nucleoside analog resistance of nsp9 K336 and K541 mutant viruses

The PV 3D K359H mutation, structurally analogous to PRRSV nsp9 K541, has been shown to regulate replication fidelity through elongation assays [56,74]. In addition, structural modeling of the PRRSV RdRP domain identified another lysine, K336, located near K541 with an outward-facing side chain within the RNA channel (Figure 5A). Sequence alignment of nsp9 confirmed that K336 is conserved across genotype 2 PRRSV strains from lineages 1, 3, 5, and 8 (Figure 5B) [66,67]. To explore whether K336 influences viral fidelity, and to assess the effects of different nsp9 K541 mutations, we constructed the nsp9 K541A, K541H, K336R, K336A, and K336R+K541R mutant viruses. These mutant viruses were confirmed by sequencing and IFA (Figure 5C) All the K336 and K541 mutant strain kept the indicated mutations till the 10th generation (Fig. S1B). In the nsp9 K336 and K541 mutant strain, the nsp9 K336A and K541A mutant strain had the most significant difference in replication (Figure 5D). The nsp9 K336A strain reached the peak titer at the 24 hour post infection, while the titer decreased after 24 h and had a significant difference from the RvJXwn at 48 h and 60 h. The nsp9 K541A strain had significant lower titer than the RvJXwn at 12 h\36 h\48 h\60 h, as well the peak titer was the lowest in all mutant strains and the RvJXwn.

Figure 5.

Resistance of PRRSV nsp9 K336 and K541 mutant viruses to nucleoside analog mutagens. (A) The spatial positions of K336 and K541 in the PRRSV RdRp domain are highlighted with arrows. Both lysines are located within the RNA channel and near the SDD active site. (B) Amino acid sequence alignments indicate that K336 is conserved across PRRSV-2 strains. (C) IFA identification of PRRSV N protein in MARC-145 cells at 48 hpi. (D) The growth kinetics of the K336 and K541 mutant strains. (E and F) The titer of PRRSV nsp9 K541/K336 mutant viruses with or without ribavirin, 5-FU or 5-AZC treatment. Each treatment was colored as mentioned above. MARC-145 cells were infected at MOI of 0.01, and infection proceeded for 24 hours(E) or 48 hours(F). The virus titers were determined by TCID₅₀ assay. (G and H)The viruses titer difference between with or without nucleoside analog treatment at 24 hpi(G) or 48 hpi(H). Data reflect results from three independent experiments, with error bars indicating SEM and statistical significance is noted(*p < 0.05, **p < 0.01, ***p < 0.001).

The mutant viruses were tested for sensitivity to nucleoside analogs at an MOI of 0.01 in MARC-145 cells. Infected cells were collected and titered at 24 hours (Figure 5E) and 48 hpi (Figure 5F). After treatment with nucleoside analogs for both 24 hours (Figure 5G) and 48 hours (Figure 5H), the nsp9 K336R and K336R+K541R mutant strains showed similar titer reductions to the K541R strain. In nearly all mutagens treatment, these reductions were significantly lower than those observed in the RvJXwn, except for the nsp9 K336R mutant strain at 48 hpi with ribavirin. Additionally, the nsp9 K336R+K541R mutant did not display increased resistance to nucleoside analogs compared to the single K541R mutation, suggesting that both lysines may perform similar roles in conferring resistance to these analogs. While the K541H did not confer the same level of resistance as K541R. Meanwhile, the inactive K541A and K336A mutations caused similar or greater viral titer differences compared to the wild-type strain at both 24 and 48 hours.

The nsp9 K336 and K541 mutation significantly altered the recombination and mutation characterization of PRRSV

Following the nucleoside analog resistance assay for the nsp9 K336 and K541 mutant viruses, we sequenced these mutant strains and analyzed their recombination patterns and mutation accumulation. The nsp9 K336R, K336R+K541R, K541H, and K541A mutant strains exhibited more non-contiguous junction types near the 5’ end of the viral genome compared to the RvJXwn, while the nsp9 K336A mutant strain had fewer junction types (Figure 6A). Based on the junction frequency results, the nsp9 K541A mutant strain exhibited the highest frequency compared to all other mutant and parental strains (Figure 6B). The junction frequencies of the nsp9 K541A and K336R mutants were 1.8- and 1.6-fold higher than that of the parental RvJXwn strain (Figure 6B). The nsp9 K336A and K336R+K541R mutants showed slightly elevated junction frequencies, while the nsp9 K541H mutant exhibited a junction frequency comparable to RvJXwn (Figure 6B). Although the nsp9 K336R and K336R+K541R mutant strains showed more nucleoside analog resistance than the RvJXwn (Figure 5F, G), they also exhibited higher junction frequencies (Figure 6B). The junction frequency of the K336R mutant was 1.6-fold higher than that of RvJXwn, but this frequency decreased to 1.1-fold when the K541R mutation was introduced, suggesting a downregulation of recombination due to the K541R mutation and these two lysines regulated fidelity through distinct mechanisms. All mutant strains accumulated more mutations during replication compared to the RvJXwn, with the nsp9 K541A mutant showing the greatest increase in mutation accumulation (approximately 2.4-fold) (Figure 6C). Given that the nsp9 K541A mutant showed the highest values in both junction frequency and mutation accumulation (Figure 6B, C), K541 appears to be a key determinant of viral replication fidelity. This aligns with the nucleoside analog resistance test results (Figure 5F, G).

Figure 6.

Genome-wide recombination and mutation analysis of the nsp9 K336 and K541 mutants. (A) The non-contiguous junction pattern of the nsp9 K336 and K541 mutant strain and RvJXwn. Non-contiguous junctions were mapped to their corresponding genomic locations and color-coded based on their frequency: the 5’ junction site is marked as the start position and the 3’ junction site as the stop position. And each dot represents a unique junction type, with warmer colors indicating less frequent junctions and cooler colors indicating more frequent junctions. (B) The frequency of junctions within each column was determined by calculating the ratio of nucleotides involved in junctions to the total number of nucleotides mapped by ViReMa. This ratio was then scaled by a factor of 10,000, resulting in junction frequency reported as the number of junctions per 10,000 mapped nucleotides. (C) The total number of accumulated mutations, with a proportion exceeding 1% at each genomic position, was counted for each strain. The NGS data for mutation and recombination frequency analysis were derived from a single, independent infection experiment, designed for deep sampling of each viral population.

Discussion

The RNA-dependent RNA polymerases (RdRPs) of positive-strand RNA viruses exhibit structural conservation, featuring similar finger, palm, and thumb domains, suggesting a commonality in fidelity determinants across these viruses [16]. Mutations in the palm domain significantly affect nucleotide discrimination, which is closely tied to elongation rates and mutation frequencies. In contrast, mutations in the upper region of the finger domain primarily influence elongation rates with a minimal impact on fidelity [7], indicating the importance of the mutation site selection. Although the structure of the PRRSV RdRP domain remains unsolved, it is plausible that PRRSV fidelity determinants can be identified through orthologous recognition by structural modeling and alignment with RdRPs from other RNA viruses.

PRRSV is recognized as a low-fidelity virus due to the absence of a proofreading exonuclease (ExoN), leading to genetic and antigenic variation in the field [75]. Generally, the replication fidelity of RNA viruses is associated with their RdRPs, and site-directed mutations can alter this fidelity. However, the fidelity-determinant residues in the PRRSV RdRP have not been extensively investigated, which is crucial for understanding PRRSV evolution and accelerating vaccine development. The fidelity mutants screened for PRRSV were primarily obtained through passaging with mutagens such as ribavirin [71,76], which was generally used in other RNA viruses fidelity mutant screening as well [77–81], requiring significant time and cost. In this study, an attempt was made to identify PRRSV fidelity determinant residues through homologous structure alignment and mutagenesis assays, based on fidelity determinants from other distantly related positive-strand RNA viruses. Candidates for the active site associated with fidelity were screened across multiple RNA virus RdRPs, based on their biochemical properties and positions within the 3D structure, and finally identified through nucleoside analog mutagens sensitivity testing and NGS analysis [9,56,77,81–83]. In contrast to previous PRRSV fidelity studies that identified residues such as A283, H421, V218, and P386 through ribavirin passaging, our structure-guided approach revealed a distinct set of fidelity-determining residues – including K336, A394, L396, R401, and K541—located in spatially discrete regions of the RdRP. These sites occupy critical positions within or near the RNA channel and active site-associated motifs, suggesting they may regulate fidelity through mechanisms different from those previously reported. Importantly, our findings demonstrate that fidelity determinants can be predicted through cross-viral structural alignment even in the absence of high sequence conservation, providing a generalizable framework for identifying functional residues in unresolved viral polymerase structures. This approach not only expands the repertoire of known fidelity-regulating sites in PRRSV but also underscores the structural conservation of replication fidelity determinants across divergent RNA viruses. However, only 4 of the 23 mutations were viable for PRRSV, with the remainder being non-viable in our reverse genetics system.

To maintain the correct orientation of the triphosphate, an intricate hydrogen-bonding network is established in the polymerase, extending to residues within the ribose-binding pocket [84]. Residues within or near this pocket, when mutated, may perturb the position of the triphosphate or reduce the efficiency of phosphoryl transfer, as observed with the PV 3D H273R mutation [85]. Since fidelity is associated with the second conformational change in nucleotide incorporation catalysis [8,86], fidelity mutants tend to exhibit changes in RNA synthesis rates, which could also be a cause of lethality and requires further verification through biochemical assays.

Currently, there is no established method for nucleotide incorporation assays of PRRSV to analyze elongation rates and nucleotide selectivity due to defective RdRP activity in vitro. Mutations in viable strains confer resistance or sensitivity to three types of nucleoside analog mutagens and result in enhanced or weakened fidelity. While the proofreading mechanism of PRRSV is not well understood, the fidelity alterations are limited compared to the nsp14-ExoN(-) mutant coronavirus, which has been shown to significantly affect fidelity [55,60].

In the order Nidovirales, a unique N-terminal nucleotidyl transferase domain is connected to the RdRp [87,88], yet its specific role in viral replication has not been thoroughly verified. Therefore, the N terminant of nsp9 was excluded from structural alignment, permitting direct comparison of the fidelity determinants with the CVB3 RdRP domain. Through mutagenesis studies, these structurally conserved residues were confirmed to influence the fidelity of PRRSV, further supporting the notion that fidelity determinants are conserved across distantly related RNA viruses. This implies that RdRPs possess similar functional determinants in terms of structure, despite low amino acid sequence identity among these domains [56]. It indicates that highly conserved proteins with unresolved structures can be modeled to identify key functional sites by aligning with proteins with solved structures. This approach significantly enhances the efficiency of screening and identifying key candidate sites in proteins lacking 3D structurew information.

Mutations at orthologous residues in RdRP can affect viruses differently. For instance, the orthologous residue at site 230 is isoleucine in CVB3 and phenylalanine in PV. When I230 of CVB3 was mutated into phenylalanine, it resulted in decreased fidelity; similarly, the F230I mutation in PV led to an increase in fidelity [7,9]. For PRRSV, the orthologous mutations A394F and A394I in nsp9 were non-viable in our reverse genetics system, although the A394 of PRRSV RdRp and the residue 230 in enterovirus both locate at the terminus of an orthologous β-pleated sheet near the active site. Concurrently, the PRRSV nsp9-K541R mutant exhibited enhanced fidelity, akin to the effect of nsp12-K359R in PV [89]. However, CVB3 nsp12-K360R and MHV nsp12-K794R-ExoN(-) are not viable in PRRSV [7,55]. The role of PV 3D K359 in motif D is presumed to be a proton donor (general acid) in catalysis [90]. According to the solved structures, this site is not only near the entrance of the NTP tunnel but also near the NTP γ-phosphate, albeit too distant (4–5Å) to directly contact the NTP γ-phosphate [91]. Based on molecular dynamics (MD) analysis, when NTP is fully integrated at the catalytic center, K359 can act as a sensor to facilitate the initiation of active site closure [92]. Introducing the mutation 3D K359R into PV, the 3D polymerase elongation rate was decreased, and fidelity was enhanced by affecting rate-limiting conformational change steps before and after the catalytic reaction [56], suggesting that PRRSV nsp9 K541R might serve a similar function to improve fidelity. For another mutant strain, the nsp9 L396S mutant strain exhibited similar resistance to ribavirin compared to the nsp9 K541R mutant strain. It has been reported that the CVB3 nsp12-V553l nsp14-ExoN (-) mutant strain was sensitive to 5-FU but resistant to 5-AZC [55], and similar findings was also reported in other mutant viruses [93,94]. To reveal the mechanisms behind these differences, the most commonly used is comparing the biochemical kinetic differences of the viral polymerase in RNA synthesis [8,82,95]. However, the nsp9 with polymerase activity in vitro was not successfully expressed. Due to the unclear RTC assembly mechanism for PRRSV, the establishment of an active in vitro elongation assay for the PRRSV polymerase remains a challenge for the field.

Furthermore, the contrasting mutational burden of the K541R mutant in MARC-145 cells versus PAMs reveals that host cellular environment may influence the manifestation of fidelity-related phenotypes. Further experiments comparing viral replication kinetics and host factor interactions between MARC-145 cells and PAMs would be needed to elucidate the mechanistic basis of these differences. A limitation of this study is that the analysis of mutation accumulation and recombination frequency was derived from a single deep-sequencing experiment. Future studies incorporating biological replicates in transcriptomic analyses could further strengthen the generalizability of these findings. It will be important for future work to systematically evaluate these fidelity mutants in primary PAMs and other physiologically relevant cell types, comparing their replication dynamics, recombination frequency, and mutation spectra, to ultimately determine the in vivo consequences of altering RdRP fidelity.

Nsp9 K336 and K541 are spatially proximal but located on opposite sides of the RNA channel, and they appear to modulate fidelity in distinct ways. The K541R mutant exhibited classic high-fidelity traits: resistance to nucleoside analogs, reduced mutation accumulation, and lower junction frequency. In contrast, the K336R mutant presented a complex phenotype: it was resistant to nucleoside analogs yet exhibited increased junction frequency and mutation accumulation. This could arise if the mutation alters the polymerase’s affinity for or dissociation from the RNA template, promoting premature detachment and re-annealing at non-contiguous sites. Such a mechanism would increase recombination/DVG formation without necessarily altering base-pairing fidelity during elongation.This suggests that K336 May not primarily affect the chemical step of nucleotide incorporation but, given its outward-facing location, might instead influence the polymerase’s interaction with the RNA template, more specifically affecting the frequency of template switching (leading to the observed increase in recombination/DVGs) and/or replication fidelity in certain genomic contexts. When K541R was introduced together with K336R (K336R+K541R double mutant), the junction frequency decreased compared to the K336R single mutant, suggesting that the high-fidelity character of K541R can partially counteract the recombination-prone phenotype conferred by K336R. This supports the notion that these two residues influence the replicase complex function via different molecular interfaces. We therefore hypothesize that K336 may be part of a structural motif that modulates template threading or polymerase stability during replication, thereby influencing recombination propensity. Future structural insights into the PRRSV RdRP and biochemical studies will be crucial to delineate their precise mechanisms of action. In the genetic background lacking exonuclease proofreading, CVB3 nsp12 mutations V553I and M611F confer resistance to 5-FU [55]. NGS of these two mutants revealed that the nsp12 V553I mutant strain had fewer accumulated mutations than the wildtype virus, whereas it was the opposite for the M611F mutant strain [55], suggesting that despite their similar capacities for nucleoside analog resistance, the mutant strains’ ability to accumulate mutations varies, and the correlation between them remains unclear.

The exonuclease activity in coronaviruses is crucial for proofreading during replication, which is why mutagenesis experiments are typically conducted by using viruses with a nonfunctional or absent exonuclease to observe the changes in fidelity [55,60]. In contrast, PRRSV lacks a similar exonuclease for proofreading, and thus, the impact of proofreading was not factored into mutagenesis experiments aimed at studying PRRSV fidelity. Consequently, further investigation into the fidelity determination mechanisms in PRRSV replication could yield valuable insights for studying related viruses that lack replication proofreading proteins.

As previously reported, mutations in fidelity-determined residues can result in viral attenuation [65,94,96–100], offering a novel perspective for vaccine development. However, the reduced replication fidelity in attenuated viruses raises concerns about the potential for virulence recovery. For instance, the RdRP of the Sabin I vaccine strain has a T362I mutation, which has been identified as a low-fidelity variant and contributes to the attenuated phenotype [97]. While our study demonstrates that mutations alter replication fidelity and recombination patterns in vitro, the impact of these mutations on viral pathogenicity in vivo remains to be directly evaluated. Given the genetic plasticity and adaptive capacity of PRRSV, any fidelity-altering mutation intended for vaccine development should be rigorously assessed in animal models to ensure that it does not inadvertently enhance pathogenicity or facilitate reversion to virulence. Further research on viral fidelity can provide theoretical support for the development of safer and more effective vaccines. In vivo studies evaluating virulence, immunogenicity, and genetic stability will be essential to assess the suitability of these fidelity mutants as vaccine candidates.

Conclusion

By utilizing structural alignment and mutagenesis, the nsp9 K541R mutation of PRRSV was found to enhance fidelity, while mutations at other sites such as K336A, A394G, L396S, and R401A resulted reducement. As well, the distinct effects of the proximal K336 and K541 mutations on viral fidelity, which locate on the opposite sides of the RNA channel, suggest they may engage in different mechanisms of action. These findings not only underscore the potential for modulating viral fidelity through targeted modifications to improved genomic stability and safety of live attenuated vaccines, they also provides evidence that fidelity determinants in RNA viruses may be structurally conserved, despite low sequence homology across different viral RdRPs. In future, it is needed to further characterize the biochemical properties of the PRRSV RdRP and its multiprotein complex, to gain a deeper understanding of the fidelity determination mechanisms and their implications for virus evolution and control.

Funding Statement

This work was supported by the National Natural Science Foundation of China [32573333, 32330106, 31772759] and China Agriculture Research System of MOF and MARA [CARS-35].

Data availability

The viral genome sequencing data was deposited in the Sequence Read Archive (SRA) database of National Center for Biotechnology Information(NCBI) with the accession number PRJNA1162805.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data that support the findings of this study are openly available in https://www.ncbi.nlm.nih.gov/bioproject/?term=PRJNA1162805.

Ethics approval

The protocol for primary PAMs preparation was approved by the Laboratory Animal Ethical Committee of CAU, with approval No. AW51905202-2–01.