Genetic diversity in vector populations influences the transmission efficiency of an important plant virus

The transmission efficiency of aphid-vectored plant viruses can differ between aphid populations. Intra-species diversity (genetic variation, endosymbionts) is a key determinant of aphid phenotype; however, the extent to which intra-species diversity contributes towards variation in virus transmission efficiency is unclear. Here, we use multiple populations of two key aphid species that vector barley yellow dwarf virus (BYDV) strain PAV (BYDV-PAV), the grain aphid (Sitobion avenae) and the bird cherry-oat aphid (Rhopalosiphum padi), and examine how diversity in vector populations influences virus transmission efficiency. We use Illumina sequencing to characterize genetic and endosymbiont variation in multiple Si. avenae and Rh. padi populations and conduct BYDV-PAV transmission experiments to identify links between intra-species diversity in the vector and virus transmission efficiency. We observe limited variation in the transmission efficiency of Si. avenae, with transmission efficiency consistently low for this species. However, for Rh. padi, we observe a range of transmission efficiencies and show that BYDV transmission efficiency is influenced by genetic diversity within the vector, identifying 542 single nucleotide polymorphisms that potentially contribute towards variable transmission efficiency in Rh. padi. Our results represent an important advancement in our understanding of the relationship between genetic diversity, vector–virus interactions, and virus transmission efficiency.

Aphid and disease management strategies for yellow dwarf disease follow strict thresholds [8].In the UK, the current threshold, the level of aphid infestation above which treatment is recommended, is the presence of a single virus-vectoring aphid in the crop during the early stages of plant growth [8,9].Once the crop reaches growth stage 31, it is able to naturally tolerate yellow dwarf virus infection [10].Similar stringent thresholds are followed in other European countries.These low thresholds probably contribute to increased application of management interventions such as insecticide treatments, directly increasing the development of insecticide resistant, or desensitized, aphid populations [11][12][13].Currently, the same yellow dwarf virus threshold applies to all vector species, and all populations within a vector species.This is an important oversight, as aphid populations are not homogenous and there is inherent diversity within vector populations that can significantly influence the behaviour and phenology of both the aphid and the virus.Indeed, the transmission efficiency of BYDV-PAV differs between cereal aphid species, and a recent review found that transmission efficiency can also vary between aphid populations within a given species [5].For example, transmission efficiency of BYDV-PAV by Rh. padi can range from 39% to 80%, 0% to 100% and 20% to 100% for wheat, barley and oats, respectively [5].The biological drivers behind this variation are poorly understood; however, intra-species diversity (genetic diversity and the presence and diversity of endosymbionts) within aphid populations are some proposed hypotheses [5].
Despite the broad effects intra-species diversity (genotype, B. aphidicola strain, facultative endosymbiont presence and strain) has on aphid phenology and behaviour, relatively few studies have examined how these traits impact aphid-virus interactions.Recent studies have started to explore the potential influence facultative endosymbionts might have on the aphid-BYDV relationship [24][25][26]; however, transmission efficiency is often not directly examined [25] or the observed endosymbiont effects cannot be disentangled from the confounding effect of aphid genotype [24].Genetic variation has been found to underpin transmission efficiency in another aphid-yellow dwarf virus combination [27], but this remains understudied for Rh.padi, Si. avenae and BYDV-PAV.Here, we use Illumina sequencing to characterize genetic and endosymbiont diversity in aphid populations and combine this with BYDV transmission experiments to examine how diversity in vector populations impacts BYDV transmission efficiency.To achieve this, we use the most prevalent BYDV strain found in mainland Europe and the UK (BYDV-PAV) and several populations of the two most important vector species, Rh. padi (seven populations) and Si.avenae (25 populations).Figure 1 represents our study system.Broadly, our results provide biological insights into the drivers behind variable transmission efficiency in an important vector-virus system.

Characterization of intra-species diversity
Aphid populations comprised 25 Si.avenae populations and seven Rh. padi populations, an additional Rh.padi population (RP-12) was included in the genotyping analysis.We included this population to increase the genetic data available for our phylogenetic analyses.All aphid populations were maintained in cup cultures (similar to Leybourne et al. [21]) under controlled environmental conditions (18 ± 2°C; 16 L : 8 D cycle) in a plant growth room on Triticum aestivum cv.Alcedo.The sampling location for all aphid populations, along with their characterized facultative endosymbiont communities, was described previously [11].
For aphid genotyping, approximately 40 aphids (mixed life-stage) were collected into 96% molecular biology grade ethanol.Samples were sent to LGC Genomics GmbH (Berlin, Germany) for DNA extraction and sequencing (150 bp paired-end reads on an Illumina NextSeq 500/550 platform).All DNA extraction, library preparation and sequencing were conducted by LGC Genomics GmbH.Data processing and single nucleotide polymorphism (SNP) characterization were carried out at the Centre for Genomics Research (University of Liverpool).
We used plink to perform linear regression and assess for SNPs associated with BYDV transmission and BYDV titre.For aphid samples, instead of using thinned VCFs, variants were thinned using '--indep 50 5 2' to account for linkage disequilibrium (LD).LD pruning was not performed for symbiont samples.Phenotype association studies were performed in plink using the '--allow-no-sex --noweb --linear --ci 0.95' options.Variants with a p-value of less than 0.05 were deemed to be of interest.
Phylogenetic distance between aphid and symbiont populations along the Newick tree and separation into distinct MDS clusters were used to assign our aphid populations into the putative aphid genotype, B. aphidicola and any associated secondary endosymbionts into putative microbial strains.

Barley yellow dwarf virus-PAV transmission experiments
Apterous adult aphids were randomly selected and placed onto BYDV-PAV-infected T. aestivum cv.Alcedo plants and left to feed for 48 h; source plants had a mean relative virus titre of 2.97 ± 0.62 and the BYDV-PAV culture held at the Julius Kühn Institut was used as a virus source [37].Following this acquisition period, five adult apterous aphids from each population  were selected and placed at the base of a new wheat plant (the BYDV-susceptible cv.Alcedo at Biologische Bundesanstalt, Bundessortenamt und Chemical Industry, growth stage 12 (BBCH)) for 48 h; virus-carrying aphids from the BYDV-PAV stock culture were used as a control.After 48 h, aphids were removed and plants were treated with insecticide.Plants were retained in the controlled environment chamber for six weeks for virus incubation.The plants were then screened for BYDV symptoms [5] and material was collected for serological detection of BYDV infection via a double antibody sandwich enzyme-linked immunosorbent assay (DAS-ELISA).The number of replicates per aphid population ranged from 21 to 24.Experiments were carried out in a controlled environment room (20 ± 2°C; 14 L : 10 D cycle).We used the highly efficient Rh. padi clone R07 [37] as a positive control in our transmission experiments.
For DAS-ELISA, 96 well polystyrene immunoassay microtiter plates were prepared by coating the plates with BYDV-specific polyclonal antibodies (IgG).BYDV-PAV IgG was prepared by the Julius Kühn Institut.The IgG concentration used was 1 : 200, diluted in ELISA coating buffer comprising: Na 2 CO 3 (1.59g l -1 ), NaHCO 3 (2.93 g l -1 ), NaN 3 (0.2 g l -1 ); pH 9.6.One hundred microlitres of IgG solution was added to each well, leaving two wells for blanks.Plates were incubated at 37°C for 4 h in a moist chamber.After incubation, plates were washed four times with wash buffer (Phosphate-buffered saline, (PBS): 40 g NaCl, 7.2 g Na 2 HPO 4 •2H 2 O, 1.0 g KH 2 PO 4 , 1.0 g KCl in 5 l, with 2.5 ml Tween; pH 7.3) using a plate washer (Tecan Hydrospeed, Crailsheim, Germany).Fifty milligrams of leaf tissue was sampled from each plant and placed in a 2 ml bead milling tube containing five steel beads.Samples were homogenized in 500 µl of extraction buffer (wash buffer + 2% polyvinylpyrrolidone K25 and 0.2% dry milk) through shaking in a Precellys ® Evolution homogenizer for 30 s at 25 000 r.p.m.A 100 µl aliquot of homogenate was placed in an IgG-coated well, and three negative controls (uninfected plant tissue) and three positive controls were included in each plate.Plates were covered and incubated at 4-6°C overnight.Plates were washed five times in wash buffer and the enzyme conjugate solution was added.The enzyme (alkaline phosphatase) conjugate solution comprised a 1 : 10 000 dilution of enzyme in extraction buffer.One hundred microlitres of enzyme conjugate solution was added to each well, and the plate was incubated at 37°C for 4 h.Plates were washed four times, and 200 µl of substrate buffer was added to each well.Substrate buffer comprised: 97 ml diethanolamine, 200 mg NaN 3 and 203 mg MgCL 2 •6H 2 O in 1 l H 2 O + 1 mg ml -1 p-nitrophenyl phosphate; pH 9.8.Plates were incubated in the dark for 60 min; endpoint extinction was measured at 405 nm using a plate reader (Tecan Sunrise).Two blank wells (substrate buffer only) were included per plate, and all wells were blank corrected.The extinction intensity is a measure of the relative virus content/titre.The threshold for a positive BYDV-PAV infection was calculated at EXT of more than 0.06 (×(mean negative control) + 3 standard deviations (STD)).

Statistical analysis
All statistical analysis was carried out using R (v. 4.3.0)[38] and R Studio (v.1.3.1093).The following additional packages were used to support data analysis and data visualization: car, v. 3.0-11 [39], ggplot2, v. 3.3.5 [40].In all models, response variables included aphid genotype, B. aphidicola strain and facultative endosymbiont presence; B. aphidicola strain and facultative endosymbiont presence were tested as nested variables within aphid genotype.In the Si.avenae models, Re. insecticola strain was included as an additional explanatory variable (nested within aphid genotype).
BYDV transmission efficiency was analysed using a general linear model fitted with a binomial distribution and a logit link.A binary value (1 = infected; 0 = uninfected) was modelled as the response variable, and each aphid species was tested in a separate model.A Type II Wald χ 2 analysis of deviance test was used to test the model.Differences in viral inoculation (ELISA titre) were examined using linear models.The BYDV titre of successfully infected plants was modelled as the response variable, and each aphid species was tested in a separate model.A type II ANOVA was used to test the model.

Genetic variation in Rhopalosiphum padi influences barley yellow dwarf virus transmission efficiency
From the Illumina data, we identified 6444 Rh. padi SNPs and used these to group the Rh.padi populations into three genetically similar clades (figure 2a,b).We used these clusters to call putative genotypes (clades) for the Rh.padi populations.We also observed clustering for the obligatory endosymbiont B. aphidicola (figure 2c,d) and called putative strains for these based on 34 SNPs.

Transmission of barley yellow dwarf virus-PAV by Sitobion avenae is broadly inefficient and not affected by vector diversity
We identified 5274 SNPs in our Si.avenae Illumina data, and the 25 Si.avenae populations clustered into several clades (figure 3a,b).In contrast with Rh. padi, we did not observe a high level of genetic diversity across the obligatory endosymbiont B.

Discussion
Our work presents an investigation into the influence that diversity in vector populations has on the transmission efficiency of an important cereal virus.We find that the two vector species examined, Rh. padi and Si.avenae, can be broadly categorized into highly efficient and moderately efficient vectors of BYDV-PAV, respectively.For the efficient vector, Rh. padi, we show that virus transmission efficiency is influenced by genetic variation, and we identify 542 SNPs that potentially influence virus  transmission.Broadly, our findings help disentangle the relationship between vector diversity and the transmission efficiency of important plant viruses and provide insights that can guide future research endeavours.
A recent review synthesized information on the transmission efficiency of yellow dwarf virus across the main cereal aphid vectors, including Rh. padi and Si.avenae [5], identifying significant variation in transmission efficiency across virus species and strains, vector species and different clonal populations within a vector species [5].Variation in transmission efficiency in different vector-virus and virus-host (plant) combinations is unsurprising, as vector-virus relationships can be highly specific.Vectors can be characterized as efficient (competent) or inefficient (incompetent) vectors for a given virus strain.For our study, we selected Rh. padi and Si.avenae as these species are considered to be important and efficient vectors for BYDV-PAV [5,41,42].However, as we observed consistently low levels of transmission efficiency of BYDV-PAV by Si. avenae, this aphid species might only be a moderately efficient vector for BYDV-PAV when compared with Rh. padi.It should be noted that although we observed low transmission efficiency of BYDV-PAV in Si. avenae and high efficiency for Rh.padi, this might differ for other BYDV species.For example, Si. avenae is known to be an efficient vector of BYDV-MAV (Tombusviridae: Luteovirus mavhordei) [5]; therefore, while we observe low BYDV-PAV efficiency in the Si.avenae lines tested here, transmission efficiency might be more variable for other yellow dwarf virus species.
For Rh. padi, we observed significant variation in BYDV-PAV transmission efficiency between the populations examined, with transmission efficiency ranging from 40% to 80%.This broadly supports previous observations that transmission efficiency varies significantly between clonal populations within a cereal aphid species [5,37,43].Three mechanisms that could potentially explain this variation were recently proposed [5].These included: (i) the indirect effect of facultative endosymbionts through altered aphid feeding and probing behaviour; (ii) the direct effect of B. aphidicola through variation in endosymbiont-derived chaperonin proteins; and (iii) aphid genetic variation and the presence of vectoring alleles.Our study represents an examination of these hypotheses in Rh. padi and Si.avenae and identifies genetic diversity as a key factor underpinning transmission efficiency in Rh. padi.We excluded behavioural plasticity as a source for varying BYDV-PAV transmission efficiency, as it is low for feeding behaviour associated with virus acquisition and transmission in Si. avenae [44].
Our observation of differential BYDV-PAV transmission efficiency across our Rh.padi genotypes complements results reported for the wheat aphid, Schizaphis graminum [27,[45][46][47].Previous research used efficient and inefficient Sc. graminum populations and two other yellow dwarf virus species, CYDV-RPV (Solemoviridae: Polerovirus) and WYDV-SGV (Solemoviridae) to examine how genetic traits influence virus transmission efficiency [27,[45][46][47][48].This process identified 'vectoring alleles' that underpin efficient CYDV-RPV transmission in Sc. graminum [27] and provides evidence that genetic diversity within vector populations is a key driver of transmission efficiency [45][46][47][48], as found here for our Rh.padi populations.We identified 542 SNPs in Rh. padi that are probably involved in underpinning variation in BYDV-PAV transmission efficiency.However, it should be noted that our observations are based on a relatively small number of aphid populations and that information on additional populations is required in order to fully elucidate the genetic traits underpinning transmission efficiency in Rh. padi.Nonetheless, to the best of our knowledge, no other studies have characterized genetic diversity within different vector populations and linked this with variation in yellow dwarf virus transmission efficiency in Rh. padi [5].Therefore, the previous insights gained in the Sc.graminum (CYDV-RPV) and Sc.graminum (BYDV-SGV) systems, and our new observations in the Rh.padi (BYDV-PAV) system, represent important advancements in our understanding of the relationship between genetic diversity, vector-virus interactions and transmission efficiency.
We found no evidence to support the two other hypotheses recently proposed [5]: (i) indirect effect of facultative endosymbionts through altered aphid feeding and probing behaviour; and (ii) direct effect of B. aphidicola through variation in endosymbiont-derived chaperonin proteins.However, some of the phenotypic traits conferred by facultative endosymbionts can act in a synergistic manner with host aphid genotype [22,49].Therefore, future work that explores endosymbiont effects on BYDV transmission in interaction with aphid genotype, while controlling for host aphid genotype, would enable a more robust examination of these hypotheses.For example, future work could disentangle potential interactive effects by investigating the potential aphid genotype × endosymbiont effects by manipulating the endosymbionts of aphids from clade III (high efficiency) and clade I (low efficiency) through elimination (antimicrobial treatment) and introduction (microinjection) of different B. aphidicola strains and facultative endosymbiont species.

Figure 2 .
Figure 2. Newick tree (a, b) and multidimensional scaling (MDS) clustering (b, d) for Rh.padi (a, b) and B. aphidicola (c, d) based on SNPs.(e) The transmission efficiency for each Rh.padi population and the internal control; colour shows facultative endosymbiont presence, and n represents the number of replicates.

Figure 3 .
Figure 3. Newick tree (a, c, e) and multidimensional scaling (MDS) clustering (b, d, f) for Si.avenae (a, b), B. aphidicola (c, d), and Re.insecticola (e, f) based on SNPs.(g) The transmission efficiency for each Si.avenae population and the internal control; colour shows facultative endosymbiont presence, and n represents the number of replicates.