Influence of mixed and single infection of grapevine leafroll-associated viruses and viral load on berry quality

Abstract

Grapevine leafroll disease is a viral disease that affects grapevines (Vitis vinifera L.) and has a severe economic impact on viticulture. In this study, the effect of grapevine leafroll-associated viruses (GLRaV) on berry quality was investigated in clones of cultivar cv. Crimson Seedless table grapes infected with GLRaV. RT-PCR confirmed the identity of the clones: clone 3236, infected only with GLRaV-3 (termed single); clone 3215, infected with GLRaV-3, GLRaV-4 strain 9 and grapevine virus A (termed mixed); and a viral free clone of the same genetic background of the infected clones (termed control). The berry quality indices of size, sugar, acidity and anthocyanin content were measured at harvest maturity. RT-qPCR was used to determine the viral load. The study was repeated over 2 year. A two-way, multivariate analysis of variance was applied with clone and year as independent variables and the measured berry quality parameters as a dependent variable. All dependent variables were significantly affected by viral infection (Wilks, λ, (2,33) = 0.033895, P-value <0.001), while only titratable acidity was affected by year. The average berry dry mass decreased (P-value <0.001). The water content of both infected clones was greater than that of the control (P-value <0.001). Both infected clones displayed reduced sugar content as a fraction of the berry dry mass (P-value <0.001). The anthocyanin and the phenol content of the infected clones were significantly reduced compared with the control clone (P < 0.001, P < 0.05, clone 3236 and clone 3215, respectively). Finally, the viral load was highly variable, and no quantitative relationship between viral load and berry composition was found.

Introduction

It is well known that the cultivated grapevine (Vitis vinifera L.) is exceptionally vulnerable to viral infection. Grapevine leafroll-associated viruses (GLRaV) are among the most widespread in vineyards (Martinson et al. 2008, Naidu et al. 2014). It has been established that six species of the family Closteroviridae are responsible for grape leafroll disease (GLD) (Maree et al. 2013, Velasco et al. 2014, Sharma et al. 2015, Adiputra et al. 2019). This disease may result in severe reduction in fruit yield, vigor and delay in fruit ripening of wine grapes (Cabaleiro et al. 1999, Naidu et al. 2014). Although extensive studies have been conducted, the mechanism of virus–host interaction that affects fruit quality is still uncertain. In addition, the majority of reports were conducted with red-berried wine varieties (Lee and Martin 2009, Lee et al. 2009, Alabi et al. 2016, El Aou-ouad et al. 2016, Montero et al. 2016b). Comparatively, few studies have been conducted on table grapes, where the effects on quality and yield are less pronounced or deleterious (Singh Brar et al. 2008). To date, studies of GLD have focused on a single virus species, predominantly GLRaV-3, because of its high impact and virulence. In comparison, the influence of GLRaV-4 strain 9 on berry quality in grapevine has scarcely been reported (Maree et al. 2013) nor has the impact of mixed infections. In both wine and table grapes, viral infection appears to delay ripening (Lee and Martin 2009, Lee et al. 2009, Komar et al. 2010), leading to reduced sugars and anthocyanins and an increase in the berry weight and volume. Alabi et al. (2016) indicate that the viral effect on sugar levels was more significant at and after véraison than pre-véraison. Thus, changes in berry morphology may be linked to source–sink effects since GLRaV-3 spreads through the phloem, and post-véraison increase is almost completely due to the mass flow of phloem solution (Wang et al. 2003, Choat et al. 2009, Zhang and Keller 2017). This study sought to investigate the nature of the influence of single (clone 3236) and mixed (clone 3215) viral infections on the berry quality of cv. Crimson Seedless V. vinifera L. which is an example of table grapes grown under commercial vineyard conditions. Furthermore, it sought to determine whether the viral copy number influenced berry quality.

Materials and method

Plant material

The study utilized two clones of V. vinifera L. cv. Crimson Seedless that were previously generated by the Department of Agriculture and Food of Western Australia and were used in a previous study. Replicates of these clones were generated by cuttings of the original infected clone and then individually grafted onto rootstocks. Clone 3236, infected with GLRaV-3 with mild symptoms, will represent the single infection group. Clone 3215, infected with GLRaV-3, GVA and GLRaV-4 strain 9, will represent the mixed infection group (Singh Brar et al. 2008, Alagappan 2011). Both infected clones were compared with cv. Crimson Seedless viral-free vines (mock-infected). Each group had six vines as a biological replicate. The vines were 15 year old, grafted onto Schwarzmann rootstock and grown in a commercial vineyard located in the Swan Valley in Western Australia (−31.827789, 115.999947). The grapevines were spaced 3.3 m between the rows and 2.4 m between the vines. The infected vines were set in one row in the vineyard. The clones were planted in sequences of three individuals, separated by a non-experimental vine, followed by another three individuals of the same clone, another non-experimental vine and the same pattern repeated for the second clone. The control vines were located in an adjacent row of the infected clones. The virus status of control clones was assessed during 2 year of the study for the presence of the disease symptoms, and an RT-PCR test was carried out to confirm the absence of viral infection in the vines. Then the result was confirmed with RT-qPCR in the experiment of the viral load. Moreover, no infection has been reported with other types of viruses to the Department of Primary Industries and Regional Development of WA during the 2 year of the study.

Berry sample collection

Five bunches per vine were randomly collected at the harvest stage of 2 year (EL38, March 2017/18). All the samples were collected on the same day and finished before 10:00 AM. °Brix and titratable acidity (TA) were measured on the sampling day. The ripening stage of the berries was established using the recommendation of OIV resolution VITI 1/2008 (OIV 2008) and UE Commission Regulation 543/2011. Table grapes are considered to be ripe at a °Brix value higher than 16 ^°Brix or when the SSC (expressed as g.L⁻¹)/TA (expressed as g.L⁻¹ tartaric acid) ratio is higher than 20, with the exception of seedless varieties where ripeness is considered at TSS 14 ^°Brix. All bunches were checked to be free from any fungus symptoms and any other physical damage. The bunches of each vine were sampled into plastic bags and immediately kept on ice. Randomly, 50 berries were selected per vine, weighed and then macerated for 5 min in a blender. As demonstrated by Peppi et al. (2006), the filtered juice was used for soluble solids and TA measurements. An extra 10 berries were kept for dry mass measurement.

Viral status

Virus identification was carried out by reverse transcription PCR using previously designed primers, where the GLRaVs were characterized using hHSP70 (Osman and Rowhani 2006, Osman et al. 2007). The GVA virus was characterized by the capsid protein (CP; Minafra and Hadidi 1994). The 18S rRNA gene was used as an internal control (Minafra and Hadidi 1994). Five petioles, free from any symptoms of fungus, mold and other physical damage, were randomly sampled per vine at the berry pea-size stage (EL31) (Coombe and Iland 2005) in December 2016. RNA was extracted using the Spectrum Plant Total RNA Kit (Sigma-Aldrich, St Louis, MO, USA) according to the manufacturer’s instructions. DNA was removed by applying the ON-COLUMN DNASE I DIGESTION SET (Sigma-Aldrich, St Louis, MO, USA). The average concentration for the extracted RNA was 92–150 μg.μL⁻¹. The RNA was diluted to 2–5 μg.μL⁻¹, and cDNA synthesis was performed using the SuperScript IV VILO Master Mix (Invitrogen, Carlsbad, CA, USA) following the manufacturer’s instructions. Previously designed primers were selected for detecting the viruses. For GLRaV-3, the pairs Lc1/F and Lc2/R have been used (Osman and Rowhani 2006). For GLRaV-4 strain 9 and 5, the pairs LR9/F, R and LR5HSPC/F, R have been used, respectively (Osman et al. 2007). Those sets were targeting the hHSP70 gene in the leaf roll viruses. On the other hand, the CP gene was targeted in the detection of GVA virus using the primer pair GVAC1/F, R that was predicted by Minafra and Hadidi (1994). The 18S rRNA was used as an internal control (Gambino and Gribaudo 2006). The reaction mixture and the reaction parameters were modified after optimization as 5 μL GoTaq Green Mastermix (Promega, Madison, WI, USA), 0.5 μL of 10 mM of each primer, 1.5 μL nuclease-free water and 2.5 μL of the cDNA. The amplification steps were as follows: 2 min at 94 °C, 35 cycles of 94 °C for 30 s, 58 °C for 45 s and 72 °C for 60 s, and a final extension at 72 °C for 7 min. The electrophoresis was carried out by loading 10 μL of the amplification mix in a 1.5% agarose gel submerged in TAE buffer (40 mM Tris base, 20 mM sodium acetate, 1 mM EDTA pH 8.0). The amplified DNA fragments were visualized on a UV transilluminator following ethidium bromide staining and photographed. Positive and negative controls for the viruses under study were included in each experiment (Sambrook and Russell 2001).

Viral load

The rachis with pedicels attached (here after, ‘stalk’) were collected with the berries at the harvest stage (EL38, March 2017/18). The RNA was extracted as described for viral identity, with the exception that traces of DNA were removed using Dnase I Amplification Grade (Sigma-Aldrich, St Louis, MO, USA) according to the manufacturer’s instructions. The cDNA synthesis was carried out as described in the manufacturer’s instructions of SuperScript IV VILO Master Mix (Invitrogen, Carlsbad, CA, USA). Then, the cDNA was diluted to supply 10 ng in a 10 μL reaction volume to be subjected to the ABI 7500 Realtime PCR System (Applied Biosystems, Foster City, CA, USA) using PowerUp SYBR Green Master Mix and previously designed primers (Osman and Rowhani 2006, Osman et al. 2007, 2008). 18S rRNA was used as a housekeeping gene (Osman et al. 2008). The reaction was prepared according to the manufacturer’s instructions.

The positive controls were constructed as plasmid vectors and contained the specific region of hHSP70 for each of the leafroll viruses, CP for the GVA virus and 18S rRNA as a housekeeping gene. These regions are short sequences between 240 and 280 bp, which contain the primer sites. They were chosen from the virus reference genomes on the National Centre for Biotechnology Information website (https://www.ncbi.nlm.nih.gov/). The accession numbers are KY821094.1 (GLRaV-3), AY297819.1 (GLRaV-4/9), AF039552.1 (GLRaV-4/5), AY244516.1 (GVA) and AF321271.1 (18s rRNA). The designed vectors were aligned and synthesized by Integrated DNA Technology Australia (IDT) (NSW, Australia), providing the PUC IDT-AMP vectors and the vector construction process. All the vectors were supplied in 4 μg lyophilized powder, and the stock solution was prepared by adding 40 μL of nuclease-free water directly to the lyophilized powder to get 100 ng μL⁻¹. The working solution with a specific concentration (10 ng μL⁻¹) was prepared by taking 3 μL from the stock solution (100 ng μL⁻¹) and diluting it in 27 μL nuclease-free water. The copy number per 1 ng was determined by using Eq. (1):

(1)

The standard curves were prepared for all viruses using the working solution to produce 10-fold serial dilutions for 5–6 points in triplicates (102–107 copies). The concentration of each dilution was measured by Qubit 4 Fluorometric Quantification and RNA Quantification, broad range Kit (Thermo Fisher Scientific Australia Pty Ltd, VIC, Australia) and then detected by RTqPCR in two independent assays per virus. Standard curves for each virus were constructed by plotting Ct values versus the logarithm of the RNA copy number (Ct vs the log of the standard sample amount) using the StepOne Software (Applied Biosystems, Foster City, CA, USA).

Berry quality

Total soluble solids (TSS): the juice filtered through the muslin cloth into the conical flask to exclude berry flesh debris. Finally, TSS was determined using a digital refractometer calibrated in °Brix (g sucrose/100 g solution, Atago, PAL-1 Digital Brix Refractometer, Tokyo, Japan). Titratable acidity (TA): the juice (10 mL) was titrated against 0.1 M NaOH to an endpoint pH 8.2. Titratable acidity was calculated using the formula and expressed as a percentage of tartaric acid (Eq. (2)) (Iland et al. 2004, Considine and Frankish 2014):

(2)

°Brix/acid ratio

For each sample, the °Brix to acid ratio was calculated by dividing the °Brix value by the percentage acidity. Berry volume: the berry size was determined. Calipers were used to determine the vertical diameter (L) and horizontal diameter (l). The volume was determined by matching the berry form to an ellipsoid using the following equation: volume (cm³) = 4 abc/3, where a = b = l/2 and c = L/2 (Río Segade et al. 2011, Río Segade et al. 2013).

Dry weight and water content

All treatments are arranged into a set of three replicates for each sample. Each replicate contains 10 berries placed in a Petri dish. The initial weight of the samples and the Petri dishes were recorded. Then, the dishes were subjected to 57 °C for 21 days in a dry vacuum oven. The weight of all treatments was measured every day until no change in weight was noticed. The dry weight was calculated as illustrated in Eq. (3), while the total moisture in berries was calculated according to Eq. (4) (Nielsen 2017):

(3)

(4)

Total anthocyanin and phenol determination

The frozen berries were thawed at 4 °C for 2 h. The accurate weight of 50 berries per sample was recorded. The berries were transferred to a 50 mL plastic beaker for homogenization using an Ultra-Turrax T25 high-speed homogenizer with an S25N dispersing head (Janke and Kunkel GmbH Co., Germany). The beaker was placed on ice and homogenized at 24,000 r.p.m. for 30 s. The shaft was cleaned, and the remaining grape tissue was returned to the homogenization vessel again and homogenized for 15 s. One g of the homogenate was placed in a 15 mL tube. Ten milliliters of 50% (v/v) aqueous ethanol pH 2 was added. The tube was mixed by inversion for 1 h and centrifuged at 3500 r.p.m. for 5 min. The supernatant was collected, and the volume was measured. One milliliter of the extract was taken and mixed with 100 mL of 1 M HCl and mixed thoroughly. The diluted extract was incubated at room temperature for 3 h. The absorbance of the acidified diluted extract was measured at 520 nm using a 1.0 M HCl blank on a SPECTROstar Omega reader (BMG LABTECH, Ortenberg, Germany). The anthocyanin mg/berry was calculated (Eq. (5)) along with total phenolics per berry (absorbance units (au) per berry) as predicted in Eq. (6) (Iland et al. 2004):

(5)

(6)

Statistical analysis

The statistical analysis was performed using the ‘R’ language (https://cran.r-project.org/). A two-way multivariate analysis of variance (MANOVA) was conducted using the viral infection type and seasonal (year) influence as independent factors and the measured berry quality characteristics as dependent variables. Tukey post hoc contrasts were used to analyse the difference between the means of berry quality parameters and the clones. A simple linear regression was calculated to predict the relationship between the berry mass and sugar mass, TA and TSS. The analysis of covariance (ANCOVA) test was applied to analyse the viral load and season (year) against the berry quality parameters. The figures were plotted using the R package ‘ggpubr’ (https://CRAN.R-project.org/package=ggpubr).

Results

Viral identity

All the samples amplified an RT-PCR product of 844 bp, representing the internal control 18S rRNA (Figure 1). The control vines failed to amplify RT-PCR products using gene-specific primers for GLRaV-3, GLRaV-4 strain 9 and GVA, confirming the absence of detectable infection of these viruses. The clone 3236 vines tested positive for GLRaV-3, but not GLRaV-4 strain 9 or GVA. The clone 3215 vines consistently tested positive for GLRaV-3 and GVA, which confirmed the mixed infection. However, only four vines amplified bands with the GLRaV-4 strain 9-specific primers. Vines that did not amplify a GLRaV-4 product were retained in the experiment because they tested positive for GLRaV-3 and GVA. Previous research (Credi and Babini 1997) suggested that disease severity increased with mixed infections in grapevine, particularly when GVA co-infected with GLRaV-3.

Figure 1.

Agarose gel electrophoresis RT-PCR amplified cDNA products of GLRaV-3, GLRaV-4 Strain 9, GVA and 18s rRNA. L: ladder DNA Promega 100 bp with fragment size 100 to 12,000 bp. V1–V6 represent the vines replicates of each clone. Vertically: 18S rRNA represented the amplification of the positive control. GLRaV3 represents the amplification of HSP70 of the GLRaV-3 (546 bp). GVA represents the amplification of CP of the GVA (429 bp). GLRaV4/9 represents the amplification HSP70 of GLRaV-4 strain 9 (393 bp), in the control, clone 3236 and clone 3215.

Berry quality

Fruits on infected vines were lighter in color and larger than those growing on uninfected vines (Figure 2). Having established the identity of viral infections in the vines, berry quality parameters were quantified for 2 year (2017/18). A two-way MANOVA was performed with the viral infection type and the year effect as the independent variables and the quantified berry quality parameters as dependent variables. The viral infection types included three levels, namely viral free (control), single infection (GLRaV-3) and mixed infection (GLRaV-3, GLRaV-4 strain 9 and GVA), while the seasonal (year) effect consisted of two levels (2017 and 2018). Findings show significant differences for all dependent variables (Wilks λ, (2,33) = 0.033895, P < 0.001) for the viral infection type, while no significant difference was observed for the seasonal effect, with the exception of TA (Table 1).

Figure 2.

The cv. Crimson Seedless berries on the harvest day, showing the three types of viral infection: (a) shows the individual berries at the harvest stage; (b–d) show the corresponding berries on the vine. Control and clones are labeled.

Table 1.

The variation of the berry quality of the infected clones compares with the control. Means, standard deviations, Fisher value and the P-value were calculated using MANOVA, where each clone comprises six replicates (n = 6).

Clones									Seasons
	Control		Clone 3236		Clone 3215				2017		2018
Berry quality	M	SD	M	SD	M	SD	F		M	SD	M	SD	F
Mass (g)	5.83	0.59	8.70	0.42	8.56	0.35	175.23	***	7.74	1.22	7.66	1.62	0.31
°Brix	19.66	0.70	18.72	0.65	18.68	0.34	11.05	***	18.84	0.67	19.19	0.76	3.39	.
TA (%)	0.51	0.02	0.57	0.02	0.59	0.05	33.04	***	0.58	0.06	0.54	0.03	18.50	***
Dry mass (%)	32.73	1.30	28.70	1.19	28.06	0.78	62.24	***	29.95	2.30	29.71	2.47	0.41
H2O (%)	67.27	0.15	71.3	0.09	71.94	0.12	62.24	***	70.05	2.30	70.29	2.47	0.41
Anthocyanin/berry (mg)	1.72	0.19	0.78	0.20	0.58	0.26	86.22	***	1.06	0.56	1.00	0.55	0.44
Phenol/berry (AU)	1.15	0.27	1.01	0.31	0.82	0.25	4.54	*	1.02	0.32	0.97	0.29	0.30

M, mean; SD, standard deviation; F, Fisher value; AU, absorbance unit.

^*** P < 0.001,

^** P < 0.01,

^* P < 0.05, P < 0.1.

The mean of the berry mass was strongly dependent on the presence of the infection but was unaffected by year. Tukey post hoc contrasts showed significant differences between the control and single infection groups (P < 0.000001) and the control and the mixed infection (P < 0.0001), but not between the two infection groups (Figure 3a).

Figure 3.

The effect of viral infection on the berry quality among the infected clones and the control. Each clone comprised 6 replicates. (a) The effect viral infection on berry fresh weight, (b). The effect viral infection on moisture percentage, (c). The effect viral infection on berry dry weight (d). The effect viral infection on total soluble solids (e). The effect viral infection on sugar content of the berry (f). The effect viral infection on sugar content as a fraction of the dry weight %., one-way ANOVA has been calculated along with the plot using ggpubr “R” Package.

Depending on the moisture percentage, a significant increase in berry water accumulation can be noted in the infected clones compared with the control (Table 1, Figure 3b). The berry water weight in grams was calculated depending on the moisture percentage. A simple linear regression was calculated to predict the relationship between the berry mass and berry water mass among the control and the infected clones. The linear regression revealed a significant relationship (F(1, 34) = 3918, P < 0.000001), with a multiple R² of 0.9914. It indicates that the berry mass increases significantly in the infected clones depending on the increase in water accumulation in the berry. However, the relationship between the dry components of the berry and the berry mass needs to be clarified. Since the percentage of the dry mass shows a highly significant difference between the infected clones and the control (Table 1 and Figure 3c), the absolute dry berry mass was calculated. A linear regression was calculated to predict the relationship between berry mass and dry berry mass. A significant relationship was found (F(1, 34) = 268.6, P < 0.00000), with a significant multiple slope value (R² = 0.8867). However, the individual slope for each clone shows that the control had a considerably higher slope value (R² = 0.91) than infected clones, with clone 3236 and 3215 R² of 0.6 and 0.66, respectively, indicating that the increase in berry mass was due to ectopic accumulation of water rather than dry mass.

The °Brix value can be defined as the mass of sucrose per 100 g of juice, although it measures all soluble solids per 100 g of juice. The MANOVA analysis revealed a significant difference in the °Brix value between the infected clones and the control, where no significant effect of the year on the °Brix value was found (Table 1 and Figure 3d). Sucrose typically represents ~95% of the TSS in grape juice. Berry sugar was calculated using the method proposed by Vila et al. (2010), Eq. (7):

(7)

The ANOVA test was conducted to explain the effect of viral infection on the amount of sugar in the berries, showing a statistical significance (F(2, 3) = 4.170, P < 0.0001). The amount of sugar/berry was higher in both infected groups relative to the control, although the °Brix value was lower in the infected clones than the control, which could be related to the berry mass. ANCOVA confirmed a significant effect of infection type on the berry sugar content after controlling for berry mass (F(3, 32) = 332.9, P < 0.001, R² = 0.969). Figure 4a shows that the relationship is strongly positive and linear. However, previous studies showed that the relationship between the berry sugar content and the berry weight followed the second-order curve (Considine 2004). Therefore, forcing the line through the origin is reasonable, and as shown in Figure 4b, the differences in sugar/berry between the infected and control groups were related to the berry weight.

Figure 4.

The relationship between the berry sugar mass (g) and average berry mass (g). (a) ANCOVA test of the sugar/berry (g) tested against the berry mass for each clone and the control (n) Individual line regression for each infection type (b). The ANCOVA test show combined relationship by forcing all the lines through the origin point. F (1, 35) = 28,989.570, p < 0.001, with an R² of 0.999, n = 6 for each clone.

Sugar represents the major component of the TSS. Therefore, analysing the sugar as a fraction of the dry berry mass against the berry mass will clarify the picture of the viral effect on the sugar content. First, the sugar was calculated per berry dry weight as shown in Eq. (8):

(8)

Then the effect of the infection on sugar as a fraction of dry mass was assessed by ANOVA. The results showed a significant difference between the infected clones and the control (F(2, 32) = 21.4, P = 0.001; Figure 3f). The value of the sugar as a fraction of the dry mass was higher in the control compared with the infected clones. The result was confirmed by Tukey post hoc contrasts, which showed a significant difference between control versus clone 3236 (P < 0.001) and clone 32315 (P < 0.001), and no significant difference between the infected clones.

Table 1 shows that the difference in TA between the control and the infected clones was significant, whereby TA is lower in the control group than in the infected clones. Moreover, the year of production had a significant effect on the TA (Table 1, Figure 5a). Typically, the TSS (°Brix) and TA bear a negative relationship throughout the course of ripening. The relationship between TSS and TA provides an indication of fruit maturity and quality (Figure 5b). Interestingly, the two groups of virus infections show a noticeable difference in the regression slope of the TSS: TA (Figure 5c). The control shows an inverse relationship between TSS and TA, as expected y = 37 − 35x, R² = 0.55. Clone 3236 reveals almost similar trend to the control, y = 30 − 20x, R² = 0.42; however, the slope was less than the control. In contrast, clone 3215 shows a considerably weaker relationship where the slope was almost flat, y = 21 − 3.8x, R² = 0.032 (Figure 5c).

Figure 5.

The effect of the viral infection on the acidity (TA) of the grape berry and relationship to TSS between the infected clones and the control. (a) The effect on the TA as a percentage in the berry juice. (b) The effect of viral infection on the TSS:TA ratio. (c) The relationship between the °Brix value and the TA among the infection type group.

It is apparent from Figure 2 that virus infection influenced berry coloration. Analysis revealed a significant difference in the amount of anthocyanin among the infection groups (Table 1, Figure 6a). Tukey post hoc contrasts showed that anthocyanin content in clone 3236 and clone 3215 was significantly less than the control (P < 0.001), but there was no difference between the infected clones. A similar trend was evident in total phenols; however, the magnitude of the difference was less pronounced (Figure 6b).

Figure 6.

The effect of viral infection on anthocyanin and phenol levels between the infected clones and the control. (a) The berry anthocyanin content (mg) per berry. (b) The phenol content (mg) per berry. AU: absorbance Unit.

The relationship between viral load and berry quality

Viral load was determined for the four assayed viruses by RT-qPCR. As shown before, the berry mass and anthocyanin were the most affected characteristics by the viral infection. A linear regression of berry mass and the anthocyanin against GLRaV-3 load in both infected clones showed a linear relationship. However, the slope value was low for both years, suggesting a weak relationship (Figure 7). An ANCOVA test was applied using the year as the primary variate and the GLRaV-3 load for both infected clones as a covariate against the berry quality parameters (berry mass, °Brix, TA, dry mass and anthocyanin), to clarify the relationship between the viral load and the berry quality.

Figure 7.

Summary of the linear regression of the viral load. (a) The linear model of the berry mass (g) against the log10 of the copy number of GLRaV-3 for two years. (b) The linear model of the berry anthocyanin (mg) against the log10 of the copy number of GLRaV-3 for two years for the infected clones (clone 3236 and clone 3215).

The results showed no significant differences between the GLRaV-3 virus copy number and the measured berry quality parameters. A similar trend was observed for both GLRaV-4 strain 9 and GVA when the ANCOVA test was performed. Interestingly, the results were similar for the three viruses, as clearly indicated by the P-value shown in Table 2. Moreover, the results indicated by the R² and adjusted R², which represent the slope, were very low and mostly null. That refers to the null relationship between the virus copy number and the measured berry quality parameters. On the other hand, the year effect has a considerable impact on TA, as evidenced by the P-value of 0.001 for all viruses. Although the P-values in the three viruses were nearly identical, the slope values (R²) in GLRaV-3 were slightly lower when compared with GVA and GLRaV-4 strain 9, as shown in Table 2.

Table 2.

The summary of the ANCOVA test including a season as the primary variate and the GLRaV-3 load for both infected clones as a covariate against the berry quality parameters.

GLRaV-3
	Mass		°Brix		TA		Dry mass		Anthocyanin
	M	P	M	P	M	P	M	P	M	P
(Intercept)	8.48	<0.001	18.55	<0.001	0.61	<0.001	28.41	<0.001	0.66	<0.001
GLRaV-3	0	0.425	0	0.949	0	0.604	0	0.197	0	0.314
Seasons (2018)	0.24	0.136	0.3	0.17	−0.05	0.002***	−0.33	0.441	−0.02	0.844
Observations	24		24		24		24		24
R2 /R2 adjusted	0.115/0.030		0.091/0.005		0.371/0.311		0.118/0.034		0.054/-0.036
GLRaV-4 strain 9
	Mass		°Brix		TA		Dry mass		Anthocyanin
	M	P	M	P	M	P	M	P	M	P
(Intercept)	8.33	<0.001	18.78	<0.001	0.62	<0.001	28.64	<0.001	0.73	<0.001
GLRaV-4	0	0.375	0	0.132	0	0.263	0	0.067	0	0.124
Season (2018)	0.27	0.197	0.12	0.535	−0.07	0.007***	−0.32	0.436	−0.05	0.757
Observations	12		12		12		12		12
R2/R2 adjusted	0.247/0.079		0.254/0.088		0.597/0.508		0.365/0.224		0.253/0.086
GVA
	Mass		°Brix		TA		Dry mass		Anthocyanin
	M	P	M	P	M	P	M	P	M	P
(Intercept)	8.46	<0.001	18.43	<0.001	0.64	<0.001	27.72	<0.001	0.45	0.005
GVA	0	0.709	0	0.071	0	0.355	0	0.03	0	0.062
Season (2018)	0.24	0.327	0.3	0.167	−0.08	0.007***	0.15	0.731	0.1	0.54
Observations	12		12		12		12		12
R2/R2 adjusted	0.187/0.007		0.336/0.189		0.578/0.485		0.460/0.340		0.344/0.198

M, mean; P, P-value.

^*** P < 0.001,

Discussion

Berry quality is an essential standard in table grape production since it is served fresh to the consumers (Rolle et al. 2012). Berry ripening progresses in two growth stages consisting of berry formation and maturation, separated by a lag phase coinciding with véraison (Coombe and McCarthy 2000). Previous studies indicated that the pre-véraison berries from virus-infected vines do not show a significant difference in quality compared with virus-free vines (Alabi et al. 2016). In contrast, dramatic differences were observed during post-véraison, suggesting that viral infection caused more significant impacts on ripening-related processes starting from véraison (Alabi et al. 2016, Montero et al. 2016a, 2016b). As a result, this study concentrated on the final stage of growth, when the virus full effect could be clearly observed on the berries.

In wine grapes, smaller berries are desired, as the anthocyanin and sugar will be more concentrated in the small berry volume (Weaver and Winkler 1952, Nuzzo and Matthews 2005, Abu-Zahra 2010, Ferrer et al. 2014, Melo et al. 2015, Chen et al. 2018). However, large berries are more desirable in table grapes considering the berry density (Río Segade et al. 2013). An increased berry size is the most obvious indication of viral infection in the cv. Crimson Seedless vines (Singh Brar et al. 2008). The berry size, or fresh weight, is regulated by cell number, cell volume and the accumulation of organic substances (sugars) in the cell vacuoles (Coombe 1976, Robinson and Davies 2000, Ollat et al. 2002). Thus, the pulp cells enlarge as a result of an influx of sugars and water into the vacuole. The vacuoles of the pulp cells form ~99% of the cell volume (Diakou and Carde 2001). In the virus-free vines, pulp cells continued to enlarge throughout ripening, undergoing significant structural changes responsible for berry softening late in the harvesting stage (Barnavon et al. 2000, Nunan et al. 2001). In comparison, virus-infected vines frequently exhibit a delay in ripening, one of the most common symptoms of viral infection (Over de Linden and Chamberlain 1970, Martínez et al. 2016). A delay in ripening is associated with a prolonged influx of water and sugar into the berry from the leaves. The water and sugar will accumulate in the mesocarp cell vacuoles, leading to an increase in berry weight (Fontes et al. 2011, Bobeica et al. 2015, Rowhani et al. 2015). The increase in berry size was clearly observed in the present study (Figure 3a).

At the same time, the ripening of grape berries is accompanied by sugar accumulation; these processes play significant roles in the quality of the berries. Sugars are accumulated in the mesocarp vacuoles, which accounts for 65–91% of the fresh weight of a ripened berry (Marty 1999, Fontes et al. 2011, Pastore et al. 2011). However, the results show a reduction in the TSS that was measured as °Brix value in both infected vines; GLRaV-3 implied a remarkable decrease in TSS. It could be argued that the fact of sugar accumulation in the berry is controlled by a feedback inhibition mechanism, whereby sugar controls the expression of sugar transporter genes (Koch 1996, Lecourieux et al. 2014). The feedback mechanism keeps the sugar level balanced with the berry mass (Lemoine et al. 2013). As evident in Figure 4b, sugar was the major contributor to the increase in berry weight. However, the °Brix value was low in the infected berry as a result of the dilution of the sugar concentration by the increase in weight due to water.

Tartaric and malic acids are the main organic acids in the berry (Cholet et al. 2016, Kliewer et al. 1967, Lamikanra et al. 1995). Both acids increased rapidly during the pre-véraison stage, reaching their highest levels near véraison, and then declined throughout ripening (Muñoz-Robredo et al. 2011). However, the data show that the TA% was significantly higher in both infected clones, most notably in the mixed clone (Figure 5a), which is a common sign of GLRaV-3 infection (Kliewer and Lider 1976, Alabi et al. 2016). The high level of TA in the infected vines was associated with delayed ripening (Martínez et al. 2016). The decline of organic acids is controlled by many factors, such as enzymatic degradation (Lakso and Kliewer 1975, Sweetman et al. 2014, Batista-Silva et al. 2018) the dilution effect, the increase in temperature and the acid salt formation (Kliewer et al. 1967, Ruffner 1982). Among these, the acid salt formation might explain the delay in the decline of the organic acids in grape berries. Inorganic salts are transported from the root to the leaves through the xylem at the first stage of berry development. After véraison, the salts were transported from the leaves and unloaded into the berries along with sugar via the phloem. The salt content of the berry will decrease because the virus disrupted phloem transport (Jayasena and Cameron 2008, Kliewer 1966, Ford 2012). The relatively low salt content of virus-infected berries will prevent acid salt formation and keep the acidity higher in the infected berries. Interestingly, the seasonal (year) effect on both TA and pH was significant. Similar results were previously reported (Lee and Martin 2009, Lee et al. 2009), which may have resulted from the influence of different temperatures over the 2 years; however, this was not explored here (Lakso and Kliewer 1975).

Previously, two types of phenolic compounds found in berries were investigated: tannin and anthocyanin (Adams 2006). At the initial stage of berry formation, tannins form in the mesocarp, whereas anthocyanin accumulates in the skin following véraison (Kennedy et al. 2007). Our data demonstrate a significant decrease in anthocyanin concentration in infected clones compared with control vines, which was particularly pronounced in clone 3215 (mixed infected clones) (Figure 6a). Additionally, the reduction was extended to phenolic compounds and demonstrated the same mechanism, but the difference was less pronounced between the control and the infected clones (Figure 6b). Although sugar accumulation in skin is very low compared with the pulp, the evidence is growing for some control of polyphenol metabolism by sugars, possibly through crosstalk with genes responsible for sugar regulation (Matsushima et al. 1989, Solfanelli et al. 2006, Zheng et al. 2009, Smeekens et al. 2010, Dai et al. 2013). Previous studies show that sugars modulate anthocyanin biosynthesis pathway genes (Solfanelli et al. 2006, Lecourieux et al. 2014, Filippetti et al. 2015). Beyond the effect of delayed ripening, the viral infection seems to have a deleterious influence on anthocyanin synthesis, as indicated by the previously reported up- and down-regulation of several particular genes (Vega et al. 2011).

Moreover, research indicates that phloem-specific viruses disturb the leaf minor vein phloem transporting process (Naidu et al. 2015). Disrupting the phloem leads attenuates sugar transport from the source (leaves) to the sink (berries) (Pawar and Rana 2019), resulting in an accumulation of sugar in the leaf mesophyll cells, which influences anthocyanin production, leading to red coloration of interveinal areas of infected leaves. These findings are consistent with studies indicating increased levels of anthocyanins in symptomatic leaves of GLD-affected red-berried grapevines (Gutha et al. 2010). Although aspects of the host–virus interaction between GLRaVs and grapevine remain ambiguous, it appears the major influence of the virus on berry quality results from an altered source–sink balance, which, in particular, affects sugar transport in the phloem vessels after véraison (Naidu et al. 2015, Alabi et al. 2016). The data presented here are consistent with this interpretation. Moreover, it provides no evidence that viral load is important, but rather the virus proteins may play a role in reprogramming gene expression. Additionally, the findings reveal that clone 3215 exhibits a modest increase in berry quality variation compared with clone 3236. This might be explained as the combined effect of mixed viral infections in clone 3215. However, no statistically significant evidence was found between both clone infections, which may suggest that, of the viruses studied, GLRaV-3 had the prevailing influence on berry qualities.

Authors' contributions

W.S. carried out the work and wrote the first draft under the guidance of M.J.C., who conceived the study and design and J.A.C., who supervised berry quality analysis and statistics.

Conflict of interest

None declared.

Funding

W.S. was supported by an Iraq Government PhD Scholarship. The research was supported by the Australian Research Council (LP130100347, FT180100409), The Department of Primary Industries and Regional Development (WA) and the Western Australia Vine Improvement Association (WAVIA).

Data availability

Data can be made available upon request from the corresponding author.