Comparative study of diagnostic methods used for monitoring of common grape vine (Vitis vinifera L.) crown gall (Agrobacterium vitis Ophel & Kerr) in Slovenia

Agrobacterium vitis causes common grape vine (Vitis vinifera L.) crown gall disease that destroyed a lot of Slovenian vineyards more than a decade ago. Eighty isolates of Agrobacterium spp. collected during monitoring in 2006 were identified as A. vitis and A. tumefacies by pehA and multiplex PCR method. Tumor-inducing capacity of these strains was assessed on test plants and with PCR methods for detection of the Ti plasmid responsible for tumor induction. With VCF3/VCR3 primer pair six false negatives and no false positives were detected. The high genetic diversity of pathogenic Agrobacterium spp. strains affects the performance of molecular methods, thus biological test should be performed where results from molecular methods are doubtful.


INTRODUCTION
Crown gall disease occurs worldwide and causes major economical losses in fruit and grapevine production (De Cleene and De Ley, 1976;Kennedy and Alcorn, 1980;Pulawska, 2010).The major part of income loss is attributed to crown gall on young grafted plants in nurseries.The disease is characterized by a tumor which is usually formed on a plant stem just above the ground.Still, the disease is rarely fatal.Mainly young or stressed plants develop more pronounced symptoms: loss of plant vigour, reduction in crop yield, or even plant death (Poncet et al., 1996;Epstein et al., 2008).The disease is problematic on perennial horticultural crops, such as grapevines, stone and pome fruit trees, and ornamental plants, where tumors weaken the plant year after year.The causal agents of the disease are pathogenic Agrobacterium spp.carrying the Ti plasmid (pTi) (Van Larebeke et al., 1974;Watson et al., 1975).The ability to cause tumors is encoded on a portion of the pTi (T-DNA) that integrates into the host genome.Upon expression, the T-DNA genes alter the level of plant hormones resulting in uncontrolled plant cell proliferation and tumor formation (reviewed in Escobar and Dandekar, 2003).
Traditional identification of Agrobacterium spp. is based on biochemical tests (Holt et al., 1994).The INCO-DC European program ERBIC18CT970198, "Integrated Control of Crown Gall in Mediterranean Countries" has presented an identification scheme for agrobacteria with minimal biochemical tests (reviewed in Shams et al., 2012).Additionally, accurate identification can be achieved by molecular methods.Eastwell et al. (1995) developed a PCR method for detecting A. vitis (Ophel and Kerr, 1990) -causative agent of crown gall of grapevines.The method targets chromosomal polygalacturonase gene that is found in A. vitis, but not in A. tumefaciens (Smith & Townsend, 1907) (Young et al., 2006).On the other hand, A. tumefaciens and A. vitis were reported to differ from the members of the genus Rhizobium and therefore can remain in the same genus (Farrand et al., 2003, Lindström andYoung, 2011).Additionally, genetically variable strains of A. tumefaciens, now termed A. tumefaciens species complex group, were clustered into genomospecies that will progressively be reclassified into new species (Mougel et al., 2002¸ Portier et al., 2006;Lindström and Young, 2011;Pulawska and Kalužna, 2012).
Effective detection of tumor-inducing agrobacteria in plant material is crucial in propagating material and efficient management of crown gall disease.
Traditionally, bacteria are isolated from plant or soil material by cultivation on selective media followed by testing their tumor-inducing capacity on test plants.According to Schroth et al. (1971), this protocol is not sensitive or robust enough in comparison to molecular methods.Most molecular methods for detection of tumorigenic isolates target tumorigenicity genes on a conserved vir region on the pTi.Sawada et al. (1995) developed VCF/VCR primers that target pTi-encoded virC1 and virC2 genes.Suzaki et al. (2004) improved the specificity of the primers (VCF3/VCR3) for pathogenic Agrobacterium strains from apple seedlings, but the primers work on A. vitis strains as well (Kumagai and Fabritius, 2008).
Grape crown gall caused substantial damage to vineyards in the winegrowing regions of Slovenia in 1999 (Šabec-Paradiž et al., 2002).Much of the following Agrobacterium-based research in Slovenia was dedicated to control and prevention of A. vitis and A. tumefaciens infections on grapevine plants and propagating material, and also to characterization of A. vitis isolates in Slovenia (Fabjančič and Milevoj, 2003).In the present study we compared the Agrobacterium identification methods used for grape crown gall disease monitoring in Slovenia.

Bacterial strains and isolates
A crown gall monitoring was conducted in 2006.
Eighty-seven symptomatic grapevine grafts (Figure 1) were collected from nurseries and vineyards from across various winegrowing regions of Slovenia.Eighty isolates of Agrobacterium spp.
were obtained from plant material on 3DG medium semi-selective for A. vitis (Brisbane and Kerr, 1983).All 80 strains were subcultured on King's B medium (KB), pure cultures preserved in meat peptone broth with glycerol, and stored at -80 °C until further use.All 80 strains were analysed in the diagnostic laboratory at Agricultural Institute of Slovenia.

Preparation of bacterial DNA
The bacterial DNA used in PCR reactions was extracted from 24 h-old colonies grown on KB medium at 27 °C.We used a standard alkaline lysis method (Sambrook et al., 1989), diluted the DNA (1:1000) in sterile distilled water and stored it at -20 °C.
Repeatability of the PCR results on the same DNA samples was verified in 2013.

Polygalacturonase gene amplification
The pehA PCR was performed in a total volume of 25 µl applying the protocol of Eastwell et al. (1995).For the PCR reaction, 1 µl of bacterial DNA template was used for PCR amplification in 1× PCR Buffer (Promega), 2.0 mM MgCl 2 , 0.1 µM each pehA primer (Table 1), 0.2 mM dNTPs, 0.25 U GoTaq Flexi DNA Polymerase (Promega).The thermal cycler was programmed for an initial denaturation at 95 °C for 3 min followed by 40 cycles of amplification (95 °C for 1 min, 55 °C for 1 min, 72 °C for 1.5 min) with 5 min of final elongation at 72 °C.The amplified fragments of 205 bp were visualized on 2 % agarose gel.

Multiplex PCR
The multiplex PCR was performed in a 15 µl reaction volume applying the protocol of Pulawska et al. (2006).All reactions were performed in 1× PCR buffer (Promega), 1.5 mM MgCl 2 , 1 µM each primer (UF, B1R, B2R and AvR) (Table 1), 0.2 mM dNTPs and 1.0 U GoTaq Flexi DNA Polymerase (Promega).The amplification conditions comprised an initial denaturation at 95 °C for 1 min, followed by 35 cycles of denaturation at 94 °C for 1 min, annealing at 67 °C for 1 min, extension at 72 °C for 1.5 min and a final extension step at 72 °C for 10 min.The amplified PCR fragments were visualized on a 2 % agarose gel.Strains belonging to A. tumefaciens gave a 184 bp product and those belonging to A. vitis gave a 478 bp product (Figure 2).

Assessing tumor-inducing capacity
Diagnosis of pathogenic strains of Agrobacterium spp. is carried out biologicaly on wounded test plants and molecularly through detection of bacterial tumour-inducing plasmid (pTi) responsible for uncontrolled plant cell growth.

Pathogenicity tests
The pathogenicity of Agrobacterium strains was determined on tomato, sunflower and kalanchoe plants.Young, four-week-old seedlings were punctured three times in the stem using a sterile entomological needle dipped in pure culture colonies grown on KB medium for 24 hours at 27 o C. Tests were performed in triplicates.Inoculated seedlings were maintained in a glasshouse at 20 -30 °C with natural lighting conditions.In the period of 3 to 6 weeks post inoculation, the plants were visually inspected for tumor formations every few days.The strains C58 and 339-26 were used as positive controls; strain K84 and water served as negative controls.The test was completed in 2007.

Data analysis
Agreement between PCR and pathogenicity test was evaluated by calculating positive and negative percent agreement with respect to imperfect reference standard, in our case the pehA method and the biological pathogenicity test.The agreement indices were calculated from twodimensional contingency table shown in Table 2. Positive percent agreement with respect to imperfect reference standard was calculated according to equation ( 1) and was interpreted as sensitivity (Se) of the method.Similarly, negative percent agreement with respect to imperfect standard was calculated according to equation ( 2) and was interpreted as specificity (Sp) of the method.For estimation of confidence limits the 95 % confidence interval (CI) was calculated where appropriate. (1) (2)

Identification of A. vitis by pehA or multiplex PCR method
Morphologically, most rhizobial colonies appeared similar to one another on a general media.Therefore, it is imperative to use selective media for isolation of A. vitis.On 3DG medium, A. vitis colonies were visible sooner (after 3 days at 27 °C) than colonies of A. tumefaciens and R. rhizogenes, which also had different colony morphologies on 3DG medium.Where no typical A. vitis colonies were found, we selected for colonies that predominated on 3DG medium.
From 87 grapevine grafts we obtained 80 Agrobacterium isolates.According to multiplex PCR (Pulawska et al., 2006) 75 isolates were identified as A. vitis and five as A. tumefaciens.
The number of identified A. vitis strains was compared to the number of pehA positive (A.vitis) strains.There was a perfect agreement (100 %) between the two methods.All pehA positive isolates had an A.vitis-diagnostic band of 478 bp in multiplex PCR.The results were verified in 2013 on the same DNA samples stored at -20 °C.
Our diagnostic laboratory has completely replaced the pehA identification method with multiplex PCR as it distinguishes between A. vitis and A. tumefaciens and differentiates them from other rhizobia in the Rhizobiaceae family in one reaction.

Pathogenicity status and pTi detection
All 80 isolates were tested for pathogenicity and abilitiy to cause tumors on stems of inoculated plants.This is a standard method for diagnosis of tumor-inducing strains and detection of latent infections (Janse, 2005).Almost 70 % of the strains were found pathogenic.One pathogenic strain was identified as A. tumefaciens causing tumors on all three test plants.In 2013 we analysed the same strains for the presence of pTi.We used a PCR method for identification of pathogenic and non-pathogenic strains of agrobacteria using primers VCF3/VCR3 with improved specificity (Suzuki et al., 2004).All strains with detected pTi were identified as A. vitis by multiplex PCR.The agreement between results from pathogenicity tests and pTi detection method was not exact (Table 3).
The sensitivity of VCF3/VCR3 primer pair was 89.1 % with six false negatives (Table 4).The only pathogenic A. tumefaciens strain was one of them.However, the specificity was 100 % with no false positives.The VCF3/VCR3 results were most compatible with pathogenicity assessment on kalanchoe and tomato test plants, though specificity was higher on kalanchoe plants (Table 4).It is not uncommon for A. vitis strains to have different host range even with identical physiological and biochemical characteristics (Tolba and Zaki, 2011).Interestingly, results from Tolba and Zaki (2011) indicate tomato as an unreliable test plant giving positive results on pathogenic A. vitis strains in only 5 of 12 isolates.At the same time, test on tomato proved specific for certain strains that caused tumors only on grapevines.In our case, none of the grapevine strains caused tumors specifically on tomato.On the other hand, six strains caused tumors solely on sunflower, and two solely on kalanchoe test plants.
The presence of pTi was diagnosed on only half of these strains (three and one).One possible explanation is the sensitivity of the primers.These might be affected by high genetic diversity within pathogenic agrobacteria which could result in false negatives.The use of a set of three plants proved crucial in pathogenicity determination, as few of the pathogenic strains exhibited preference toward one host plant.If we had used only tomato or sunflower test plants, we would have observed fewer pathogenic strains and determined lower specificity of the VCF3/VCR3 PCR method in comparison to pathogenicity test results (Table 4).The biological pathogenicity test is laborious and time-consuming.Pathogenicity is affected by environmental factors like temperature (Hamilton and Fall, 1971) and plant age (Binns and Thomashow, 1988).Also, the absence of tumors does not necessarily imply the absence of pTi.This is where molecular methods provide an additional confirmation.We repeated pathogenicity tests three times in three different seasons (spring, summer and autumn) using young to mature plants (results not shown).The most consistent results were obtained on young plants in late spring and early autumn when the average air temperature in the greenhouse was around 25 °C.Also, the interpretation can be doubtful when the PCR shows the absence of pTi, but the test plants develop tumors.Although PCR techniques for simultaneous identification of pathogenic and nonpathogenic A. vitis are available (Kawaguchi et al., 2005), the traditional pathogenicity test is still a standard technique in strain pathogenicity determination.

CONCLUSION
Pathogenic A. vitis strains predominated among isolates of A. vitis from Slovenian grapevine grafts.Only one strain of A. tumefaciens was found pathogenic.In identification of A. vitis we obtained matching results using pehA or multiplex PCR primers.Therefore, we suggest using multiplex PCR (Pulawska et al., 2006) for reliable identification of A. vitis and A. tumefaciens on grapevine.Further, we detected most of the pathogenic strains with VCF3/VCR3 primers.Based on our results, one might conclude that VCF3/VCR3 PCR could replace pathogenicity tests, but due to the false negatives, we conclude that biological pathogenicity test is still an invaluable tool in plant bacteriology.

Figure 1 :
Figure 1: Grapevine grafts showing crown gall symptoms on a heel (A) and on a graft union (B) (Photos: I. Zidarič).

Table 1 :
Primers pair sequences used in our study.
Conn 1942 or A. rhizogenes (Riker et al. 1930) Conn 1942, which are rarely found in grapevine tumors.A decade later,

Table 2 :
Two-dimensional contingency table for calculating agreement indices between two methods.

Table 3 :
Summary of results by two methods for Agrobacterium spp.pathogenicity assessment.

Table 4 :
Sensitivity (Se) and specificity (Sp) of VCF3/VCR3 primer pair with respect to pathogenicity test on test plants.