Pathogenic CGMMV strains MC-1 and MC-2 were isolated from leaves of cucumber plants during the outbreak of the infection in greenhouses in the Moscow region. The pathogenic CGMMV MC-1 strain produced green mosaic and mottling on cucumber leaves, while the pathogenic strain CGMMV MC-2 in addition to green mosaic and mottling, caused distortion and blistering.

The attenuated CGMMV strain VIROG-43M was obtained after a number of passages of the initial VIROG- 43 isolate; it represents a more stable analogue of the VIROG-43 strain. VIROG-43 was obtained from the pathogenic CGMMV strain by chemical mutagenesis using nitrous acid as a mutagen followed by selection of attenuated strains. The attenuated VIROG-43M strain was selected by a number of passages of the diluted virus VIROG-43 on plants with local necrotic (Chenopodium amaranticolor) and systemic (Cucumis sativus) reaction to infection. The presence of the attenuated virus in symptomless plants was controlled by DAS-ELISA. Finally, the virus was isolated from cucumber leaves and propagated on the Kurazh F1 variety ( Gavrish , Russian Federation ). Plants were inoculated with viral particles and grown for 3 weeks in a greenhouse under controlled temperature conditions (25˚C day/18˚C night). Viral particles were isolated and purified from infected cucumber leaves by stepwise precipitation with ammonium sulfate [7] .

Viral RNA was isolated by RNA adsorption on silica gel in the presence of guanidine thiocyanate (GTC) [7] . An ammonium sulfate precipitate of viral particles was dissolved in 2.5 M GTC solution, containing 24% ethanol. High-molecular-weight RNA was adsorbed on glass fiber filters (GF/F, Whatman) and eluted with RNase-free water. The first cDNA strand was synthesized with RevertAid first strand cDNA synthesis kit ( Fermentas , Lithuania ) and random hexamer primers following the manufacturer’s instructions. The second strand was synthesized, and the fragments were amplified using Taq DNA polymerase and specific primers. To design primers, we preliminarily analyzed nucleotide sequences of CGMMV isolates from the NCBI GenBank. Viral genomic RNA was sequenced at least two times in both directions from overlapping PCR fragments. If any variation in the primary structure of the studied strains occurred, the regions were sequenced more than three times. Amplified fragments were eluted from agarose gel with a Diatom DNA elution kit ( Laboratoriya IzoGen , Russia ) and sequenced using an ABI PRISM BigDye Terminator v. 3.1 kit. The reaction products were analyzed on an ABI PRISM 3100-Avant automated DNA sequencer. The sequence data for MC-1, MC-2 and VIROG-43M are available under NCBI GenBank accession numbers FJ848666, GQ495274, and GQ495275, respectively.

The phenotypic stability of the VIROG-43M strain was assayed under laboratory greenhouse conditions. The experiment was repeated three times. In each replication, 33 cucumber plants were inoculated with the pathogenic MC-1 strain and 66 plants with the attenuated VIROG-43M. 0.5 µg/ml of each viral strain was used for inoculation. After two weeks, one half of plants initially inoculated with VIROG-43M were challenged with the pathogenic MC-1 strain. The number of symptomless plants was determined 1.0, 1.5, 2.0 and 2.5 months after first inoculation.

The degree of cross-protection was measured on the Kurazh F1 variety ( Gavrish , Russian Federation ) of cucumber plants grown in commercial greenhouses. For this experiment, stock solution of the attenuated VIROG- 43M strain was diluted with water, supplied with carborundum and sprayed on 10 - 12-day-old cucumber seedlings at 1 to 2 fully-expanded-leaf stage. One hectare of inoculated with the VIROG-43M strain and one hectare of non-inoculated control cucumber plants were analyzed throughout 3 months of vegetation during summer and winter crop rotations. The same trial was conducted in commercial greenhouses located in Leningrad, Krasnodar and Moscow regions with three other cucumber varieties: Tristan F1 (Enza Zaden, Holland), Rais F1 (Gavrish, Russia), and Mirabelle F1 (Seminis, Holland). The results are given in the corresponding section and Table 1 and Table 2.

Cucumber plants of the Kurazh F1 variety (Gavrish, Russian Federation) were grown in laboratory greenhouse under controlled temperature and light conditions (25˚C day/18˚C night). When plants reached the stage of three fully expanded leaves, two bottom leaves were inoculated with 0.5 µg/ml of each viral strain. On the 3^rd, 7^th, 10^th and 14^th days post inoculation (d.p.i.), plant leaf discs were sampled and subjected to DAS-ELISA. For each sample (each CGMMV strain), ten plants were grown and inoculated with the corresponding strain. On the 3^rd d.p.i., samples for DAS-ELISA were taken from six similarly developed plants. At one time point, two upper non-inoculated leaves were analyzed; from each leaf one leaf disc was taken. The experiment was repeated three times. ADGEN Identikit CGMMV ( Neogen , Scotland , UK ) was used for DAS-ELISA according to the

n/a, not available.

manufacturer’s manual. Reactions were read on an MR-96A Microplate Reader ( Mindray , China ) at 405 nm. For negative control, leaf discs from healthy plants were sampled at each designated time point. Systemically infected with a pathogenic strain and homogenized cucumber leaves supplied by ADGEN were used as positive control.

The phylogenetic tree was built with MEGA6 software [18] on the base of CLUSTAL W alignment of CP nucleotide sequences. The evolutionary history was inferred by the Maximum Likelihood method based on the Tamura-Nei model [19] . Initial tree for the heuristic search was obtained by applying the Neighbor-Joining method to a matrix of pairwise distances estimated using the Maximum Composite Likelihood (MCL) approach. The tree was drawn to scale, with branch lengths measured in the number of substitutions per site. All positions with less than 95% site coverage were eliminated. That was, fewer than 5% alignment gaps, missing data, and ambiguous bases were allowed at any position. There were a total of 320 positions in the final dataset.

The procedure for VIROG-43M isolation was described in Material and Methods. The efficiency of the strain as a biocontrol agent was tested on the Kurazh F1 cucumber variety ( Gavrish , Russian Federation ) in laboratory and commercial greenhouses. Experiments in laboratory greenhouse allowed us to determine phenotypic stability of the VIROG-43M strain, its interfering capability against pathogenic strains, and cross-protection level. We discovered that VIROG-43M exhibited high phenotypic stability since 100% of cucumber plants inoculated with this attenuated strain showed no visible symptoms throughout two months of observation (Table 1). However, by the 3^rd month of vegetation, the number of symptomless plants decreased to 98%, most likely due to mixed infection in 2% of plants. VIROG-43M also exhibited significant interfering capability, since throughout two months of vegetation, 95% of cucumber plants inoculated initially with VIROG-43M showed no disease symptoms. By the end of vegetation, the number of symptomless plants reduced, but still was rather high (85%) in comparison with 0% of symptomless plants inoculated initially with the pathogenic MC-1 strain (Table 1).

VIROG-43M was also tested in a number of commercial greenhouses on different cucumber varieties. The level of cross-protection was extensively studied on the Kurazh F1 variety ( Gavrish , Russia ) in a commercial greenhouse located in the Moscow region during summer and winter rotations. The results are shown in Table 2. It is clearly seen from this table that at the end of vegetation in winter rotation only 1.5% of plants treated with VIROG-43M exhibited mosaic symptoms, while in control plants, this number increased to 15%. In summer rotation, the observed indicators increased to 3.2% for inoculated plants and to 34% for non-inoculated ones. We also tested other cucumber varieties grown in commercial greenhouses located in diverse regions of Russia (Table 2). We found that by the end of winter rotation, Tristan F1 (Enza Zaden, Holland) and Marabelle F1 (Seminis, Holland) varieties treated with VIROG-43M exhibited mosaic symptoms only in 0.5% and 1.5% cases, in contrast to 15% and 13% in untreated plants. The Rais F1 variety ( Gavrish , Russia ) treated with VIROG-43M during summer rotation showed mosaic symptoms in 3.0% cases, in contrast to 19% in untreated plants. It should be also noted that using VIROG-43M during winter rotation, yield gain reached 19.8% on Kurazh F1 ( Gavrish , Russia ) and 16.5% on Mirabelle F1 (Seminis, Holland ) varieties in comparison to untreated cucumber plants. The obtained results show that VIROG-43M can be applied in greenhouses for cucumber plant protection for two main reasons: this attenuated strain is phenotypically stable and shows high level of cross-protection against pathogenic strains resulting in significant yield gain.

The RNA genomes of MC-1, MC-2, and VIROG-43M CGMMV strains consist of 6422 nucleotides and have the overall structure similar to other tobamoviruses: the first and the second ORFs encode the short replicase (129-kDa) and the long replicase (186-kDa) proteins; the third and the fourth ORFs encode the 29-kDa movement protein (MP) and the 17.4-kDa CP, respectively (Figure 1).

Comparison of the genomes of MC-1 and MC-2 pathogenic strains showed that they differed in 8 nucleotides in the replicase gene and one mismatch in the MP gene. Only the C-2152 - T-2152 mismatch led to amino acid substitution P-698 → L in MC-2 strain in comparison to MC-1 strain (Figure 1). Comparison of the genomes of the pathogenic MC-1 and attenuated VIROG-43M strains revealed six mutations including two silent mutations located in the replicase and MP genes and four missense mutations in the replicase gene (Figure 1). Comparison of the nucleotide sequences of the pathogenic MC-2 and attenuated VIROG-43M strains revealed nine mutations in the replicase gene, six of which were silent and three were missense. It is interesting that these three missense mutations led to the same amino acid substitutions as those observed in the genome of the MC-1 strain (G-86 → S; S-534 → F; P-1362 → L). However, one MC-1-specific mutation leading to the P-698 → L substitution was absent from the MC-2 genome (Figure 1).

For VIROG-43M, MC-1 and MC-2, the accumulation dynamics in plants was determined by DAS-ELISA. Cucumber plants were grown, inoculated and sampled as described in Materials and Methods. The results of the experiment are presented in Figure 2. It follows from these results that on the 3^rd d.p.i., the attenuated and pathogenic strains accumulated in plants in trace amounts, but even at this stage, the mean value for each strain already differed. From the 3^rd till the 7^th d.p.i., a significant increase in virion CP for all viral strains was detected; however, VIROG-43M accumulated in significantly less amounts than the pathogenic strains. The maximum rate of virus accumulation was observed from the 7^th till the 10^th d.p.i., when mean CP levels for all CGMMV strains reached a plateau. Moreover, on cucumber plants inoculated with the pathogenic strains, the first mosaic symptoms could be clearly observed. On the 14^th d.p.i., a little decrease in CP level was recorded for MC-1 and VIROG-43M strains, although this value increased for MC-2 strain. Thus, from the 3^rd till the 14^th d.p.i., the attenuated VIROG-43M accumulated in plants in significantly less amounts than the pathogenic strains; the level of pathogenic strain accumulation differed from the 3^rd till the 10^th d.p.i. and by the 14^th d.p.i., the difference between both pathogenic strains was insignificant.

As mentioned above, CGMMV strains cause diseases in Cucurbitaceae plants in different countries. By analyzing NCBI GenBank data we retrieved 98 strains of CGMMV originated from 17 countries. The largest number of CGMMV strains (as great as 34) were described in China , next comes South Korea (13 isolates), followed by India (9 isolates) and Japan (7 isolates). In other countries less than 6 strains were characterized.

We performed phylogenetic analysis with all CGMMV strains retrieved from the NCBI GenBank. Note that

nucleotide sequences of the coat protein (CP) genes were used for phylogenetic tree construction since this particular gene was sequenced in most strains deposited to GenBank. Eighty-three of 98 CGMMV strains were taken for analysis. The remaining 15 strains either did not have the nucleotide sequence of the CP gene or the CP gene was not completely sequenced. The phylogenetic tree built using MEGA software [18] is shown in Figure 3. It is evident that all CGMMV strains group into two main clades: the small clade with 8 representatives including Moscow region, Russian strains (the clade is indicated with a bold line in Figure 3) and the large clade with the remaining 75 representatives. Some subclades of the large clade are entirely Asian groups including CGMMV isolates from China, Taiwan, Japan, South Korea and Indonesia (the clade is indicated by bold and light grey line in Figure 3). Other clades represent combined groups with Asian, European, Canadian and Australian CGMMV isolates.