Bacterial halo blight of coffee crop: aggressiveness and genetic diversity of strains

: Bacterial halo blight, caused by Pseudomonas syringae pv. garcae , is an important disease of coffee crop occurring in Brazil and other countries. In recent years, outbreaks of this disease have damaged several coffee crops in Brazil. Aggressiveness and genetic diversity of 25 strains of P. s . pv. garcae , obtained between the years 1958 and 2011, in 23 cities of São Paulo and Minas Gerais states, as well as three strains from Kenya were evaluated in this study. The strains were inoculated on coffee seedlings cultivar Mundo Novo, and their genetic diversity was evaluated by ERIC-PCR, REP-PCR, and their combination. All the strains were pathogenic to the coffee seedlings; the results of pathogenicity tests, in both experiments, could be divided in four aggressiviness classes (highly aggressive; aggressive; moderately aggressive and less aggressive). The Kenyan


INTROduCTION
Brazil is the world´s largest producer and exporter of coffee. The Arabica coffee (Coffea arabica L.) is cultivated mainly in Minas Gerais, and São Paulo States (ConAB 2015). Bacterial halo blight, caused by Pseudomonas syringae pv. garcae, is an important disease of coffee in Brazil and outbreaks of this disease have occurred in this country in recent years (Rodrigues et al. 2013;Zoccoli et al. 2011).
The disease is characterized by lesions on leaves, flowers, pin-head berries, and die-back of twigs and branches. The most characteristic symptom of the disease is the presence of brown necrotic lesions on the leaves, surrounded by a chlorotic halo. Growth and coalescence of the lesions on the leaves cause their drop and subsequent defoliation of branches. At the end of the rainy season, the symptoms are confined to the branches that exhibit sometimes severe die-back. Bacterial halo blight also affects seedlings in nurseries, causing lesions on leaves and die-back of the seedlings (Costa et al. 1957).
This disease was first described in 1955, affecting coffee crops in the county of Garça, State of São Paulo, Brazil (Amaral et al. 1956) and was considered of minor importance for approximately 17 years after its emergence, when only isolated cases were reported (Kimura et al. 1973). From 1973 to 1975, a high incidence of the disease was observed in the state of Paraná in nurseries and coffee plantations that were recovered from frost 4 (Kimura et al. 1973;Mohan 1976). Bacterial halo blight has spread across the country and the pathogen has been detected in the States of Paraná, São Paulo and Minas Gerais (Malavolta Junior et al. 2008). Recently, damages caused by the disease were reported in some regions of the State of São Paulo as well as in regions with irregular topography and high altitudes in the State of Minas Gerais, such as Alto Paranaíba, Triângulo Mineiro and Sul de Minas (Zoccoli et al. 2011;Rodrigues et al. 2013).
The rep-PCR technique has been considered an important tool in phylogenetic studies of bacterial populations and shows high correlation with other more laborious techniques such as AFLP and DnA hybridization (Rademaker et al. 2000). This technique is based on the amplification of three families of repetitive DnA sequences that are present in the genome of several bacterial species, the repetitive extragenic palindromic sequence (REP), the enterobacterial repetitive intergenic consensus (ERIC) and the BoX elements (Louws et al. 1994). Rep-PCR has been frequently used to assess the genetic diversity of several plant pathogenic bacteria such as Xanthomonas axonopodis pv. juglandis (Scortichini et al. 2001(Scortichini et al. ), P. syringae pv. syringae (natalini et al. 2006) and studies of inter-and intraspecificity in strains of P. syringae (Vicente and Roberts 2007;Çepni and Gürel 2012;Gašić et al. 2012).
Although bacterial halo blight occurs in Brazil since 1955, it is unknown if the recent outbreaks of the disease are due to changes in the population of the pathogen or to increases in the aggressiveness of new strains or to other unknown factors. Therefore, aggressiveness and genetic diversity of 25 Brazilian strains of P. syringae. pv. garcae from several coffee producing areas, obtained between the years 1958 to 2011, were investigated in this study.

MATERIAL ANd METhOdS
Twenty-five strains of P. syringae. pv. garcae, isolated from coffee crops cultivated in the States of Minas Gerais and São Paulo, Brazil, were investigated in this study. Three strains from Kenya were included in the analysis for comparative purposes, since bacterial halo blight was also reported as an important disease in that country (Kairu 1997). The strains were obtained from the Phytobacteria Culture Collection of Instituto Biológico (IBSBF) and their geographical origin, year of isolation, and climate data of the localities of origin are shown in Table 1.
The aggressiveness of the strains was evaluated in two experiments carried out in September and october of 2015, in a greenhouse located at Centro Experimental Central, Instituto Biológico, Campinas, SP, with seedlings of the susceptible coffee cultivar Mundo novo. Four pairs of true leaves of each seedling were inoculated with bacterial suspensions (approximately 10 8 CFU·mL -1 ) prepared by the cultivation of the strains in nutrient agar medium (nA) for 48 h at 28 °C. The inoculations were performed by spraying the suspensions until runoff on leaves that were previously punctured with sterile entomological needles (four punctures per leaf ). Three control plants were inoculated with sterile distilled water. After the inoculations, the plants were kept in a moist chamber for 72 h and then transferred to a greenhouse at room temperature. The severity of the disease was assessed by the number of lesions and the area affected by the disease in the second and third pairs of leaves of the seedlings at 30 days after the inoculation. To evaluate the area affected by the disease the leaves were photographed and, in order to be measured in the interactive application Leaf Doctor (Pethybridge and nelson 2015), their photographs were modified using Adobe Photoshop to create a black background that surrounded all the individual leaves. The interactive application Leaf Doctor was used to distinguish diseased from healthy plant tissues and to calculate the percentage of disease severity (Pethybridge and nelson 2015).
Two independent experiments were carried out in a completely randomized design with four replicates, each replicate represented by one seedling. The results were submitted to analysis of variance and the averages were compared by the Scott Knott test at the level of 5% of probability.
In the molecular characterization, the amplification reactions were conducted with primer sets: ERIC1R (5´ -ATG TAA GCT CCT GGG GAT TCA C -3´); ERIC2 (5´ -AAG TAA GTG ACT GGG GTG AGC G -3´); REP1R-I (5´ -III ICG ICG ICA TCI GGC -3´); REP2-I (5´ -´ICG ICT TAT CIG GCC TAC -3´), and BoX A1R (5' CTA CGG CAA GGC GAC GCT GAC G -3´) (Louws et al. 1994). PCR was conducted in a MyCycle thermal cycler (Bio-Rad) in 25 µL reaction volume containing: 100 ng of genomic DnA; 0.8 µM of ERIC primers; 0.2 mM of deoxynucleoside triphosphates; 2.5 mM of magnesium chloride; and 2.5 units of Taq DnA polimerase (Fermentas) in a reaction buffer. The following cycling conditions were used: one cycle of initial denaturation at 95 °C for 7 min; 30 cycles consisting of 94 °C for 1 min; 52 °C for 1 min; 65 °C for 8 min; and a single final extension at 65 °C for 16 min. REP-PCR reactions were performed in 25 µL volume containing: 100 ng of genomic DnA; 0.5 µM of each primer; 0.2 mM of deoxynucleoside triphosphates; 2.5 mM of magnesium chloride; and 2.5 units of Taq DnA polimerase in a reaction buffer. The amplification conditions included one cycle of initial denaturation at 95 °C for 6 min; 30 cycles consisting of 94 °C for 1 min; 40 °C for 1 min; 65 °C for 8 min; and a single final extension at 65 °C for 16 min.
Aliquots of REP-PCR (8 µL) and ERIC-PCR (10 µL) products were separated by electrophoresis on a 2.5% agarose gel stained with ethidium bromide 10 mg·mL -1 in TAE buffer (0.04 M Tris-acetate/0.001 M EDTA) and then visualized and photographed under UV light using a digital system (Alpha Innotech 2200).
A dendrogram was constructed using fragments ranging from 200 to 1400 bp, scored for presence (1) or absence (0), using the Jaccard (S J ) coefficient and UPGMA (Unweighted Paired Group Method) algorithm from the nTSYS-PC program, version 1.70 (Rohlf 1992).

RESuLTS ANd dISCuSSION
All P. syringae pv. garcae strains investigated herein were pathogenic to the coffee seedlings, but 12 of them, obtained from 1978 to 2011, were more aggressive. Considering the leaf area affected by the disease, in the first experiment the strain IBSBF 1293 was the most aggressiveness, followed by the strains IBSBF 65, 1664, 2212, 3046 and 75. In the second experiment, the strains IBSBF 2212 and 3046 were more aggressive followed by the strains IBSBF 1664 and 1293. By observing simultaneously the results of percentage of area affected by the disease and number of lesions per leaves, in both experiments, the isolates can be divided in four categories: HA -highly aggressive; A -aggressive; MA -moderately aggressive and LA -less aggressive. According to this results, the higher aggressive strains were IBSBF 1293, 1664, 2212, e 3046. The aggressiveness classification of all the other strains are described on Table 2. Symptoms of bacterial halo blight were not observed in the control treatments.
This is the first study in which the severity of bacterial halo blight was assessed with the interactive application Leaf Doctor (Pethybridge and nelson 2015). The use of the application was somewhat laborious because all the leaves needed to be photographed, but it reduced the human error involved in estimating the leaf area affected by the disease and also made possible the experiment be well documented, therefore it can be a valuable tool for assessing this disease. There was a significant correlation between the number of lesions per leaf and the area affected by the disease in the first (R 2 = 0,776847) and second experiment (R 2 = 0,699389).
Although all strains investigated herein have been obtained from severely affected coffee crops, some strains showed high and others low levels of aggressiveness in the coffee seedlings. Variations in aggressiveness of bacterial strains have already been reported in other pathosystems and are rather expected. natalini et al. (2006) found variations in the aggressiveness of P. syringae. pv. garcae strains obtained from a pear (Pyrus communis) orchard. Mirik et al. (2011) observed that strains of P. cichorii obtained from lettuce (Lactuca sativa), tomato (Solanum lycopersicum) and Schefflera arboricola were more aggressive to the hosts from which they had been isolated, although they were pathogenic to all hosts. In our study, the genetic clustering of the P. syringae pv. garcae strains showed no correlation with the aggressiveness of the isolates. In contrast, Vicente and Roberts (2007) found that rep-PCR grouped, with some exceptions, the more aggressiveness strains of P. syringae pv. morsprunorum race 1 isolated from cherry trees. It was observed for P syringae pv. garcae that aggressiveness can be lost by successive subcultures (Moraes et al. 1975 In this study, the strains did not show variations in the BoX-PCR (data not shown) but exhibited variability in the ERIC-and REP-PCR patterns. The genetic relatedness of P. syringae pv. garcae strains, investigated by ERIC-and REP-PCR, yielded fingerprints with 20 and 24 bands ranging from 1900 bp to 80 bp and 1825 bp to 200 bp, respectively. Banding patterns generated by ERIC-and REP-PCR revealed a high genetic diversity among the strains. The dendrogram constructed with ERIC allowed the separation of the strains into two major clusters with approximately 37% similarity. Cluster I was represented by four strains from the State of São Paulo isolated between the years 1958 and 1978, with altitudes ranging from 524 to 684 m. Cluster II, divided into two subclusters, "a" and "b", allocated the remaining strains with approximately 45% similarity between them. The subcluster "a" was represented by the three strains from Kenya, and the subcluster "b" was further subdivided into two other subclusters, "i" and "ii", with approximately 53% similarity. The subcluster "ii" allocated strains from the counties of Guaxupé, MG, Andradas, MG, and Varginha, MG, (70% similarity), while the subcluster "i" grouped five strains from Minas Gerais 57% similarity), isolated from 1978 to 2011, and the remaining 12 from the State of São Paulo isolated from 1998 to 2009. one strain from Albertina, MG, was allocated with São Paulo strains, and the genetic similarity between them probably has occurred due the exchange of vegetal material between these cities which are geographically close (Figures 1 and 2).
Banding patterns generated by REP-PCR also revealed a high genetic diversity among the strains. In this analysis the strains were separated into two clusters: Cluster I was represented only by the IBSBF 2883 strain from Kenya, with  approximately 20% similarity to Cluster II, which allocated the remaining strains and was subdivided into subclusters "a" and "b", with about 49% similarity to each other. The subcluster "a" included only five strains with 60% similarity, two from Kenya (IBSBF 249 and IBSBF 3037), the pathotype strain (IBSBF 248 P ), also one strain from Guaxupé, MG, and another from ouro Fino/MG. The subcluster "b" grouped the remaining 22 strains, which could be further divided into two other subclusters, "i" and "ii". Subcluster "i" grouped only the strains isolated from Minas Gerais with approximately 63% similarity, while subcluster "ii" only those from São Paulo (similarities ranging 60 to 93%), with exception of Albertina, MG, as also observed in ERIC-PCR analysis. According to this analysis, the strains from São Paulo (cluster II, subcluster "b" and "ii") were mainly separated from those of Minas Gerais (cluster II, subcluster "b" and "i") (Figures 3,4).
In the combined analysis, the strains were separated in two clusters ( Figure 5). Cluster I was only composed by the three strains from Kenya, with approximately 37% similarity with the other strains of this study. Cluster II comprised the remaining 25 strains, divided in two subclusters, "a" and "b", with approximately 50% similarity with each other. The subcluster "a" was composed by four strains from the State of São Paulo in the period from 1958 to 1978 (IBSBF 65, IBSBF 75, IBSBF158 and IBSBF 248 P ). The subcluster "b" allocated 21 strains from the States of São Paulo and Minas Gerais and was subdivided in two subclusters, "i" and "ii", with approximately 57% similarity between them. The subcluster "i" was composed only by strains from the state of Minas Gerais and "ii" by those from São Paulo, with exception of IBSBF 3032 from Albertina/MG ( Figure 5).
In the combined ERIC-and REP-PCR analysis, strains from Guaxupé/MG, Varginha, MG, and Andradas, MG, obtained from 1997 to 2009, grouped in the same subgroup ( Figure 5). These counties belong to the same coffee production region in Sul de Minas, with altitudes varying from 865 to 881 m, Cfb and Cfa climate, annual average temperature of 19,7 to 20,2 o C; and 1200 to 1500 mm average annual rainfall (Table 1).
The strains from São João da Boa Vista, São Sebastião da Grama, Águas da Prata, Garça, Bragança Paulista, Divinolândia, all from the State of São Paulo, as well as Albertina, from Minas Gerais, grouped in similar subgroups in the REP-PCR, and in the combined ERICand REP-PCR dendrograms. These counties, with the exception of Garça (IBSBF 3015), are relatively close to each other, have altitudes varying from 764 to 1021 m, average temperature from 18,2 to 20,1 o C, and 1,493 to 1,590 mm annual average rainfall, as well as Cfa and Cfb climate (Table 1).
Strains from Cristais Paulista, Franca and Altinópolis, that are relatively close to each other, as well as Serra negra and Caconde, that are more distant, grouped in the same sub-group in the combined ERIC-and REP-PCR dendrogram. These counties belong to the same coffee producing region, Mogiana, and show average temperature ranging from 18,1 to 20,1 o C, average annual rainfall of 1,479 to 1,557 mm, Cfb to Cfa climate classification (Climate date org. 2016) ( Table 1).
In this study, the strains did not show variations in the BoX-PCR (data not shown) but exhibited variability in the ERIC-and REP-PCR patterns. Isolates from the same pathovar may have identical REP, BoX and ERIC fingerprints or they may differ in some banding patterns (Louws et al. 1994). Çepni and Gürel (2012) found that ERIC-PCR could not differentiate strains of P. savastanoi pv. savastanoi from strains of P. syringae. pv. tomato and P. s. pv. phaseolicola, but the combination of BoX-PCR with REP-PCR allowed the grouping of strains of different species and hosts in diverse clusters. Gašić et al. (2012) (2007), while studying strains of P. syringae. pv. syringae from cherry trees and other hosts and P. syringae. pv. morsprunorum isolated from cherry trees, found that strains of P. syringae. pv. syringae could vary in their rep-PCR patterns, and strains from different hosts grouped together, but strains from P. syringae. pv. morsprunorum were more aggressive and less variable. Kaluzna et al. (2010) found that BoX-PCR provided a better differentiation of the genetic diversity of P. syringae. pv. syringae strains and P. syringae. pv. morsprunorum races 1 and 2 than ERIC-PCR. In this study, strains from Kenya grouped separately from Brazilian strains in the ERIC-and REP-PCR analysis. The low genetic similarity observed between these strains could be related to their geographic origin. Kairu (1997), previously related differences between strains isolated from Brazil and Kenya, in relation to aggressiveness and biochemical characteristics such as UV fluorescent pigment and bacteriocin production. In a similar pathosystem, Ferrante and Scortichini (2010) also observed that P. syringae pv. actinidiae strains, the causal agent of an outbreak of bacterial canker on Actinidia chinensis (yellow kiwifruit) in central Italy, grouped separately by rep-PCR from strains obtained in other countries.
In this study, the ERIC-PCR technique allows to cluster the strains of P. syringae pv. garcae by period of isolation, while with the REP-PCR technique the strains were grouped by geographical origin. Similar results were observed for Xanthomonads of tomato by Louws et al. 1994.
The combination of ERIC with REP-PCR showed the genetic relatedness of the Brazilian strains of P. syringae. pv. garcae. With some exceptions, the strains obtained after the year of 1997 could be grouped by geographic origin. This may have occurred, probably, because nurseries distribute coffee seedlings in the regions where they are produced, and the exchange of seedlings between different regions may occur, but is rather infrequent. As infected seedlings are the main source of spread of bacterial halo blight to new coffee crops, the strains could evolve separately in each region. This is a relevant information found in this study and implies that control measures aiming to reduce the disease in coffee seedlings nurseries can impact the spread and management of the disease in new coffee crops. Although some of the strains were grouped by aggressiveness classification, such as, in the combined analysis (ERIC/REP-PCR), the Cluster II, subcluster "a" were allocated three strains (IBSBF 65, 75 and 158) as aggressive, and the three strains of Cluster I (IBSBF 249, 2883 and 3037) as moderately aggressive. Even though some strains were grouped, these results could not be correlated. A study with a major number of strains might give a better correlation of aggressiveness and genetic diversity of P. syringae pv. garcae.
Similar results were observed by Scortichini et al. (2001), while studying strains of Xanthomonas arboricola pv. juglandis from Persian walnut. The authors observed that, although the genetic similarity of those isolates was high, the strains grouped in three different clusters related to their Italian or Greek geographic origin. Gutiérrez- Barranquero et al. (2013) also reported that the ERIC-and REP-PCR clustered separately P. syringae pv. syringae strains isolated from mango from strains obtained from other hosts, and also grouped them by their geographic origin.
Since 1997, the majority of the P. syringae pv. garcae strains (80.9%) was originated from regions near or above 800 m altitude, considered high altitude regions in Brazil. In higher altitude regions, the climatic conditions are very suitable to the coffee crop and are favorable to produce beverage with higher quality. However, in these areas, the milder temperatures and frequent incidence of winds are also favorable for bacterial halo blight. Considering that P. syringae pv. garcae strains isolated in the last years clustered by their region of origin, recent outbreaks of the disease could be related to nurseries distributing infected seedlings in regions with favorable climatic conditions, but further studies are still necessary to validate this hypothesis.

CONCLuSION
All the Pseudomonas syringae pv. garcae strains were pathogenic to coffee seedlings but 12, obtained from 1977 to 2011, were more aggressive on Coffea arabica cv. Mundo novo.