Bacteriophages were discovered more than a century ago and estimated to the most diverse and abundant biological entities on earth. They play an important role in controlling bacterial community (Gómez and Buckling, 2011; Koskella and Brockhurst, 2014), nutrient cycling (Weinbauer, 2004; Wilhelm and Suttle, 1999) and bacteria genome evolution (Howard-Varona et al., 2017; Morgan et al., 2010). Although the billions of bacteriophages infecting different classes of bacteria exist in the environment, most characterized bacteriophages are the members of the Caudovirales, which are either virulent or temperate bacteriophages, depending on their types of infection: lytic and lysogenic (Dy et al., 2018). The infection cycle of bacteriophages is begun by the adsorption of bacteriophages with the special receptors located in the cell surface of susceptible bacteria. Upon irreversible attachment, they inject their genomic DNA (gDNA) into the host cell cytoplasm (Fig. 1). In terms of lytic replication cycle, after gDNA injection, bacteriophages utilize ribosomes of the host bacterium to manufacture phage proteins. The bacterium also provides resources for bacteriophage genome replication and production of virion-related protein components (Dy et al., 2018). At the late stage of the lytic replication cycle, bacteriophages encode holins and lysins, known as endolysins, to lyse the bacterium for release of the phage progenies. In contrast to the lytic cycle, the lysogenic replication cycle has been known as the integration of bacteriophage genome into the bacterial cell chromosome—termed as prophage—or existence as an episomal element, and then they replicate and transfer to daughter bacterial cell (Fortier and Sekulovic, 2013). Prophages can spontaneously alter to a lytic cycle and kill their host spontaneously (Nanda et al., 2015) due to certain environmental stresses, metabolic condition of host bacteria, or antibiotic treatment (Davies et al., 2016). In general, lytic bacteriophages have been used for disease management of bacterial diseases because they can directly kill pathogens, resulting in protecting plants from those pathogens, while temperate bacteriophages may be used for bacterial-pathogen diagnosis because of their genome integration with bacterial chromosome. If the various fusion proteins, including green fluorescent protein can be used in lytic bacteriophages, bacterial pathogen with phage proteins can be easily detected.

The pioneer in phage therapy was Ferick Twort (1915) and Felix d’Hellex (1921), who observed some small agents parasitizing bacteria in growing culture and named as “bacteriophage”. It was recognized as potential antimicrobial agents soon after that (Hermoso et al., 2007). However, the interest in phage therapy was rapidly reversed and displaced by the discovery into new broad-spectrum antibiotics in the 1940s. Bacteriophages, in terms of association with plant pathogenic bacteria, were first discovered by Mallmann and Hemstreet, who demonstrated the inhibited growth of Xanthomonas campestris pv. campestris, by treatment with filtered decomposed cabbage (Mallmann and Hemstreet, 1924). Subsequently, Kotila and Coons suggested that the isolated bacteriophages could prevent soft rot on slices of potato tuber and carrot caused by Pectobacterium atrosepticum and Pectobacterium carotovorum subsp. carotovorum, respectively (Coons and Kotila, 1925; Kotila and Coons, 1925). Bacteriophage effects on plant bacterial pathogen under open conditions were first shown that bacteriophage-treated corn seeds displayed a reduced incidence of Stewart’s wilt disease incited by Pantoea stewartii by 16.5% (Thomas, 1935). However, early scientists found that bacteriophage treatment in the field to control bacterial diseases in plants was less effective than newly discovered broad-spectrum antibiotics (Goto, 2012). Due to this reason, the interest in phage therapy was rapidly faded, despite their promising results for disease management.

Due to the control efficacy, antibiotics and bactericidal chemicals have been major components of management strategies for controlling bacterial plant diseases for decades (Agrios, 2005). However, their negative effects on environment and human health were rapidly emerged (Hermoso et al., 2007). Moreover, the prevalence of antibiotic-, pesticideor copper-resistant bacteria such as Erwinia amylovora (Manulis et al., 2000; Stall, 1962), Pseudomonas syringae (Hwang et al., 2005; Masami et al., 2004; Mellano and Cooksey, 1988), Xanthomonas campestris pv. juglandis (Lee et al., 1994), Xanthomonas citri spp. citri and Xanthomonas alfalfa spp. citrumelonis (Behlau et al., 2011), as well as very slow development of new effective antibiotics, twisted scientists’ attention toward other potential biocontrol agents for control of bacterial plant diseases. Combined with consumers’ demand for organic and antibiotic-free products resulted in the rehabilitation of phage therapy applications in plant disease management.

As stated above, only lytic bacteriophages were given a priority for this aspect with immediate lysis ability as well as avoiding the probable negative effects of lysogenic bacteriophages such as enhancing the virulence of their host bacteria (Frobisher and Brown, 1927; Groman, 1953, 1955), promoting the ecological fitness of bacteria during infection (Flockhart et al., 2012; Fortier and Sekulovic, 2013), and protecting host bacteria from the infection of other lytic phages (Davies et al., 2016). Stonier et al. (1967) reported that fewer than 10 bacteriophage particles showing the clear plaques present at the beginning of 21-h induction period were able, at times, to inhibit completely tumor induction by highly virulent Agrobacterium tumefaciens corn strain B6. In the following years, other bacteriophages also found to target some other plant pathogenic bacteria such as Xanthomonas campestris pv. pruni (Civerolo, 1973; Civerolo and Keil, 1969; Saccardi et al., 1993), A. tumefaciens (Boyd et al., 1971), Xanthomonas oryzae (Kuo et al., 1971), Pseudomonas solanacearum currently renamed as Ralstonia solanacearum (Tanaka et al., 1990), and E. amylovora (Schnabel et al., 1998).

Even though many laboratory-based assays indicated the significant potential of bacteriophages as biocontrol agents, the applications of bacteriophage for plant disease management in field still need more technical improvement to overcome several limitations or challenges. Bacteriophages have been mostly applied directly to the rhizosphere by soil drench or sprayed to the phyllosphere of plants (Fig. 2). Gill and Abedon (Gill and Abedon, 2003) figured out several factors influencing the success of phage therapy in both environments. In rhizosphere, mostly soil, the available water is essential for free-bacteriophage diffusion. Indeed, the biofilm-trapped bacteriophages or reversibly absorption to particles are commonly found in soil, which lacks a continuous aqueous phase. In this case, bacteriophages become less mobile, sometimes permanently inactivated in low soil pH and may be hard to encounter a suitable host.

The phyllosphere is not a safe place and leads to a sharp reduction of the bacteriophage population (Balogh, 2002; Balogh et al., 2003; Civerolo and Keil, 1969; Iriarte et al., 2007). Many studies in field and laboratory conditions demonstrated that bacteriophages are easily inactivated by exposure to high temperature, high and low pH (Gašić et al., 2018) and sunlight irradiation (especially UV-A and UV-B spectrum) (Gašić et al., 2018; Iriarte et al., 2007). In the open field conditions, the population of bacteriophage mixtures was rapidly decreased and practically eliminated 36-48 h after spraying (Tewfike and Desoky, 2015). The short-period persistence of bacteriophages in the leaf surface is the main limitation of phyllosphere-bacteriophage application. Several approaches have been investigated to improve the efficacy of controlling plant pathogens in the phyllosphere. Encapsulated bacteriophages with some kind of formulation such as corn flour, skim milk, casein, sucrose, congo red, and lignin (Arthurs et al., 2006; Balogh, 2002; Balogh et al., 2003; Behle et al., 1996; Ignoffo et al., 1997); use of carrier bacteria for bacteriophage propagation in the target environment (Balogh, 2006; Boulé et al., 2011; Tanaka et al., 1990) or treatment in the early morning or in the evening (Balogh et al., 2003; Iriarte et al., 2007) showed promising results in bacteriophage-longevity of extension and bacteriophage-treatment efficacy. The control efficacy of bacteriophage-based treatment can be also influenced by many other factors: bacterial flora and bacteriophage density, virion decay rates, bacteriophage enrichment in the target environment and the timing of treatment and the surrounding environment (Gill and Abedon, 2003).

An additional challenge of biological control using bacteriophages in agriculture is a low correlation between bacteriophage characteristics in vitro such as host range, plaque size, and in vitro lytic activity and actual control efficacy in field (Bae et al., 2012; Bhunchoth et al., 2015; Gašić et al., 2018). Indeed, while in vitro bioassays showed the stability in various environmental factors and reduced diseasesymptom on leek leaves against P. syringae pv. porri, no significant difference between bacteriophage-treated and untreated plants in the field trial (Rombouts et al., 2016). The strategical use of ΦRSL1, which did not rapidly kill cells in vitro, compared to other isolated-lytic bacteriophages, which killed rapidly in vitro, showed efficient control the R. solanacearum population as well as movement in tomato root (Fujiwara et al., 2011). Moreover, its protection capability against R. solanacearum remained for 4 months, while other highly lytic bacteriophages did not induce similar plant-protecting effects. Filamentous bacteriophage XacF1, which formed small and turbid plaque during plaque assay, caused significant defects in extracellular polysaccharide production, mobility, growth rate as well as virulence to host bacteria—X. axonopodis pv. citri (Ahmad et al., 2014).

To effectively control and reduce the spread of bacterial diseases, specific and rapid detection is important. The extreme specificity to host bacterial strains, the possibility of massive production and resistance to critical conditions, potential discrimination between live and dead bacteria have led the bacteriophages as potential tools for bacterial detection (Farooq et al., 2018). A number of publications have demonstrated the possibility of bacteriophages as a biosensor for plant pathogen detection during their host infections. Sutton and Katznelson (1953) aimed to isolate bacteriophages for diagnosis and identification of some seed-borne pathogenic bacteria. Four polyvirulent bacteriophages for P. pisi isolated from peas successfully discovered the existence of bacterial blight pathogen out of 14 Pseudomonas species in pea seeds and plant tissues from the infected field (Sutton and Katznelson, 1953). Such way was defined as phage typing schemes, which traditionally screen the sample with the specific phages (A in Fig. 4). However, to isolate a specific phage for the particular pathogen is laborious and time-consuming (Singh et al., 2012).

With the possibilities and of genetic engineering techniques, bacteriophages can be engineered as a reporter phage to transfer or insert a reporter gene into target bacteria (B in Fig. 4) (Farooq et al., 2018). The expression of such genes marks the bacteria as a signal marker for bacterial diagnosis (Burnham et al., 2014). Indeed, recombination of luxAB gene encoding the bacterial luciferase was inserted into the bacteriophage PBSPCA1 as a diagnostic for the detection of Pseudomonas cannabina pv. alisalensis (Schofield et al., 2013). PBSPCA1::luxAB was sensitive to detect P. cannabina pv. alisalensis in liquid culture within 20 minutes as well as in diseased plant specimens. The same reporter gene was also introduced into bacteriophage Y2 to detect the fire blight pathogen, E. amylovora. The engineered bacteriophage Y2::luxAB rapidly and effectively detected the pathogen at lower number of viable bacteria, approximately 3.8 × 10³ without enrichment in both in vitro or and plant materials (Born et al., 2017).

Kutin et al. (2009) developed a sensitive, effective protocol for detecting R. solanacearum based on the rapid self-replication of bacteriophage with the quantitative PCR (qPCR) (C in Fig. 4). A bacteriophage M_DS1, which specifically infected 61 of 63 R. solanacearum isolates, was selected for the development of the bacteriophage-indirect assay. qPCR indirectly detected R. solanacearum via the detection of a bacteriophage DNA fragment. The method sensitively detected the presence of the pathogen in potted plants, and the detection limit was near 10² cfu/g in 0.1 g of leaf tissues and 103 cfu/ml in drainage water from the pot. In the infected soil, the detection limit was approximately 10² cfu/g (Kutin et al., 2009).

Bacteriophages have been explored as potential agents to manage bacterial diseases in plants and also to specifically detect plant-pathogenic bacteria, as highlighted in this paper. Multiple investigations of the potential application of bacteriophages in the case of bacterial disease control have been studied and observed many promising results. However, almost successful applications of bacteriophages were performed in controlled conditions like greenhouses, while agricultural production mainly occurs in an open environment where the environmental factors are constantly changeable and uncontrolled. Therefore, more field trials have to perform to fully implement its efficacy in open conditions. Moreover, even though many promising results were published throughout last decades, there are few commercial bacteriophage-based products that have reached the market for the control of several bacterial plant diseases such as AgriPhages for bacterial spot or speck of tomatoes and peppers and fire blight of apple and pear trees, Erwiphage for fire blight of apple trees, Biolyses for soft rot disease of potato tubers (Buttimer et al., 2017). The efficacy of bacteriophage application is also impaired by several environmental factors so it is also critical to the development of delivery strategies or formulations for commercial purposes.

Developing standard criteria for selecting bacteriophages is also needed more attention for phage therapy. Many evidences demonstrated that current criteria may be effective for several situations but it also remained some exceptions (Ahmad et al., 2014; Fujiwara et al., 2011; Rombouts et al., 2016). Only lytic bacteriophages have been utilized for plant disease management nowadays, but there is still a big question mark over the potential and risk of temperate bacteriophages. Although the natural-temperate bacteriophages were not ideal as biological agents for plant disease control because of their replication cycle, they can be modified to become virulent or work as a delivery vehicle for genetic elements for restoration of antimicrobial susceptibility or virulence-factor disruption (Balogh et al., 2010). Moreover, in the case of phage-based pathogen detection, the engineered phages aim to introduce a marker gene into the target bacterial genome. Therefore, no matter whether reporter phages are lytic or lysogenic, it still potentially detects the targeted bacterial pathogen (Farooq et al., 2018).

The ectopic expression of phage-based proteins in plants exhibited the enhancement of plant resistance to pathogenic bacteria (Dong et al., 2008; Wittmann et al., 2016). However, the use of transgenic plants may present a challenge in certain countries and to consumers. Therefore, it needs more detailed analysis to optimize the efficacy as well as minimize potential side effects.

In conclusion, although agrochemicals such as antibiotics and copper are still mainly used for control of bacterial plant diseases in field, there is a significant potential of bacteriophage usage to reduce the amount of agrochemicals or to replace those agrochemicals for the control of bacterial diseases in plants. For this, more bacteriophages should be collected for diverse bacterial pathogens, and more field trials instead of trials in the controlled conditions are necessary.