Fire blight – a global threat to fruit production

Apple (Malus spp.), pear (Pyrus spp.), quince (Cydonia spp.), raspberries or blackberries (Rubus spp.), and many other crops of the rose family (Rosaceae) are susceptible to infections by the Gram-negative bacterium Erwinia amylovora, a member of the Erwiniaceae family within the Enterobacterales order. E. amylovora causes fire blight, a devastating disease that annually results in high economic losses to growers worldwide (Bonn and van der Zwet 2000; Vanneste 2000). Fire blight was initially discovered in North America in the 1780s and has since spread to Europe, Africa, Australasia, and the Middle East (Bonn and van der Zwet 2000; Vanneste 2000). It seems the disease is established worldwide. With worsening global warming, further spread of fire blight into more unaffected countries and regions is reasonable.

E. amylovora is generally transmitted by wind, rain, and insects to the blossoms of susceptible plants during the flowering period (Emmett and Baker 1971; Johnson and Stockwell 1998). After colonization of the hypanthium and pistil, it invades the ovaries and spreads further through the xylem vessels of the infected plant (Castiblanco and Sundin 2018; Vanneste 2000; Koczan et al. 2011). Secondary infections can occur if bacteria enter the plant via wounds or after shoot infections. In the invaded xylem vessels E. amylovora produces large amounts of exopolysaccharides (EPS) such as amylovoran, levan, and bacterial cellulose (Ayers et al. 1979; Bereswill and Geider 1997; Koczan et al. 2009; Castiblanco and Sundin 2018; Gross et al. 1992; Geier and Geider 1993). EPS production leads to clogging of the vasculature and prevents sufficient transport of nutrients and water in the infected parts of the plant (Vanneste 2000). Eventually yellowish infectious ooze is secreted out of infected stems or fruits. In addition, E. amylovora injects effector proteins into the plant cell’s cytoplasm using a type III secretion system (Kamber et al. 2017; Castiblanco et al. 2018). The effector proteins cause necrosis of the cells (Bogdanove et al. 1998; Nissinen et al. 2007; Triplett et al. 2009). Therefore, symptoms of fire blight resemble blackened tissue, leaves, and stems. Hence, the name fire blight due to the “burnt” appearance of the tree. Additional virulence factors of E. amylovora have been described elsewhere in more detail (Koczan et al. 2011; Yang et al. 2020).

Control of fire blight becomes more and more challenging. Antibiotics, including streptomycin, kasugamycin, and oxytetracycline, are routinely applied in some places to control the disease (McManus et al. 2002; Slack et al. 2020). However, due to the increasing incidence of antimicrobial resistance and the occurrence of multiple drug resistant bacteria in clinical settings, the application of streptomycin for the control of E. amylovora was banned in Europe. Pruning and clearing of infected plants and the eradication of entire orchards are necessary to prevent spread of E. amylovora after infection in the field. Prevalence of streptomycin resistant (SmR) E. amylovora declined from very high in 2006 and 2007 to very low incidence in 2013 and 2014 in California, USA (Förster et al. 2015). From 1280 isolates collected from apple orchards in New York, 34 isolates (2.65%) exhibited strAB genes located on transposon Tn5393 on the nonconjugative plasmid pEA29 (Tancos et al. 2016). In New York prevalence of SmR E. amylovora was monitored over 6 years (2015–2020). SmR E. amyolovora was only detected in two counties and was always associated with only one CRISPR profile. The authors conclude that the SmR E. amylovora originally introduced into NY state was never fully eradicated (Wallis et al. 2021).

Bacteriophages offer unique properties to prevent infections by E. amylovora

Bacteriophages (short phages) of the Caudoviricetes class are tailed viruses which infect bacteria for their own replication and progeny virion synthesis. An infection usually starts with the adsorption of the phage to the bacterial surface by tail fiber (TF) or tail spike proteins (TSP) that are connected to the phage’s base plate (Klumpp et al. 2023). In case of E. amylovora surface structures such as the lipopolysaccharide (LPS), amylovoran, and bacterial cellulose were identified as primary phage receptors so far (Born et al. 2014; Knecht et al. 2022a2022b). In general, any other structure that is presented at the host cell’s surface could be used by phages for adsorption. Flagella (Bertozzi Silva et al. 2016; Shin et al. 2012), pili (Guerrero-Ferreira et al. 2011), and diverse outer membrane proteins (Marti et al. 2013; Lagonenko et al. 2015; van den Berg et al. 2022) were identified as phage receptors in bacteria. After adsorption the phage DNA is translocated across the cell wall into the cytoplasm of the host cell. After circularization the phage DNA is replicated, and novel infective virions are assembled intracellularly. At the end of the infection cycle the host cell is lysed to burst release infective virions into the environment.

A phage that is applied as a biopesticide should be non-transducing and strictly virulent (Hagens and Loessner 2010). Temperate phages are generally not applied, because a lysogenized cell will be resistant against superinfection by the same phage, a phenomenon known as superinfection exclusion or homo immunity suppression (van den Berg et al. 2022). In addition, temperate phages often encode virulence factors which could contribute to the virulence of pathogenic host bacteria (Yano et al. 2023). Interestingly, temperate phages infecting Erwinia spp. have only rarely been described (Roach et al. 2015; Andrade-Domínguez et al. 2018; Zlatohurska et al. 2019; Zrelovs et al. 2020). If phages are non-genetically modified and non-transducing they are “generally recognized as safe” from the admission authorities (Hagens and Loessner 2010). The receptor specificity of a phage should be known to better understand the adsorption process and to select phages with different receptor specificities for the composition of phage cocktails to avoid rapid development of phage resistance (Abedon et al. 2021). Another advantage of combining different phages into a cocktail is that the host range is extended (Shaidullina and Harms 2022). Preferential phages ideally infect all types of strains within the target species. In E. amylovora many strains can be differentiated into different CRISPR element types. Depending on the spacer content most apple and pear strains isolated in the eastern US can be distinguished from western US strains (Rezzonico et al. 2011; McGhee and Sundin 2012; Parcey et al. 2022). Interestingly, these strains exhibit different phage susceptibilities (Gayder et al. 2019). Finally, phages should be easy to propagate in large culture volumes with high yields to meet economic requirements.

A major advantage of bacteriophages over standard antibiotics is that they cannot cause significant dysbiosis because they specifically infect host bacteria only (Federici et al. 2023). Phages therefore leave the natural microbiome, that can contain many useful organisms, unharmed. Specificity of phage infection is mostly due to the coevolution of phages with their host bacteria and is mainly mediated by the adsorption of the virion to the host cell (Taslem Mourosi et al. 2022) and the development of phage defense mechanisms (Pfeifer et al. 2022).

Many diverse mechanisms of phage resistance are described in the literature including downregulation, mutation, and masking of phage receptors (Riede and Eschbach 1986; Forde and Fitzgerald 2003; Harvey et al. 2017), DNA restriction and modification (RM) systems (Tock and Dryden 2005), CRISPR/Cas based immunity systems (Barrangou et al. 2007; Dupuis et al. 2013), and abortive infection systems (AIS) (Dy et al. 2014). Recently, several novel mechanisms of phage resistance, e.g., BREX, DISARM, cGMP-AMP signalling, and viperins, were identified (Goldfarb et al. 2015; Ofir et al. 2018; Doron et al. 2018; Cohen et al. 2019; Morehouse et al. 2020; Bernheim et al. 2021).

Phages evolved to overcome phage resistance by changing receptor specificities (Mahichi et al. 2009; Chen et al. 2017; Boon et al. 2020; Holtzman et al. 2020; Zhang et al. 2022). RM systems are circumvented by deleting endonuclease cleavage sites from phage genomes or by modifying phage DNA bases. CRISPR/Cas systems can be rendered inactive by phage encoded anti-CRISPR proteins that inhibit Cas endonucleases (Samson et al. 2013; Borges et al. 2017; Safari et al. 2020).

A large collection of phages infecting E. amylovora is available from many different laboratories worldwide (Akremi et al. 2020; Arens et al. 2018; Besarab et al. 2022, 2022; Boulé et al. 2011; Born et al. 2011, Buttimer et al. 2018; Erskin 1973; Esplin et al. 2017; Gill et al. 2003; Kim et al. 2020; Knecht et al. 2018; Lagonenko et al. 2015; Lehman et al. 2009; Meczker et al. 2014; Müller et al. 2011; Park et al. 2018; Ritchie and Klos 1977; Sabri et al. 2022; Schnabel and Jones 2001; Sharma et al. 2018, 2019; Zlatohurska et al. 2019, 2023). They have been extensively studied in vitro. In general, the host range is monitored by standard plating techniques and the lytic activity is demonstrated in liquid culture (Adams 1959). However, the host range can also be studied with the use of qPCR, which can greatly expedite the process and allow for larger and more detailed studies. Previously, the host range of ten phages belonging to four different genera was measured against a globally sourced collection of 106 E. amylovora and E. pyrifoliae strains (Gayder et al. 2019), and also against 30 P. agglomerans and other orchard epiphytes (Gayder et al. 2020). Perhaps the most significant finding from this was that the highest degree of phage resistance was found in strains isolated from Western North America compared to the rest of the world.

Some phages have been examined by electron microscopy to verify classification (Gill et al. 2003). However, due to the limits of classical classification, the new phage classification system is now based on genomic data. It is currently changing as the International Committee on Taxonomy of Viruses (ICTV) is adding novel phage families to the classification scheme. Currently (March 2023), phages infecting E. amylovora can be distinguished in at least 11 different families with 27 different phage genera (https://ictv.global/taxonomy). Supplementary Table 1 summarizes all sequenced phages infecting E. amylovora known to date.

Phages can exhibit enzymatically active tail spike proteins (TSP) such as depolymerases, as structural proteins of the baseplate at the end of the phage tail (Born et al. 2014; Latka et al. 2017; Knecht et al. 2020). These proteins specifically bind to polysaccharide moieties of the host cell wall. After binding the substrate is cleaved. In phages infecting E. amylovora TSPs are encoded by members of the Agricanvirus, Elunavirus, Eracentumvirus, Nezavisimistyvirus, Waedenswilvirus, and the potential new Keyvirus genera. In phages phiERA103, phiEa1h, phiEa100, L1, and S2 the TSP specifically cleave amylovoran, the major constituent of the bacterial capsule (Müller et al. 2011a, b; Born et al. 2014; Knecht et al. 2018). Phages Bue1, Ray, Hena1, phiEa2809, phiEaH1, Micant, Loshitsa2, Ea9-2, Rexella, Frozen, Gutmeister, VyarbaL, Key, and Zoomie encode so far non characterized TSPs (Meczker et al. 2014; Lagonenko et al. 2015; Esplin et al. 2017; Knecht et al. 2018; Besarab et al. 2020, 2022; Sharma et al. 2019; Zlatohurska et al. 2019). Since Bue1, Ray, Key, and likely many others do not exclusively infect E. amylovora, but also diverse strains of Pantoea spp., the specificity of the encoded TSP should be analyzed in more detail. From phage RAY the EPS depolymerase was expressed in E. coli and spotted on a lawn of P. vagans. The formation of a halo indicated enzymatic activity that was not evident on a lawn of E. amylovora (Sharma et al. 2019). Another phage, the Schitovirus S6, is exceptional, because it encodes an enzymatically active cellulase, Gp95, that is likely located at the phages’ baseplate. The enzyme cleaves bacterial cellulose, a minor constituent of the E. amylovora capsule (Knecht et al. 2022b). Since phages that interact with amylovoran or cellulose exhibit synergistic effects in reducing viable cell counts together with capsule independent phages, they can be elegantly added to a phage cocktail (Born et al. 2011).

As capsule independent phages, members of the Loessnervirus genus could be considered as these phages are LPS dependent and exhibit a broad host range with a high efficacy of plating (Knecht et al. 20221). They exclusively infect E. amylovora only (Born et al. 2011).

In many studies so called “jumbophages” that could infect both E. amylovora and strains of Pantoea spp., were identified (Yagubi et al. 2014; Esplin et al. 2017; Arens et al. 2018; Buttimer et al. 2018; Sharma et al. 2018, 2019; Kim et al. 2020). The name is derived from the exceptionally large genome size exceeding 200 kb. Receptor specificities have not been reported yet, indicating that many phages are still awaiting their full characterization. However, the proportion of Erwinia infecting phages that are jumbophages is abnormally high compared to other target hosts (Thompson et al. 2019). The reason for this is not known but presents interesting avenues for further research.

In addition, Akremi et al. (2020) described the isolation and characterization of filamentous phages infecting E. amylovora. These phages exhibit neither a lytic, nor a temperate life cycle, but establish a persistent association with the host (Akremi et al. 2020). They do not lyse the host, because they are extruded from the infected cell. However, in the host bacterium an infection by a filamentous phage can cause a loss of fitness and therefore attenuate the bacteria (Akremi et al. 2020).

After in vitro analyses the efficacy of phages to inhibit E. amylovora in planta needs to be determined. Many phages with promising capabilities in reducing viable cell counts of the host bacteria in vitro, exhibit different properties in planta (Knecht et al. 2022b). Matrix effects could alter the adsorption kinetics of phages and render host cell infection more inefficient. So far, the interaction of phages with E. amylovora was mostly studied in standard growth media and buffers, but not in nectar, xylem sap, or similar solution containing higher concentrations of sucrose, glucose, and fructose. The host bacteria may also behave differently in planta because growth and metabolism are dependent on changing environmental conditions (Lee et al. 2022; Schachterle et al. 2022). This may lead to altered gene expression and presence of phage receptors at the host cell surface, which will dramatically impact phage infection. An example is phage S6 that depends on bacterial cellulose as primary receptor for adsorption (Knecht et al. 2022b). The bacterial cellulose synthesis operon (bsc) in E. amylovora is likely constitutively expressed, but the amount of cellulose that is produced by the bacteria may vary dependent on culture conditions. It seems that cellulose is needed by the bacteria to better establish biofilms in xylem vessels and to withstand xylem sap flow (Castiblanco and Sundin 2018). It is not clear yet how much cellulose is produced during the colonization of the blossoms. Phage S6 could not protect blossoms from infection by E. amylovora (Knecht et al. 2022b). In the same context the amount of amylovoran differs in strains of E. amylovora and depends on the availability of nutrients (Schollmeyer et al. 2012). However, even if synthesis of amylovoran is reduced during blossom infection it still seems to be enough for adsorption of amylovoran-dependent phages (Knecht et al. 2022a). Loss of amylovoran synthesis would abolish phage infection (Born et al. 2014) and will in turn also attenuate virulence of the bacteria – another reason for selecting amylovoran specific phages for fire blight control.

How to prevent fire blight with phages – considerations for practical application

The use of phages for the control of E.amylovora was first investigated as early as 1973 (Erskine 1973; Ritchie and Klos 1977). Since then, different models have been established to study the virulence of E. amylovora and its interaction with phages. Immature fruits can be used because they can be easily collected and stored and are comfortable to work with. Fruit slices are generally inoculated with contaminated buffer that is pipetted onto the sliced surface. This model seems to be useful to study the interaction of E. amylovora and phages within fruit. Infection, virulence, or inhibition of E. amylovora is measured semi quantitatively by documenting size and appearance of necrotic spots, and by judging symptom development. Müller et al. (2011b) reported on the reduction of disease symptoms on apple flowers and immature pears. Four different phages were tested. Phages phiEa1h and phiEa100 reduced the recovered pathogen by only 40%, while phiEa104 and phiEa116 revealed 90% symptom reduction (Müller et al. 2011b). Akremi et al. (2020) demonstrated that a cocktail of four filamentous phages significantly reduced symptom development in a pear bioassay.

Using detached blossoms, the primary infection of a plant is mimicked, which is superior to the immature fruit model, because it is closer to the situation in the field. However, growth of molds often interferes with the assay, which limits the usefulness of this approach. Schwarczinger et al. studied the efficacy of phage H6 and H5B which significantly reduced viable cell counts of E. amylovora by at least 45%. The authors also demonstrated that the efficacy of the phages depends on the plants’ cultivar (Schwarczinger et al. 2011; Nagy et al. 2012). In another study reduction of viable E. amylovora by phage Y2::dpoL1-C was inoculum-dependent. On blossoms inoculated with 102 CFU/ml, Y2::dpoL1-C caused a reduction of the final cell number by ca. 6 logs. Interestingly, no bacteria could be recovered from 95% of the flower samples (Born et al. 2017). The study revealed the usefulness of engineered phages for biological control. Here, phage Y2 was artificially equipped with the TSP gene of phage L1 (EPS depolymerase dpoL1-C). The resulting phage Y2::dpoL1-C did not exhibit an altered burst size or latency period compared to the wild type phage. Infection of E. amylovora by Y2::dpoL1-C resulted in the production of soluble TSP proteins which were not integrated into the Y2 virion, but were burst released together with the phage from infected cells to enhance efficacy of the phage adsorption and replication in later infection cycles. However, since Y2:: dpoL1-C is a recombinant phage it would be regarded as a GMO with particular requirements for approval.

Yet another way to study the efficacy of phages for fire blight control was demonstrated by Sabri et al. The authors injected E. amylovora into the stem of pear plants followed by the injection of phage ET-IT22 at the same site. After 40 days post inoculation the treated plants remained as asymptomatic as the streptomycin control (Sabri et al. 2022). Nagy et al. applied phages phiEa104 and H5K to the roots of apple seedlings and demonstrated that infective virions could be re-isolated from the stem above ground level thereby reducing symptoms of fire blight significantly (Nagy et al. 2015). It seems that these phages are efficiently translocated through the plant’s tissue.

The most conclusive way to demonstrate the efficacy of phages in agriculture are greenhouse or field trials. However, the experimental setup for such trials is not trivial and should be more standardized for a better comparison of results. In general, phages should be used to prevent an infection during the blossom period rather than curing infected trees when the bacteria have already spread into the xylem. To prevent the pathogen from blossom colonization the phage needs to be applied before the bacteria are transferred. Treatment of blossoms is routinely performed by spraying. Obviously, the timing and dosing are essential for optimal efficacy. After the phage is sprayed, it should not be inactivated by any matrix component or UV radiation but remain viable and infectious. Currently it is not well understood for how long phages remain viable after spraying whole trees. As phages are non-motile, they only randomly hit the host bacteria by passive diffusion. Therefore, many phages are needed for effective control, because a blossom, e.g., the treatment area, should be covered homogeneously and densely. If coverage is too low, incoming E. amylovora may not be infected by a phage, because they simply do not meet each other.

To enhance diffusion and bacteria-phage interaction, a sufficient volume of the spraying solution should be applied. Surfactants could be used to enhance the bioavailability of the phages, wetting of the blossom surface and pistil, and to prevent removal of phages after rain fall. Therefore, formulation of the phage cocktail is required to maximize efficacy. On the other hand, if too much liquid is sprayed, the phage might drip off the treated plants and might be lost. Therefore, the optimal concentration of phages and liquid volume needs to be experimentally determined and the mode of spraying or nebulizing should be considered carefully before application. Ideally a homogenous aerosol produced by a nebulizer is used for treatment to cover a square centimeter with 1–3 × 108 plaque forming units (pfu) of each phage used (Hagens and Loessner 2010). The lesser the total amount of phages is, the more restricted pathogen control will be. Nebulizers offer additional advantages, because they can produce charged droplets which could electrostatically interact with the treated surface. However, the suitability of such approaches has not been determined yet with respect to Erwinia control using phage.

The Svircev lab in Canada established an alternative approach to deliver phages to blossoms. Before spraying phages are mixed with a particular strain of Pantoea agglomerans that is used as a carrier (Boulé et al. 2011; Lehman 2007). The bacteria occupy the same ecological niche on the stigma but are nonpathogenic and compete with E. amylovora during blossom colonization. The advantage of using bacteria as a phage carrier is, that the phages will multiply within the infected carrier bacterium. Because many isolates of Pantoea spp. produce carotenoids (Kumar et al. 2019; Choi et al. 2021), the multiplying phages could be protected from detrimental UV radiation and likely other environmental factors. After carrier lysis the phages will be released onto the colonized blossom, which may result in a direct delivery to the colocalized targeted E. amylovora. The authors reported on the use of phages ΦEa1337-26, ΦEa2345-6, ΦEa2345-19, and ΦEa21-4 which were used together with the carrier P. agglomerans Eh 21.5 on detached pear blossoms (Boulé et al. 2011). In each case severity of fire blight symptoms was decreased significantly. In the presented experiment phages ΦEa2345-6 and ΦEa21-4 revealed a similar efficacy as streptomycin.

The use of Pantoea as a carrier was followed up with a qPCR approach to study population dynamics of different phages, carrier strains, and pathogen strains in liquid culture (Gayder et al. 2020). Of particular significance was the difference in efficacy of different phages depending on the pathogen strain tested. Even though E. amylovora is a highly homogeneous species there is still a great deal of variation in susceptibility to different phages (Gayder et al. 2019; Parcey et al. 2020). The analysis revealed the highest degree of phage resistance in strains isolated from Western North America. Additionally, a strain of P. agglomerans that appeared to be phage resistant was no longer resistant in the presence of another strain. The host range of some Erwinia phages on epiphytic strains showed that there is a high degree of cross-species host interactions. There also seems to be a sort of dichotomy between their preferences for either Erwinia or Pantoea; some of the Eracentumvirus phages infect only Erwinia, the Agricanvirus Ea35-70 only infects Pantoea, and the Kolesnikvirus and Johnsonvirus phages have potential to infect both nearly equally (Gayder et al. 2019, 2020). The implications of this in terms of biocontrol are unknown, other than the potential use of P. agglomerans as a phage carrier for delivery to the blossom (Lehman 2007). Since these phages are not specific to E. amylovora, but do also infect Pantoea spp., more attention should be given to investigating these interactions with natural microflora of a blossom. It is not known how these phages might interact with other epiphytic bacteria. These bacteria could deplete the free phages due to adsorption, or if enough are infected as well could drastically change the microbial composition of the blossom.

Considerations for testing phage biopesticides in field trials

Overall, there is inconsistency when it comes to field trial results, and this likely contributes to the lack of publications of these results compared to the number of phages isolated and labs working on this. Therefore, some considerations should be given to standardize various aspects of field trials in the future, and some of these are indicated in the summary Fig. 1. Additionally, particular emphasis should be given to consider the end user of these products: the grower. We would suggest a formal round-table discussion be organized at the future Fourth International Symposium in Fire Blight of Rosaceous Plants to discuss the potential of implementing some of the following suggestions.

Fig. 1
figure 1

Major considerations for the future of field trials of phage-based biopesticides for the control of E. amylovora

Full size image

One of the biggest variations amongst researchers is which pathogen strains are used and how they are applied. Many different E. amylovora strains are used by different fire blight researchers around the world to study other control methods or biological functions, and they likely would use the same strains for their phage research. Many use some of the reference strains (Ea273, Ea110, or CFBP1430), some use mutant strains resistant to antibiotics such a rifampicin or nalidixic acid (Ea153N, Ea110R, Ea273R), and some can use local isolations which may be completely uncharacterized (Sundin et al. 2009; Stockwell et al. 1998; Khan et al. 2012; Ait Bahadou et al. 2018). Even though E. amylovora is an exceptionally homogenous species genetically (Zeng et al. 2018; Rezzonico et al. 2011; Parcey et al. 2020), there is still significant variation in phage susceptibility between hosts. As mentioned above, this could greatly bias results towards phages that are effective against a particular strain, but that might not necessarily be ideal against an epiphytic population that the growers will be facing. As we move further into the “-omics” era with a growing focus on the microbiome, perhaps we too should consider a standardized library of strains which more accurately represent the diversity of the pathogen. This could help to set a benchmark for the comparison of results of field trials and even in vitro experiments, as currently comparisons between results are nearly impossible.

Another huge variation in field trials is the methods by which E. amylovora infection is induced. The main methods being: (1) the spray application of a suspension of the pathogen one or more times during bloom opening; (2) using pollinators to spread the pathogen around open blossoms as they search for pollen; and (3) relying on a naturally occurring infection in the orchard. Spray applications seem to be the most common method, especially in North American trials, where generally 106 to 108 CFU/mL of the pathogen is sprayed onto open blossoms. This essentially guarantees a high enough degree of infection in controls to reliably calculate treatment efficacy. However, such a highly concentrated inoculation applied fairly evenly across the whole blossom surface with so much liquid is likely not reflective of the vast majority of natural infections. This creates a very different environment for phage-pathogen interactions compared a natural infection, which may be causing phage treatments to fail when they could potentially be very effective when used as a product by a grower. However, because of climatic differences in many geographic regions worldwide pathogen inoculation standardization could be challenging. As an alternative, the percentage of infection in the water control could be standardized. Natural pollinators, e.g., insects such as bees (Apis spp.) or bumble bees (Bombus spp.) are another method of artificially inoculating trees, where either separate trees are sprayed with inoculum for bees to pick up, already infected trees are nearby, or bees are inoculated directly before release to spread the pathogen (Kammerecker et al. 2021). This method may be more representative of a natural infection but requires additional planning and expenses to set up the bees, is not as uniform as spraying due to differing proximity to the inoculum source and poses more risk of contaminating trees outside of the experimental plot.

Finally, researchers can simply spray their treatments and hope that infection occurs naturally in their plot. While some regional restrictions on the direct application of E. amylovora may force some to go with this approach, generally it is not preferred because there is significant risk that little or no infection will occur and there will be no results acquired that year. However, this method would most closely replicate natural conditions. It could also be argued that given how much time is wasted on poorly planned field trials and unpublished attempts anyway, if researchers in the field came together and committed to this method over time we could probably come out further ahead in the long run with more informative and comparable results.

Another aspect of field trials for phage biopesticides to consider is the choice and application of the phages themselves (Fig. 1). Because so little is really known about the phage-host interactions on the blossom we are often just guessing how to assemble and apply these products. It is perhaps instinctual to want to spray as many phages as possible as often as possible during bloom. This poses a large problem, as lab scientists may jump into these projects without considering the perspective of a grower, and if the product is tedious or inconvenient compared to other products then there is already a barrier to convince them to use it. If a more standardized approach for field trials was implemented, experiments could be designed to finally start addressing the issues and questions that so far have held back this technology.

The future of phages for fire blight control has great potential, but years of inconsistency and poorly designed trials have surely held the field back. We must come together in a more united manner or risk irreparably damaging its image to growers and applied researchers altogether. A large-scale comparative and collaborative effort across different countries and climates with standardized methods with the involvement of local extension experts and growers would allow us to dissect the roles that weather, climate, geography, etc. have on the success of phage biocontrol and gain an overall better understanding of what is required to make a highly effective phage biopesticide.

The role of additives in the formulation of phage cocktails

Diverse environmental factors decrease the persistence of bacteriophages under field conditions

Different environmental factors such as UV radiation, temperature and availability of moisture could potentially influence the phages’ performance in field application (Jones et al. 2012). The most significant factor is UV radiation, causing phage loss due direct DNA damage (Born et al. 2015). UV derived phage loss can either be circumvented by avoiding diurnal peaks of radiation, choosing shorter application intervals or the addition of UV protectants to the formulation (Jones et al. 2012). Another critical factor is the presence of moisture on plants, being crucial for phage-bacteria interactions (Jones et al. 2012). Iriarte et al., however, found that in contrast to ambient temperature and fluorescent light, the stability of phages against bacterial spot disease in tomato was only minorly affected by desiccation. Only a slight reduction of phage populations was observed after 60 days. On the other hand, ambient temperature and fluorescent light had a pronounced effect on non-formulated phages (Iriarte et al. 2007). Additionally, it can be argued that the effect of desiccation might be more pronounced in phages applied and acting on the leaf surface than on blossoms. The stigma, the infection point of Erwinia amylovora, provides a moist environment. Due to the presence of hygroscopic substances, the stigma attains 98% relative humidity (Thomson 1986). Additionally, amino acids, glycoproteins and carbohydrates are present on the stigma (Thomson 1986), which might further stabilize phages. Therefore, Erwinia phages are unlikely to be inactivated by desiccation upon application to the stigma and are presumably protected by the stigma’s intrinsic environment.

Other chemical components present in the phyllosphere may also impact phage survival. Copper-based pesticides are particularly damaging to phages, and after the application of such pesticides, a treatment-free interval before the next phage application is necessary (Jones et al. 2012). It was observed that copper treatments applied closer than 4 days prior to phage application strongly impacted stability of phages against bacterial spot disease on tomato (Iriarte et al. 2007). Considering that copper is used as a control agent against fire blight, it needs to be ensured that in field settings, neighboring plots treated with copper do not affect effectivity of phage application. Consequently, intermittent application strategies including copper and phages are not advisable.

Luo et al. summarized the effect of different environmental stresses on the stability of phages against Pseudomonas syringae pv. actinidiae (Psa), the causative agent of kiwifruit canker. Interestingly, it was found that depending on each specific strain of phage, different levels of stability were observed (Luo et al. 2022). In this context, it would be advisable to evaluate each Erwinia phage for its individual behavior under various environmental stresses. General conclusions on the stability of phages may not be drawn. In the case of Psa phages, varying rates of phage infectivity depending on pH and temperature were observed (Luo et al. 2022). Whether this is also the case for phages infecting E. amylovora would need to be experimentally determined. It was also found that generally, DNA phages are more susceptible to UV radiation than RNA phages, and dsDNA and dsRNA phages are more resistant to UV radiation than ssDNA and ssRNA phages (Luo et al. 2022). Most phages infecting Erwinia possess genomes consisting of dsDNA (Walker et al. 2020; Thompson et al. 2019) although individual phages are known to consist of ssDNA (Akremi et al. 2020).

It would be advisable to determine the stability of Erwinia phages under diverse environmental conditions. Of these, UV radiation and temperature appear to be the most relevant. Furthermore, combining phages with UV protectants has the potential of stabilizing phages in planta, while the addition of surfactants can enhance application coverage and lead to an increased efficacy of treatments.

Considerations for the development of commercially competitive formulations

When evaluating substances as potential candidates for the formulation of phage cocktails some practical factors need to be considered. One critical factor is the concentration at which a substance shows activity. Ideally, substances should be active at low concentrations, to facilitate practical aspects such as handling, storage, and distribution. Most (non-biological) plant protection agents are applied at final concentrations of 0.5% or lower, and newly developed formulations should ideally be within this range. If activity of a substance is only given at high concentrations such as 10%, the feasibility for commercialization and field applications is limited. Another important factor is the commercial price. Here, it needs to be evaluated if the addition of a costly additive significantly improves the efficacy of the cocktail, or if the same effect, with less cost, could be achieved by applying a higher dosage of phages. Costs should be kept within the range of already commercially available plant protection agents against fire blight to be competitive. Alternatively, an increased cost of a phage-based agent should be matched by comparably high efficacy. In addition to a substance’s effectiveness, the compatibility with both phages and plants needs to be ensured. The stability of phages in the presence of the candidate substances can be evaluated under laboratory conditions, while potential phytotoxic effects should be examined in planta. Considering these factors at early stages of product development and research, including laboratory tests and field trials, allows for targeted development of a commercially competitive phage based biopesticide.

Formulations for improved phage persistence under UV radiation

UV radiation is grouped according to different wavelengths. While UV-C (100–280 nm) is absorbed by the ozone layer and does not reach the earth’s surface, UV-B (280–320 nm) is considered the most damaging to phages due to direct DNA damage (Iriarte et al. 2007). When identifying candidate substances for UV protection, it is advisable to perform laboratory tests in preparation for field trials. A first indication of UV protection capacity is the absorption spectrum of the candidate substance. Here, a high absorption at UV-B wavelengths 280–320 nm is an indicator of eligibility. For some substances, this information is already available, while for others, this can be experimentally determined. Bitton et al. observed that the UV-protective capacity of a given substance correlated with its specific absorption rather than its absorbance when studying substances for the protection of Klebsiella aerogenes against UV-C (Bitton et al. 1972).

While field trials are the ultimate test to examine a substance’s performance under realistic conditions, they are often difficult to interpret due to big variations between replicates and influences by additional environmental conditions. Therefore, it is advisable to first perform laboratory UV-B tests to estimate a substance’s potential. Hence, some substances could be excluded due to low effectiveness while others may show high potential, making them promising candidates for field trials. When performing laboratory UV-B tests, it is recommended to adjust the dose to the desired protection capacity under field conditions, as described previously (Kaiser et al. 2019). Iriarte et al. observed, that based on a regression curve, the impact of UV radiation on phage decline was stronger when UV intensity was higher within the same cumulative UV dose (Iriarte et al. 2007). This indicates that laboratory experiments, utilizing a high UV radiation intensity to mimic a defined UV dose corresponding to field conditions, might underestimate the effect of UV protection of a given substance or overestimate a phage’s susceptibility to UV radiation. In this scope however, a substance performing well under laboratory conditions under high UV intensities could show an even higher effect when subjected to lower UV intensities under field conditions. Utilizing lower UV intensities corresponding to the ones present on the field might come with the challenge that other environmental factors such as ambient temperature or evaporation of the test substances would complicate the interpretation of such results. Therefore, laboratory tests can give a good indication of a substance’s capacity of UV protection, however, field conditions cannot be precisely mimicked.

To this date, a few laboratory and field studies have been performed, evaluating the effectiveness of potential candidates for UV protection of phages and other biological control agents (Table 1). However, comparability of the different studies is limited, as experiments were performed under different UV wavelengths, irradiation doses or relying on natural sunlight. A study by Born et al. demonstrated, that under laboratory conditions, Erwinia phage Y2 can effectively be shielded from the detrimental effects of UV-C radiation by adding UV absorbing, natural substances. In this study, both complex substances such as plant extracts as well as defined compounds were tested. Among the plant extracts, 10% beetroot juice, 5% red pepper juice and 10% carrot juice were tested. From the category of protein rich substances, casein (10 mg/ml) and soy peptone (50 mg/ml) were examined. Furthermore, an aromatic amino acid mixture of phenylalanine, tryptophane and tyrosine in equal parts (5 or 50 mM each), astaxanthin (3 mg/ml) and Tween 80 (5%) were evaluated. All tested compounds showed a significant effect on phage stabilization under UV-C light, the effect of Tween 80 however being very low (Born et al. 2015). It must be noted, that, with the exception of the aromatic acid mixture and astaxanthin, the substances were evaluated under high concentrations unrealistic for the application under field conditions. Furthermore, the substances were evaluated under UV-C light. To confirm the potency of these UV-protective candidate substances, these experiments should be repeated under UV-B light, adjusting the UV dose to match field conditions, as described previously (Kaiser et al. 2019).

Balogh et al. have tested the effect of UV-absorbing substances under both greenhouse and field conditions. Here, phages against Xanthomonas campestris pv. vesicatoria (Agriphage, OmnilLytics, Inc., Salt Lake City, UT) showed increased efficacy in disease control by the addition of different combinations of pregelatinized corn flour (PCF, 0.25–0.5%), sucrose (0.5%), casecrete (0.5%) and skim milk (0.75%). The three different formulations significantly increased phage longevity on the plant surface. While this effect was pronounced in greenhouse trials, it was greatly reduced under field conditions. Additionally, it was observed, that a regimen avoiding treatments during peaks of irradiation lead to improved outcomes. Compared to the non-formulated phage preparation, disease incidence was lower, and phage titers on leaf samples measured 36 h post application were 1000-fold higher in the plots including phage formulation than in the non-formulated applications (Balogh et al. 2003).

Skim-milk based phage formulation was also tested by Iriarte et al. in a field study on control of bacterial spot on tomato. Phage application was performed at different time points during the day while UV-A and UV-B radiation were collectively monitored. Here, it was demonstrated that UV-A and UV-B radiation correlated with phage titer decrease, and skim milk (0.75%) + sucrose (0.5%) formulation reduced phage titer loss. Interestingly, the skim-milk + sucrose formulation also provided protection against phage titer loss induced by ambient temperature, fluorescent light and desiccation in addition to protection from UV-radiation (Iriarte et al. 2007).

Obradovic et al. found phage (Agriphage) against tomato bacterial spot formulated with skim milk (7.5 g/liter) + sucrose (5 g/liter) or casecrete (5 g/liter) + PCF (2.5 g/liter) + sucrose (5 g/liter) developed by Balogh et al. (Balogh et al. 2003), to be highly effective in field trials (Obradovic et al. 2004). On the other hand, later greenhouse trials on the same pathosystem showed inconsistent results after the single application of unformulated phage (Obradovic et al. 2005). While no direct comparisons can be drawn, the skim milk + sucrose seemed to have, at least in this case, stabilized phage effectiveness.

UV protective formulations for other biological control agents

Beside the studies evaluating substances for the protection of phages against UV radiation, some studies have been performed on other biological control agents such as bacteria, fungi or entomopathogenic viruses (Table 1). A review by Burges and Jones, focusing on the formulation of entomopathogenic viruses, bacteria and protozoa for biocontrol, provides a wide overview of various additives tested in formulations (Burges and Jones 1998). A broad list of potential UV protectants gives an overview on additional substances to be considered in future screenings for phage cocktail formulations. The mode of action of the protective substances varies from prevention of damaging wavelengths reaching the agents, to antioxidants and enzymes scavenging hyperactive oxygen.

Absorbents scavenge damaging UV wavelengths and only allow the passage of harmless wavelengths. Natural absorbents providing protection at low concentrations at 0.1% or lower include amino acids, such as tyrosine and tryptophane, B vitamins, such as folic acid, nitrogenous metabolic compounds, and ascorbic acid. However, the problem with many of these substances is that they would not be economically feasible for field application. Carbon products also provide good protection from UV light due the effect of carbon acting as an oxygen sink.

For some substances, both an effect on UV protection as well as alteration of spray application behaviors were observed. For example, molasses additionally influences the viscosity and drop size of tank mixes, while milk and albumin can also serve as wetters/stickers and humectants. Drop size influences the effectiveness of UV protection, as the thickness of protective substance surrounding the biocontrol agent particle will influence the protective effectivity (Burges and Jones 1998).

Entomopathogenic fungi have been of great interest as biocontrol agents and several studies have been undertaken to improve the spores’ viability under field conditions, including UV-B radiation. A study by Kaiser et al. tested a wide range of UV-B absorbing substances both under laboratory and field conditions. For this purpose, a comprehensive, effective screening method to evaluate UV-B absorbing substances under laboratory conditions was developed. The protocol adjusted irradiation time with UV-B sources in the laboratory to UV-B doses present on a typical treatment day in the field. In addition, protective substances were evaluated in field trials. A wide variety of substances was tested. Among these, the overall best identified candidates were humic acid sodium (10%), humic acid potassium (10%) and lignin (10%), showing good performance under both laboratory and field conditions, humic acid sodium being the most effective substance under field conditions (Kaiser et al. 2019). The study exemplifies how UV doses in the laboratory can be adjusted to meet realistic field conditions. This principal could be transferred to the testing of phages and protective candidate substances. In theory, the tested humic acids and lignin could present interesting candidates for phage cocktail formulation. However, it needs to be considered that the here tested concentration of 10% is far above the ideal range. Additionally, both humic acids and lignin are of brownish color and therefore unsuitable for spray application on apple flowers, as they might lead to discoloration of petals and hinder pollination.

In addition to lignins, minerals have been evaluated for their UV protective capacity. Arthurs et al. tested the capacity of minerals to shield granulovirus of the codling moth, Cydia pomonella, from solar radiation. Formulations containing kaolin clay (28.1 kg/ha) or calcium with boron (28.1 L/ha) provided effective disease control, however no difference to the control (phage + surfactant) was detected (Arthurs et al. 2008). Different clays (0.3–0.6%) were also evaluated for their protective capacity against UV-C in Klebsiella aerogenes. Some clays, in particular nontronite, potassium-montmorillonite, zinc-montmorillonite and ferric sulfate increased the bacterial survival rate significantly (Bitton et al. 1972). Furthermore, a formulation of Verticillium lecanii spores both containing UV-protective montmorillonite (1%) and humectant polyoxyethylene (1%) was tested by Lee et al. for the control of cotton aphids. Spore survival under UV-B and UV-C light could be greatly enhanced by the formulation, and in greenhouse trials, aphid densities were strongly reduced following application of formulated spores, while in non-formulated spores, aphid densities increased (Lee et al. 2006). Mineral substances could be of interest for the protection of phages against UV radiation. Additionally, it needs to be verified that these minerals do not change the visual appearance of apple flowers.

Apart from the potential candidates for UV protection found in the scientific literature, some commercial products designed to both protect plant foliage from sun-borne damage on days of intensive irradiation, and to prevent the photolytic degradation of certain plant protection agents, are available. Two of these products, SolarPROTECT (Renovita Wilen GmbH, Wilen b. Wil, Switzerland) and Lufix (Stähler Suisse SA, Zofingen, Switzerland) are based on pine terpenes, while SOLAR (Aqua Aid Europe B.V, Breda, The Netherlands) is based on titanium dioxide and plant nutrients. These commercially available substances are interesting candidates as additives for phage cocktails and should be experimentally evaluated.

In summary, the portfolio of tested, effective formulations protecting phages from UV-B radiation is limited. The identified substances were often applied at concentrations higher than desirable for commercial products. Therefore, systematic evaluation of these substances for phages and identification of additional substances, showing high effectivity at even lower concentrations, would be of high interest in scope of phage cocktail commercialization.

Table 1 Summary of different additives used for UV protection
Full size table

Surfactants

To exploit a phage cocktail’s maximal potential efficacy, addition of surfactants to the formulation should be considered. Surfactants, also called wetters, lower the surface tension of water droplets. Thereby, they facilitate a more evenly distributed spray application enabling better coverage, especially on hydrophobic surfaces, and facilitate improved adherence of water droplets on foliage. Additionally, they can reduce desiccation of the applied drops (Zeisler-Diehl et al. 2022). In phage cocktails, these surfactants could maximize the plant surface covered with phages during the spray application. Their characteristic of reduced droplet desiccation could additionally stabilize phages on plant surfaces.

Stickers can improve adherence of biological control agents to the phyllosphere, reducing activity loss in case of rain fall (Burges and Jones 1998). Different materials such as molasses, gums and resins can serve as stickers. Some stickers such as lignin, casein, flour, gluten, milk and molasses can also serve as UV protectants. However, it has to be considered, that some stickers may also increase viscosity spray droplet size, altering spray application dynamics (Burges and Jones 1998). Baculoviruses, GVs and NPVs are known to withstand wash-off by rain fall (Burges and Jones 1998). For polyhedral inclusion bodies of Baculovirus heliothis, it was experimentally determined that only 6% of activity was lost after simulated rainfall (Ignoffo et al. 1997). In a field trial, Iriarte et al. observed that even non-formulated phage persisted despite of rainfall. Additionally, phages formulated with skim milk or skim milk + sucrose persisted longer than the non-formulated ones (Iriarte et al. 2007). However, it is unclear if the addition of skim milk and sucrose lead to an increase in resistance to wash-off by rain in addition to protection against UV radiation. Behle et al. tested casein formulation (0.5%) of Bacillus thuringensis and found that the formulation provided protection from wash-off, allowing for higher residual insecticidal effectiveness compared to the unformulated control. Additionally, protection from light-dependent degradation was observed, although the effect was not significant (Behle et al. 1996). Therefore, casein might also be a candidate to improve adherence of phages to the phyllosphere, in addition to its potential protection against UV-B light.

Burges and Jones provide an extensive overview of wetters and stickers tested with different biocontrol agents. Most baculoviruses were shown to be compatible with most commercial surfactants (Burges and Jones 1998). During the selection of wetters and stickers for field trials including phages, the stability of phages in presence with these additives should be monitored. Additionally, a compromise in drop size needs to be found for the final application. While decreased drop size facilitates even spreading, the capacity of UV protection could be impaired with decreased droplet size, providing less shielding to the active agent in the solution (Burges and Jones 1998).

Agriphage also holds a compatibility list, available on their website, of phages with different plant protection agents, in which all surfactants except one (composed of yucca and garlic extract) showed good compatibility with phage. On the other hand, it was reported that some surfactants decreased viral stability and activity (Chattopadhyay et al. 2002). In summary, surfactants considered to be used in field applications in combination with phage cocktails should be evaluated for their effects on phage virulence and stability under laboratory conditions first.

Conclusion

Unfortunately, field trial data demonstrating the efficacy of phages in fire blight control, to optimize inoculation and application parameters, and to study the effectiveness of a particular formulation is scarce. The trials are very time consuming and dependent on the seasonal environmental conditions, which may vary excessively in a season. Hence, a collaborative research project should be initiated to study the efficacy of well described phages in field trials around the world using predefined comparable setting. Such a project should ideally cover several years to allow enough repetitions for sophisticated statistical analyses.

Author-recommended Internet Resources

https://www.agriphage.com/wp-content/uploads/2021/11/Compatibility-List_11-15-2021.pdf.

https://ictv.global/taxonomy.