Archives of Virology

, Volume 159, Issue 9, pp 2253–2261

Efficiency of bacteriophage therapy against Cronobacter sakazakii in Galleria mellonella (greater wax moth) larvae


  • Reza Abbasifar
    • Canadian Research Institute for Food SafetyUniversity of Guelph
  • Andrew M. Kropinski
    • Department of Molecular and Cellular BiologyUniversity of Guelph
    • Laboratory for Foodborne ZoonosesPublic Health Agency of Canada
  • Parviz M. Sabour
    • Agriculture and Agri-Food CanadaGuelph Food Research Centre
  • James R. Chambers
    • Agriculture and Agri-Food CanadaGuelph Food Research Centre
  • Joanne MacKinnon
    • Laboratory for Foodborne ZoonosesPublic Health Agency of Canada
  • Thomas Malig
    • Laboratory for Foodborne ZoonosesPublic Health Agency of Canada
    • Canadian Research Institute for Food SafetyUniversity of Guelph
Original Article

DOI: 10.1007/s00705-014-2055-x

Cite this article as:
Abbasifar, R., Kropinski, A.M., Sabour, P.M. et al. Arch Virol (2014) 159: 2253. doi:10.1007/s00705-014-2055-x


Cronobacter sakazakii, an opportunistic pathogen found in milk-based powdered infant formulae, has been linked to meningitis in infants, with high fatality rates. A set of phages from various environments were purified and tested in vitro against strains of C. sakazakii. Based on host range and lytic activity, the T4-like phage vB_CsaM_GAP161, which belongs to the family Myoviridae, was selected for evaluation of its efficacy against C. sakazakii. Galleria mellonella larvae were used as a whole-animal model for pre-clinical testing of phage efficiency. Twenty-one Cronobacter strains were evaluated for lethality in G. mellonella larvae. Different strains of C. sakazakii caused 0 to 98 % mortality. C. sakazakii 3253, with an LD50 dose of ~2.0 × 105 CFU/larva (24 h, 37 °C) was selected for this study. Larvae infected with a dose of 5 × LD50 were treated with phage GAP161 (MOI = 8) at various time intervals. The mortality rates were as high as 100 % in the groups injected with bacteria only, compared to 16.6 % in the group infected with bacteria and treated with phage. Phage GAP161 showed the best protective activity against C. sakazakii when the larvae were treated prior to or immediately after infection. The results obtained with heat-inactivated phage proved that the survival of the larvae is not due to host immune stimulation. These results suggest that phage GAP161 is potentially a useful control agent against C. sakazakii. In addition, G. mellonella may be a useful whole-animal model for pre-screening phages for efficacy and safety prior to clinical evaluation in mammalian models.


Cronobacter sakazakii, known as yellow-pigmented Enterobacter cloacae prior to 1980, and Enterobacter sakazakii until 2007, is a ubiquitous, opportunistic pathogen found in both the environment and a variety of foods [14]. Although the epidemiology and reservoir of this bacterium are unknown, it has been shown that contaminated milk-based powdered infant formulae were the source of sporadic cases and outbreaks of Cronobacter infections, which cause sepsis, necrotizing enterocolitis, brain abscess and meningitis in neonates and infants [416]. Cronobacter spp. have been ranked by the International Commission on Microbiological Specifications for Foods [17] as a “severe hazard for restricted populations, life-threatening or substantial chronic sequelae or long duration.”

The mortality rate for neonate/infant infections with this bacterium can approach 80 %, with antibiotics not having a dramatic affect on clinical outcomes [15, 18, 19]. Almost all patients surviving infections of the central nervous system (CNS) experienced delays in mental and physical development [16]. In addition to its effects on neonates and infants, this nosocomial pathogen has been reported to produce rare instances of infections such as urosepsis, bacteremia and pneumonia in adults, especially in the elderly [16, 2024].

Consideration of bacteriophages (phages) as alternative control agents has been rekindled because of their high specificity and effectiveness in killing bacterial pathogens, including those which are antibiotic resistant, without affecting the host commensal microbiota. Phage biocontrol of pathogens including Salmonella, Campylobacter and Listeria in the food industry is recognized in the US and Europe [2531] and may be particularly useful to control Cronobacter because of its intrinsic antibiotic resistance [16].

Several studies have confirmed the efficacy of phage therapy in animal models for the treatment of bacterial pathogens such as Pseudomonas aeruginosa [3234], Escherichia coli [3539], Staphylococcus aureus [4042], Klebsiella pneumoniae [43, 44], and Campylobacter jejuni [45]. In general, mammalian models, such as mice, have been favored to study human infectious diseases due to their similarity to humans in anatomy, physiology, immune response, and pathogen susceptibility. There are, however, disadvantages in the use of these models, including high monetary costs, ethical issues, and difficulty in obtaining sufficient numbers of animals to achieve meaningful results [46]. As an alternative, invertebrates including insects have been used in pathogenesis studies [47, 48]. Some of the benefits of using insects as experimental models to study human pathogens include i) the possibility of using large numbers of individuals, ii) short life cycle; iii) ease of manipulation (reducing time and cost of maintenance), iv) quicker infection process, leading to more-rapid results, v) fewer ethical concerns related to the administration of pathogens [47], and vi) similarity between humans and insects in infection and immune responses, such as phagocytosis and production of antimicrobial peptides [49].

Among the invertebrates used, Galleria mellonella, the greater wax moth, is of special interest because of its ability to survive at 37 °C (important in studying temperature-sensitive virulence of pathogens), and its relatively large larval size, which allows inoculation of precise quantities of pathogen by syringe [47]. Like the mammalian system, Galleria possesses a complex innate immune system that includes phagocytic hemolymph cells. Furthermore, the humoral response includes the inducible production of lysozyme and low-molecular-weight cationic antibacterial peptides. Lastly, “regardless of the bacterial species, results obtained with Galleria larvae infected by direct injection through the cuticle consistently correlate with those of similar mammalian studies: bacterial strains that are attenuated in mammalian models demonstrate lower virulence in Galleria, and strains causing severe human infections are also highly virulent in the Galleria model” [50].

The objective of this study was to isolate specific phages against C. sakazakii and to evaluate their therapeutic efficacy and safety against this bacterium using the G. mellonella larva as a whole-animal model. To date, a few Cronobacter (Enterobacter sakazakii) phages have been isolated and characterized [5159]. However, there is little published evidence showing the effectiveness of phage therapy for Cronobacter infections in animal models; only one study was published on phage therapy of Cronobacter turicensis causing urinary tract infections in mice [60]. To our knowledge, the current study is the first phage therapy against C. sakazakii in an animal model.

Materials and methods

Bacterial strains and culture conditions

Twenty-one Cronobacter strains (including 14 C. sakazakii strains) used in this study were obtained from the culture collection maintained by the Canadian Research Institute for Food Safety (CRIFS), Franco Pagotto (Public Health Agency of Canada, Ottawa, ON) and Roger Stephan (Institute for Food Safety and Hygiene, Zurich, Switzerland). Cronobacter muytjensii 51329 was purchased from the ATCC (Manassas, VA, USA). Cronobacter strains were surface plated onto tryptic soy agar (TSA; Fisher Scientific, Ottawa, ON) and incubated at 37 °C. Colonies from 18-hour cultures were collected aseptically and added to SM buffer (100 mM NaCl, 50 mM Tris-HCl, pH 7.5, 0.002 % (w/v) gelatin, 8 mM MgSO4·7H2O) in order to prepare a bacterial inoculum of ~1.0 × 1010 CFU/mL. The inoculum was then serially diluted in SM buffer. These suspensions were used for experimental infection.

Bacteriophage isolation and characterization

Phage GAP161 was isolated from untreated sewage from the Guelph Wastewater Treatment Plant, Guelph, ON, Canada) [61] by enrichment with clinical isolate C. sakazakii HPB 3253 as the host. Samples were centrifuged (5000 × g for 15 min), and the supernatant was filtered through a 0.45-μm filter (Corning Inc., NY, USA). Twenty milliliters of an 8-h inoculum of Cronobacter isolates at 37 °C in tryptic soy broth (TSB; Fisher Scientific, Ottawa, ON, Canada), was added to 200 mL of filtered supernatant, plus 20 mL of bacteriophage broth and 20 mL of TSB containing 2 mM CaCl2. This mixture was incubated at 37 °C overnight and centrifuged at 5000 × g for 15 min at 4 °C (Beckman Avanti J-20 XPI, Beckman Coulter Inc., Mississauga, ON, Canada), and the supernatant was treated by adding 1 % v/v chloroform. Bacteriophage broth was prepared as described previously [61] by adding 100 g peptone (Difco Laboratories, Detroit, MI, USA), 30 g beef extract (Difco), 50 g yeast extract (Fisher), 25 g NaCl (Fisher), and 80 g potassium dihydrogen phosphate (BDH Laboratory, Toronto, ON, Canada) to one liter of distilled water.

For isolation of the phages, the overlay method [62] was used, with molten (47 °C) top agarose (5 g low-melt agarose/L TSB containing 2 mM CaCl2). After overnight incubation at 30 °C, the plates were examined for the presence of plaques. Single plaques were picked and eluted in SM buffer and re-plated. This was repeated three times to achieve purified phages. Selected phages (based on the host range) were examined by electron microscopy. The phages were negatively stained with 2 % uranyl acetate for 30 s and examined by electron microscopy (energy-filtered TEM, EFTEM, LEO 912AB model operated at 100 kv, Carl Zeiss, Jena, Germany).

Phage GAP161 was selected, and the phage preparation was purified using PEG precipitation and two rounds of CsCl equilibrium gradient purification centrifugation as described [63]. The phage was dialyzed against SM buffer using dialysis cassettes with a 3500 molecular weight cutoff (Thermo Scientific, Fisher Scientific, Mississauga, ON). The phage titre was determined by the overlay method, and the phage was stored at 4 °C prior to being used in the wax moth larvae experiments.

G. mellonella assays

Larvae were purchased from a commercial supplier (Recorp Inc., Georgetown, ON, Canada), and those weighing between 200 to 300 mg were selected. They were stored in wood chips and paper towel at 12 °C. In order to reduce the stress of temperature change, larvae were kept at 37 °C for 24 h prior to the experiment. Bacteria or phage (4-μL aliquots), when applicable, were injected into the hemolymph at the base of the second set of thoracic legs using disposable 1-mL tuberculin syringes, 32-gauge needles, and a micro-injector [64]. Phages were injected in the right side, and bacteria in the left, where applicable. Following injection, G. mellonella larvae were immediately placed in an incubator at 37 °C in the dark. To study the virulence of C. sakazakii in G. mellonella, fresh inoculum (~1.0 × 1010 CFU/mL) of 21 different Cronobacter strains was injected into the larvae and scored 24 h postinfection (p.i.) at 37 °C. Larvae were scored as dead when they had no reaction to touch with thumb forceps. To determine the 50 % lethal dose (LD50), a series of 10-fold serial dilutions from 4 × 107 to 0 CFU of C. sakazakii strain HPB 3253 in SM buffer were injected into larvae. A control group received 4 μL of only SM buffer to assess any potentially negative effects of the injection process. Three groups of ten larvae were injected for each dilution, and larvae were incubated at 37 °C, monitored periodically, and scored as alive or dead up to 48 h p.i. Each experiment was repeated three times independently, and the LD50 value for G. mellonella was calculated using the method of Miller and Tainter described by Randhawa [65].

In order to test the persistence of phage GAP161 in larval hemolymph over time, the phage was injected into larvae (4 μL, 2.0 × 109 PFU/mL), and the titer of phage in hemolymph was measured at 0, 24 and 48 h after injection. For the zero time point, hemolymph was collected from the larvae 20 min after injection. At each time point, 20 μL of hemolymph was collected from 10 larvae, combined in a microcentrifuge tube and serially diluted. To determine the phage titre, three groups of 10 larvae were used for each time point, and the titre was determined by the overlay method [62].

For the challenge trials, 4 μL of bacteria at the concentration of 5 × LD50 was administered into the hemolymph (~1 × 106 CFU/larva). In the treatment groups, larvae were injected with 2.0 × 109 PFU/mL of phage GAP161 (~8.0 × 106 PFU/larva), resulting in a multiplicity of infection (MOI) of 8. This MOI was chosen based on the results of preliminary trials (results not shown). Larvae infected with C. sakazakii 3253 (Infected Group) were immediately treated with phage GAP161 by separate injection (time 0). Furthermore, in order to determine the prophylactic effect, phage was injected into larvae at 1 h and 0.5 h prior to the time of infection. Also, to determine the length of time that treatment could be delayed, the phage was administered at various intervals up to 4 h after bacterial infection (time +1 h, +2 h, and +4 h). The untreated group received 4 μL of SM buffer instead of phage.

In addition, there were three groups of uninfected larvae, which received i) no injection (not infected, not injected with phage or buffer; Nil group), ii) one injection of SM buffer following an injection of phage to determine any potential negative effect of the virus (phage group), or iii) two injections of SM buffer to measure the effect of multiple injections on the larvae (sham group). Following injection, G. mellonella larvae were immediately incubated at 37 °C for 48 h and monitored periodically, at which times the larvae were scored as alive or dead. In the third experiment, survival was measured at 23.5 h instead of 24 h.

To verify that the effects of phage therapy were related to a nonspecific immune activation response, and not the phage itself, a control group was tested with heat-inactivated phage to determine the ability to rescue larvae infected by C. sakazakii. Phage GAP161 (2.0 × 109 PFU/mL) was inactivated by heating at 85 °C for 30 min. Following this heat treatment, no viable phage was detected using the overlay method. Larvae were injected with a 5 × LD50 dose of C. sakazakii 3253 (~1.0 × 106 CFU) and immediately treated with heat-inactivated phage at an MOI of 10.

Statistical analysis

In all experiments conducted in this study, each group had a total of 30 larvae and consisted of three replicates, 10 larvae per replicate, and every experiment was repeated three times independently. Survival data were analyzed using SAS/STAT version 9.2 (SAS Institute Inc., Cary, NC, USA). The GLM procedure with “Experiment, Time, Treatment and Treatment*Time” in the model was used to evaluate these effects (survival at 24 and 48 h). Of the many comparisons of the various treatments that could be made, single degree of freedom contrasts involving pairs of or groups of treatments were used to examine specific effects. In addition, survival curves were plotted by the Kaplan-Meier method using GraphPad Prism (GraphPad Software, Inc., La Jolla, CA, USA).


Properties of Cronobacter phage vB_CsaM_GAP161

Phage vB_CsaM_GAP161 was very active against Cronobacter strains, lysing 64.3 % of the C. sakazakii strains tested, half of C. malonaticus, plus all the C. dublinessis and C. muytjensii strains tested (Table 1). It showed the broadest host range and strongest lytic activity of all the phages that we have characterized. GAP161 is a member of the family Myoviridae, displaying an elongated cylindrical head 110 nm long by 74 nm wide, and a tail 113 nm long by 17 nm wide (Fig. 1). Its 178-kb genome, encoding two tRNAs and 275 proteins, is peripherally related to enterobacteria phage RB16 [57]. Based on tests of its broad host range and strong lytic activity, phage GAP161 was selected as a therapeutic agent.
Table 1

The host range of Cronobacter phage GAP161 against 21 Cronobacter strains based on the spot test (16 h incubation at 30 °C)

Phage isolates


C. sakazakii 2855

C. sakazakii 2870

C. sakazakii 2871

C. sakazakii 2876

C. sakazakii 3253

C. malonaticus 3263

C. sakazakii 3199

C. muytjensii 51329

C. sakazakii 3290

C. dublinensis 3169

C. malonaticus 3267













Phage isolates


Lytic score

C. muytjensii 3270

C. universalis 3287

C. sakazakii 130/3

C. sakazakii 236/04

C. sakazakii 324/04

C. sakazakii 354/03

C. sakazakii 974/03

C. sakazakii 1084/04

C. sakazakii 1103/03

C. malonaticus 1154/04













The results were recorded as 3 (very clear zone of complete lysis on the bacterial lawn), 2 (clear/turbid zone of lysis), 1 (turbid lysis) and 0 (no lysis)
Fig. 1

Electron micrograph of phage vB_CsaM_GAP161 negatively stained with 2 % uranyl acetate. The bar represents 100 nm

Cronobacter-Galleria-phage testing

Our results showed that Cronobacter strains, some derived from the International Life Science Institute North America Cronobacter (Formerly Enterobacter sakazakii) isolate set [66], differed in virulence, ranging from 0 to more than 98 % mortality in larvae when injected with 4 × 107 CFU (Fig. 2). Mammalian studies will be required to verify whether the results of the two systems correlate.
Fig. 2

Mortality rates of different Cronobacter strains compared with SM buffer or no injection in G. mellonella larvae 24 h p.i. at 37 °C (~4.0 × 107 CFU) with standard deviations

To study the effect of bacteriophage on C. sakazakii infection during 48 h incubation at 37 °C, the C. sakazakii HPB 3253 strain was selected as the infectious agent because of its high virulence and its sensitivity to bacteriophages.

C. sakazakii 3253 infection caused severe illness and death in G. mellonella larvae injected with an LD50 dose of approximately 2 × 105 CFU (Fig. 3). At a dose of 5 × LD50, the mortality rates were up to 96.6 % and 100 % in the control infected groups at 24 and 48 h p.i., respectively (Table 2; Figs. 3, 4). Administration of phage GAP161 at an MOI of 8, in the opposite side of the larvae to the bacterial injection site, resulted in mortality as low as 16.6 % after 24 and 48 h p.i. The survival of phage-treated groups that received the phage before or immediately after infection was significantly higher than untreated groups (p ≤ 0.001). Treatment with the bacteriophage at 1, 2, and 4 hours postinfection did not reduce mortality. In general, survival rates observed at 24 hours were not significantly different from those at 48 hours. When the heat-inactivated phage was administered, no significant difference was observed between infected untreated larvae and those treated with the heat-inactivated GAP161, with both showing the same mortality rate of 83.3 % at 48 h p.i.
Fig. 3

G. mellonella larvae infected with C. sakazakii HPB 3253. A) Right to left: normal, sick, and dead. Images B and C show larvae 6 h after infection: B) Larvae treated with phage GAP161 immediately after infection, C) Larvae that received SM buffer immediately after infection

Table 2

Least square means (LSM) of the survival of G. mellonella larvae following challenge with C. sakazakii HPB 3253 and/or phage GAP161 administration










 24 h


 48 h



 Nil group


 Sham group


 Phage group


C. sakazakii


Sa Ph −1


Sa Ph −0.5


Sa Ph 0


Sa Ph +1


Sa Ph +2


Sa Ph +4


These results were obtained by using general linear model (GLM) in SAS

* Standard error = 4.49. LSM with different lower-case superscripts are different at P ≤ 0.05; those with different upper-case superscripts are different at P ≤ 0.001
Fig. 4

Survival of G. mellonella larvae infected with C. sakazakii and treated with phage GAP161 at different times prior to and after infection. Nil group, no injection; phage group, one injection of SM buffer following an injection of phage; sham group, two injections of SM buffer. Challenged groups were infected with C. sakazakii HPB 3253 and treated with phage GAP161 at various times before infection (−1 h and −0.5 h) or after infection (0 h, +1 h, +2 h, and +4 h)


The variation in virulence of different Cronobacter strains observed in this study, as determined by larval mortality, agrees with a previous study on pathogenesis of this bacterium [66, 67], which showed significant differences in adhesion, invasion, and toxin production among strains. Strain 2855 has a low ability, and strain 3290 a high ability, to adhere to and invade human brain microvascular endothelial cells. In our study, infection with the same strains resulted in 0 and 93 % G. mellonella larval mortality, respectively.

The survival rates of groups treated prior to or immediately after infection with phage GAP161 were significantly higher than the survival rate of the untreated infected control group. When heat-inactivated phage was used, the survival rates were the same as for larvae inoculated with the pathogen alone, indicating that the survival of larvae is not due to immune stimulation in the host but to the effect of phage antibacterial activity. The lack of significant difference among the nil, sham, and phage groups reveals that neither the phage nor the injection method used had an effect on the health of the larvae. Analysis of the complete genome sequences of phage GAP161 and screening of its proteome against a database of 83 bacterial toxin proteins have revealed the absence of toxin-encoding genes in phage GAP161 [57].

There were small but significant differences among the results of the three repeated experiments. These could be due to variations between batches of larvae due to the growth conditions such as temperature, moisture, and diet, as well as the age of the larvae. The effects of these differences were mitigated to some extent by choosing only healthy-looking larvae and sorting them by weight for use in experiments.

In our study, the phage was applied at various intervals prior to C. sakazakii infection in order to determine if this virus had a prophylactic effect. The results suggest that the time of administration of phage plays a key role in therapy of C. sakazakii infection in larvae. When GAP161 was applied 1 h before infection, the survival rates were higher than in the groups administered phage at 0.5 and 0 h pre-infection. The differences among these treated groups were not statistically significant, but the survival rate in all of the treated groups was significantly higher than in the untreated group. The longer the time that phage administration occurs postinfection, the greater the increase in the mortality rate. Although the survival rate in the group that received phage 1 h p.i. was higher than that of the infected untreated group, the differences were not statistically significant. Mortality rates were higher in the groups treated with phage at 2 and 4 h p.i. than in the untreated group, but the difference was not statistically significant.

The significant differences observed in the survival of animals treated with phage prior to or immediately after infection with Cronobacter confirm that GAP161 is highly efficient in preventing infection in larvae. Likewise, Seed et al. [68] showed that two out of three phages that were tested provided the best protection when applied immediately after infection with B. cepacia; however, treatments after 6 and 12 h resulted in fewer survivors. The higher mortality rates in groups treated with phage GAP161 later than 1 h p.i. is likely due to the fact that by the time the phage has been administered, the bacteria have already caused irreversible damage to the larvae. At the time of these injections, the animals were in an advanced stage of morbidity as visually manifested by a color change from yellow to gray, dark brown, or black, and signs of dehydration. In addition these moribund larvae were more sensitive to the invasiveness of the injection, as it caused bleeding of hemolymph.

The treatment of larvae infected with C. sakazakii with phage, helps the larvae to survive not only in the short term (24 h p.i.) but also for longer time periods, as there is no significant reduction in survival between 24 h p.i. and 48 h p.i. (Fig. 4; Table 2).

This study confirms earlier work illustrating the usefulness of G. mellonella larvae as an inexpensive animal model for evaluating the efficacy of phage therapy [68]. Greater wax moth larvae have also been employed to evaluate other antimicrobial compounds. For example, G. mellonella infected with Cryptococcus neoformans were treated with different antifungal agents [69]. The combination of amphotericin B plus flucytosine, which is the recommended treatment in humans for this fungal infection, provides the best protection in G. mellonella. These studies, along with our results, suggest that G. mellonella is a valuable model to study antimicrobial agents, including bacteriophages.


Our studies indicate that Cronobacter phage GAP161 lacks recognizable virulence determinants and is effective in reducing morbidity and mortality caused by Cronobacter in the Galleria system. Prior to its intended end use, it will be necessary to test its efficacy in controlling Cronobacter in infant formula, and to perform clinical evaluations in mammalian models.


The authors thank Hans-Wolfgang Ackermann for examination of the electron micrographs, and Franco Pagotto and Roger Stephan for providing bacterial strains. This work was financially supported by the Natural Sciences and Engineering Research Council of Canada (NSERC) and Dairy Farmers of Ontario (DFO). We thank June Chadwick and Peter Aston (Queen’s University, Kingston, ON) for the micro-injector and Gary Dunphy (McGill University, Ste. Anne de Bellevue, QC) for practical advice on handling and injecting Galleria larvae.

Conflict of interest

The authors declare that they do not have any conflict of interest.

Copyright information

© Springer-Verlag Wien 2014