Galleria mellonella as a Suitable Model of Bacterial Infection: Past, Present and Future

1Research on animal models is essential to get more information about human infections and host-pathogen interactions. Animal experimentations had been going on for a long time since Aristotle and Hippocrate were practicing animal dissections to elucidate how the human body functions. The rise of microbiology, and Robert Koch postulates have greatly enhanced the use of animal models to demonstrate the pathogenicity of microorganisms. Vertebrates such as murine models constitute traditional host-models for the study of human pathogens because of a high degree of similarity in metabolism, body temperature, or immune response. Currently, about 75 to 100 million vertebrates per year are employed for scientific purposes, primarily mice and rats. In recent years, voices have been raised to protest against the untimely use of animals in research. As a result, regulation has become considerably tougher. The first European directive on animal protection dates back to 1986 and is routinely reviewed. It is mainly based on the 3Rs rules: Replace, Reduce and Refine, as described by Russel and Burch. Ethical rules require authorizations that can induce considerable delays in approval. In addition, there are other challenges facing the scientific research: working on vertebrate models needs specific training, adequate permits and equipment, animal adaptation times and had significant costs. So, the use of a large number of mammals can be challenging for financial, technical and ethical reasons. One of the outcomes is the rationalizing of the number of animals that could result in unreliable and unpredictable data. This legislation on the animal conditions and welfare is necessary, and requires the scientific community to adapt. The in vitro approach to examine human pathogens is not suitable and does not mimic the natural niche. It is demonstrated that the expression of virulence factors is not the same between in vitro growth, including media mimicking the host’s environment, and in animals or humans. This is true for Staphylococcus aureus, a major human pathogen, but also for other bacterial species. Consequently, extrapolations from in vitro to more complex biological conditions are therefore subject to limits of interpretation. As an alternative, scientists are increasingly conducting animal experiments with alternative models such as invertebrates to allow easier determination of host effects. Various invertebrate models exist and are used to study host-pathogen interactions such as the fruit fly Drosophila melanogaster, the nematode Caenorhabditis elegans, and the greater wax moth Galleria mellonella. These invertebrates do not have nociceptors and are thus insensitive to pain; consequently, they are no restrictive ethical rules unlike for vertebrates. Their low cost as well as their ease of use contribute to their expanding popularity. Although lacking an adaptive immune system, invertebrates share with vertebrates a large number of orthologic genes responsible for general functions thus explaining that their innate immune system is similar. G. mellonella, the most recent invertebrate model has been less studied than the others, nevertheless its use as a model for deciphering virulence and antimicrobial efficacy is increasing. It is now considered that this larvae model constitutes a suitable model for studying human pathogens. This review aims to describe the current knowledge of the G. mellonella model and points out directions for standardization and new perspectives for its utilization.

2The insect G. mellonella, also called honeycomb moth or greater wax moth, is part of the Lepidopetra order, the Pyralidae family, and the Galleriinae subfamily. It is described worldwide where beekeeping is practiced, and lives naturally in beehives where it feeds on wax and pollen, causing bee galleriosis. This is a typical holometabolous insect with a full metamorphosis cycle whose larval stage (caterpillars) is of interest to the scientific community as a model of infection.

3Four distinct live stages are described: eggs, larvae, pupae and adults ( Figure 1 ), with an approximate duration of a complete cycle of 8 weeks at 29-39°C and high humidity. Complete metamorphosis is affected by both biotic and abiotic factors including competition for food, diet quality, cannibalism, temperature and relative humidity.

4Initially, during oviposition, females lay 50-150 clustered-eggs whose hatching varies with temperature. Indeed, there is a better development at warm temperatures than at cold ones, and eggs do not tolerate extreme temperatures. They have a spheroidal shape, and are white to light pink color. The development into larvae is temperature dependent, and last between 3 and 30 days. Indeed, at a temperature comprising between 24-27°C, the duration is 3-8 days whereas at 10-16°C, the duration is about 30 days. Larvae measure 1-23 mm, are creamy-white with a reddish head and became darken as it grows. They undergo about 8-10 molting stages. The larva consists of a number of segments, divided into 3 anatomical parts: head, thorax, and abdomen. Six legs are found at the thoracic part and prolegs are present on the abdominal area ( Figure 2A ). Anatomically, the larval integumentary system comprises a thick coating called cuticle, under which is found a thin epithelial layer ( Figure 2B ). The inner cavity consisted of the fat body as well as the hemolymph, i.e. the larval circulatory system. The digestive system and the silk gland are located in the fat body ( Figure 2C ). The dorsal region, considered to be crucial by coordinating the immune response, is called the “new immune tissue”. The ventral region corresponds to the nervous system, and consists of several ganglia ( Figure 2C ). Larvae at last instar produce silk which is used to form cocoons. At this moment, corresponding to the stage of pre-pupae, the larvae stop eating and become less mobile. Pupae are immobilized in cocoons and do not eat during this period. They are initially white to yellow then turn brown and finally take a dark reddish color. Adult moths color varies from reddish-brown to pale cream color. They are sensitive to light, and they live usually at night. Like the pupae stage, adult moths do not feed. Male and female adult moths have many differences. Indeed, males are slightly smaller and lighter in color than females, and they live 21 days whereas females live 12 days.

5The development of next-generation sequencing (NGS) has led to a better understanding of microbial communities, referred as microbiota, in both humans and animals. It is now established that the microbiota constitutes an interface with the host, particularly with its immune system, and is a source of many interactions between bacteria. The G. mellonella larvae microbiota is poorly characterized, and few data are available. The first studies to investigate the larval microbiota described it as very simple with a main species, Enterococcus faecalis. By studying the microbiome composition on various body sites of G. mellonella larvae, Allonsius et al. showed a predominance of Enterococcus species, as well as slight variation of bacterial composition of hemolymph, fat body, skin or feces. They also provided differences between two groups of larvae according their origin: bait larvae (i.e. larvae used as reptile food or fishing bait, commercially available), or research grade larvae (i.e. larvae whose rearing and storage conditions are well defined with standardized procedures). Bait larvae presented a greatest microbial diversity in the hemolymph and on the skin, probably due to previous treatments with hormones or antibiotics comparing to research grade larva. This hypothesis is in agreement with Krams et al., who report that diet diversity influence the gut microbiome composition. The different larval metamorphosis modified the composition of the larval digestive microbiota: more the development stage was advanced, more Enterococcus species were predominant, and conversely, Enterobacterales and staphylococci became undetectable. This is an important concept because G. mellonella larvae were used at the final instar-stage in virulence studies of microorganisms.

6Like other insects, G. mellonella lacks an adaptive immune system but its innate system shares many similarities with that of mammals. It includes a cellular response in which hemocytes – immune cells close to mammal neutrophils – are key actors responsible for cellular events, and a humoral response with soluble effector molecules ( Figure 3 ).

7Hemocytes are predominantly found in the hemolymph, the analogue of the mammalian bloodstream, but are also subcuticular, scattered in the fat body and around the digestive tract. Hemocytes concentration fluctuates during life, and is also affected by stress caused by microorganisms. During an infection, activated hemocytes migrate to the injured site. Broadly, insects possess many types of hemocytes with particular morphological, histological and functional features. At least, six were identified within G. mellonella, called prohemocytes, plasmatocytes, granular cells, coagulocytes, sphelurocytes and oenocytoids. They are implicated in different physiological functions: phagocytosis, nodulation, encapsulation, clotting, melanization, and are linked to the humoral response through soluble effector molecules. Prohemocytes are progenitor cells, having the property to differentiate into several cell types. Plasmatocytes and granular cells are the predominant immune cells, and are key members of cellular immunity due to their role in phagocytosis, nodule formation and encapsulation. Phagocytosis first involves granulocytes, which use opsonins to recognize microorganisms responsible for degranulation, thus allowing plasmatocytes to adhere to pathogens. Complete phagocytosis provokes killing of pathogen by mechanisms including production of reactive oxygen species (ROS). Nodulation is mediated by the cooperation of hemocytes, which leads to the formation of several successive layers of immune cells surrounding recognized pathogens, resulting in aggregates known as nodules. Next step is activation of the melanization process inside the nodules, responsible for death of foreign bodies. Encapsulation concerns large microorganisms such as nematodes or protozoa, but is not reported for bacteria. Encapsulation process involves granular cells and plasmatocytes that interact in sequential steps: granular cells recognize and attach to microorganisms leading to the release of specific peptides that then attract plasmatocytes. This process leads to the formation of superimposed hemocyte cells surrounding pathogens, which are destroyed by the release of specific molecules by the two cell types. Coagulocytes are involved in the hemolymph coagulation (clotting) corresponding to the first line of defense after exposure to a pathogen. Spherulocytes corresponding to non-adhesive cells that transport and secrete several cuticular components but their function is still little explored. Oenocytoids contain and secrete precursors of the phenoloxidase, the pro-phenoloxidase, and are involved in the melanization pathway. In addition, like mammal neutrophils, they are able to secrete extracellular nucleic acid traps which are involved in microorganism sequestration and coagulation activation.

8G. mellonella produces many pathogen-associated molecules opsonins that target bacteria cell-wall components such as lipopolysaccharide (LPS) or lipoteichoic acid (LTA). At least, four opsonins classes are described: apolipophorin-III (apoLp-III), peptidoglycan recognition proteins (PRGPs), cationic protein 8 (GmCP8) and hemolin. ApoLp-III, a molecule involved in the lipid transport, is the most characterized, and acts as a pleiotropic effector of larval humoral immunity. ApoLp-III is both engaged in the production of ROS and promotes the synthesis of antimicrobial peptides. In insects, PRGPs are involved in the hydrolysis of bacterial peptidoglycan. Six of them were identified in G. mellonella using transcriptomic data, but their specific functions are not elucidated yet. Similarly, others opsonins such as GmCP8 and hemolin are known to bind LPS or LTA, but their functions are unclarified to date. Nevertheless, hemolin concentration in hemolymph increases during S. aureus infection, suggesting a putative role in immunity.

9Antimicrobials peptides (AMPs) are ubiquitous components that play a major role in G. mellonella innate immunity. AMPs repertory consists of 20 peptides, at least, whose the common feature is to possess a broad spectral microbial activity. They are found in hemocytes, fat body, salivary glands, reproductive and digestive tracts, at different concentrations both in non-infected and infected larvae. Two types of AMPs are described, namely anionic and cationic antimicrobials.

10Cationic peptides are the most characterized and the main AMPs found within G. mellonella. They are divided into 3 groups according to their structure: linear α−helical peptides (e.g. cepropins, moricins), peptides with disulfide bridges (e.g. gallerimycin, galiomycin) and peptides with proline and/or glycine residues (e.g. Gm proline-rich peptide 1, gloverin). AMPs actions result in a unique effect: the induction bacterial cells lysis by ion leakage, involving different mechanisms. For example, cecropins and moricins lead to the formation of cytoplasmic membrane pores whereas gloverin or Gm proline-rich peptide increase membrane permeability by blocking the synthesis of vital membrane proteins. As a common characteristic, they exhibit a broad antibacterial spectrum, even if some of them act preferentially on Gram-negative or Gram-positive bacteria or on filamentous fungi. These observations are in agreement with the specific activation of the larval immune in response to a determined pathogen.

11Anionic peptide 1 and 2 (AP1, AP2) are to date the only known member of anionic AMPs. AP2 is found in the hemolymph of unstimulated larvae as well as infected larvae at a high and constant level, and possesses a low activity against Micrococcus luteus and yeasts.

12Many human pathogens secrete a wide range of metalloproteinases especially thermolysin-like metalloproteinases from the M4 family, which cleave proteins implicated in the immune response. G. mellonella produces an insect metalloproteinase inhibitor (IMPI) that is the only microbial-induced metalloproteinase inhibitor identified to date. IMPI can inhibit thermolysin-like metalloproteinases, and contributes to the innate immune response by counteracting secreted virulence factors.

13Lysosymes were the first antimicrobial proteins that have been described from Galleria. They constitute a family of muramidase-like proteins able to hydrolyze strongly the peptidoglycan, of Gram-positive bacteria but moderately in Gram-negative bacteria. Nevertheless, they act synergistically with others opsonins such as apoLp-III, leading to an increase of damages in E. coli. Lysozymes are present in hemolymph of non-infected larvae and their concentration increase in the presence of foreign bodies. They modulate the larval microbiota since a significant increase of Enterobacterales is observed in larvae knocked-down for lysozyme production.

14During the infection process, soluble effectors molecules bind LPS or LTA bacterial components, inducing the release of pro-phenoloxidase by oenocytoids that are then activated in phenoloxidase (PO) through the serine protease cascade. PO oxidizes the phenolic compounds into quinones that are then metabolized into melanin, explaining the black spots in infected larvae. Melanization pathway participates in the antimicrobial activity against bacteria or fungi. The degree of melanization seems to be a dependent on both strain virulence and inoculum. On one hand, enteroaggregative E. coli triggered the melanization process while non-pathogenic E. coli did not. On the other hand, larvae infected with a S. aureus strain at 106 UFC were rapidly and totally melanized whereas with 104 UFC, there is no sign of melanization.

15Since invertebrates are deprived of T and B lymphocytes and do not produce antibodies, the lack of adaptive immunity apparently limits the relevance of this model to study microorganism virulence. Nevertheless, a specific response called ‘trained immunity’ has been described, which consists of an infection containment mediated by hemocytes and AMPs. This mechanism is evidenced by the infection at sub-lethal concentrations followed by exposure to higher concentrations of the pathogen, resulting in more resistant larvae. This immune-like memory has not been clearly elucidated yet but it demonstrates the complexity of the immune response to foreign microorganisms. In G. mellonella, this mechanism has been described in particular in larvae-fungi interactions. The two most described AMPs, gallerimycin and galiomycin, play a role in this pre-immunization since it was shown that their concentrations increased after stimulation with a non-lethal dose of Candida albicans. It was also demonstrated that pre-immunization of G. mellonella larvae with LPS or heat-killed Listeria monocytogenes enhanced larval survival by pre-immune activation inducing an increase of hemolymph antimicrobials. Results were similar when larvae were pre-immunized and infected with Klebsiella pneumoniae. Infection by 105 CFU of S. aureus, but not 104 CFU, induces melanization and death. This tolerance effect would be related to primary immunization by staphylococci as recently confirmed.

16Here, we describe and discuss several studies implying the G. mellonella model as an alternative infection model, focusing on relevant human pathogen bacteria. Frequently, this larvae model is used as a screening model because of multiple advantages described above. For example, to test the hypothesis that a gene or a system contributes to virulence, research teams monitor larval mortality over time with both the wild type strain and the mutated one, allowing for fast results. Our objective is using methods currently available, to focus on host pathogen interactions in the G. mellonella model. Virulence assessment can be performed according to several approaches: clinical observations, bacterial burden, immune response activation through AMP expression, hemocyte density, phenoloxidase activity, histopathological data and monitoring of bacterial gene expression ( Table 1 ).

17Deciphering of intracellular bacteria interactions with the host is a major part of the acquisition of knowledge on virulence. The G. mellonella model was assessed to investigate these interactions. Consistencies between mammalian models and G. mellonella were observed with Legionella pneumophila infected larvae. The system Dot/Icm type-4 secretion system (T4SS) responsible for the translocation of proteins into host cells was determinant for the pathogenesis because of there was no mortality with a T4SS-deficient ΔdotA mutant. Indeed, no internalized bacteria were visualized and the mutant was quickly cleared with no replication. In contrast, infection caused by wild-type strain resulted in a rapid proliferation, an activation of the immune response implying hemocytes aggregates and nodule formation, an intra-hemocyte survival and bacterial multiplication within vacuoles. One of the effectors of the Dot translocation system, SdhA, was also crucial for the virulence since a knockout strain was rapidly killed resulting in reduced larvae mortality. These results were then confirmed in a murine lung infection model. The G. mellonella model was also shown to be a reliable surrogate model for the virulence study from another intracellular bacterium, L. monocytogenes. Authors demonstrated similarities in host-L. monocytogenes interactions between the G. mellonella model and vertebrates: intracellular bacteria were detected, larval mortality was correlated with bacterial growth, and the pathogenicity island vgc encoding the major virulence determinants was essential for the successful of the infection. Host-pathogen interactions were also highlighted by a rapid activation of immune-responsive genes such as gallerimycin and IMPI. Same authors revealed that this bacterium can infect brain larvae with evidence of melanization containing bacteria in the cerebral tissue. Furthermore, they identified an activation of stress and neuronal repair markers following the infection thus suggesting a connection between host-response and brain damage. In the last years, involvement of microRNAs (miRNAs) into host-bacteria interactions had emerged, and miRNA profile expression pattern could be altered during an infection, resulting in a different expression pathway of targeted mRNAs that regulate the host immune response. In the G. mellonella host, L. monocytogenes was able to modify miRNA expression with at least a different expression profile for 90 miRNAs of which mRNA-targets were identified for some of them. Some mRNA targets participate to the larvae defense response thus indicating a modulation of the immune response. For example, the miRNA miR-133 was downregulated after L. monocytogenes infection leading to an upregulation of its target, the MAP kinase system allowing an activation of the insect prophenoloxidase cascade.

18G. mellonella is also an alternative model to investigate enteric bacteria pathogens. Indeed, similarities are described between intestinal epithelial cells from larvae and from mammalian digestive tract. By an oral infection to respect the natural route of contamination, with a mixture of Bacillus cereus spores or vegetative cells and toxins, a synergistic effect was observed, mimicking a gastrointestinal infection. These results highlighted that spores resisted to stress induced by host larvae and could turn into vegetative bacteria explaining the increase of the bacterial count observed. Pathogenicity would be explained by the action of the toxin, enhancing the multiplication of bacteria in the gut, which then spread by breaking the intestinal barrier. Others enteric bacteria pathogens were tested in this model such as Yersinia pseudotuberculosis, Campylobacter jejuni, Yersinia enterocolitica, Shigella spp.. Surprisingly, the majority of studies did not practice oral infection but systemic infection by injected larvae with a standardized inoculum. Barnoy et al. testing the Shigella virulence by oral force feeding observed no clinical larval manifestations whereas larvae died quickly after injection. Results were similar with C. jejuni or Y. pseudotuberculosis infected larvae, which were also competent for an intra-hemocyte survival.

19All the studies described above prove an activation of the larval immune system regardless of the pathogen of interest. Some have shown a decrease in hemocyte count after infection but the fate of hemocytes is unknown, particularly for morphological and/or ultrastructural modifications or protein markers. This issue has been explored through the interaction between P. aeruginosa and hemocytes. Eighteen hours after the infection, hemocyte modifications were detected including swollen or naked nucleus, damaged organelles and condensed chromatin. All these alterations increased over time. LCB3 protein and caspase were detected, indicating the progression of hemocytes towards a programmed cell death or apoptosis. Conversely, no hemocyte variation was observed after a Streptococcus pneumoniae infection with larvae-killing doses and increased intra-hemocyte production of oxygen free radicals. Since this increase should lead to apoptosis and necrosis, a lower number should be expected. Authors hypothesized a steady-state between the production of hemocytes and their destruction by bacteria. Interplay between K. pneumoniae and larvae pointed out the immune response modulation for the benefit of bacteria that enabled coexistence within the host. Infection was followed by bacterial proliferation, signs of hemocyte damages, and decrease of hemocytes number. Curiously, no intra-hemocyte K. pneumoniae were detected, and the production of AMP was not activated. The same observation in experimental K. pneumoniae pneumonia in mouse macrophages, suggests an escape way to survive in the host.

20Available methods and techniques as well as increasing knowledge on this infection model are converging towards for more sophisticated studies revealing the complete potential of the greater wax moth. Larvae display intrinsic autofluorescence, which limits the use of fluorescent proteins to monitor the evolution of the infection over time, so this challenge was resolved by using bacterial gene bioluminescence and/or bacterial RNA expression. Using bioluminescence, the expression of the L. monocytogenes major virulence factors was measured, revealing a significant induction of temperature-dependent gene expression. The K. pneumoniae capsule polysaccharide (CPS) is an important virulence factor in the pathogenesis of pneumonia and urinary tract infections, the Galleria model was used to support these facts by mimicking a systemic infection. The CPS has been proved crucial also in invertebrates as larvae lethality was decreased with a Δcps strain and associated with a reduced bacterial survival after 24h. The monitoring of cps bioluminescence expression levels and genes encoded lipid A remodeling validates these findings, and revealed an expression peak during the first hours of infection. Bioluminescence combined to bacterial RNA expression highlighted the importance of ribonucleotide reductases (RNR) by monitoring the expression of genes encoded for these RNR toxins in Pseudomas aeruginosa infected larvae. We have recently used this model to investigate the role of S. aureus small regulatory RNAs (sRNAs), provide evidence that the sRNA expression profile in infected larvae differs sharply from in vitro. sRNAs were tightly regulated at the different stages of infection. Some sRNAs are linked to virulence but for the most part, their function has not been elucidated, so the G. mellonella host would be a reliable tool to screen their implication into virulence. Others S. aureus virulence factors were also in vivo monitored in a context of device associated infections. Few genes were upregulated in vivo as expected such as atl, icaA, and sarA unlike fib, fnbA and fnbB that are normally involved in biofilm biogenesis. This last comment could be related to the lack of fibrin and fibronectin in G. mellonella larva.

21Experimental studies of infected burn wounds in mammals are unwieldy due to ethical considerations. The G. mellonella infection model has been tested recently for this purpose. It was possible to simulate a burn wound by applying a preheated metal element to the larval surface. Spontaneous mortality was correlated to burn wound surface and fluid resuscitation e.g. injection of a saline solution. Then, infection by different pathogens, P. aeruginosa, S. aureus, and Acinetobacter baumannii, results in increased mortality with a bacterial dissemination. This model could therefore constitute a new alternative to analyze biofilm formation and to test compounds in this context.

22Recently, the study of bacterial biofilm infections was successfully adapted in vivo in the G. mellonella model especially with S. aureus. Two techniques were employed to simulate S. aureus infection on indwelling devices: toothbrush bristles insertion and stainless steel or titanium implants. First, abiotic surface insertions did not result in excess larval mortality, and secondly, by scanning electron microscopy, it was possible to observe S. aureus biofilm on indwelling devices and to determine the main steps of biofilm genesis such as attachment, proliferation and detachment. Interestingly, in all cases, larval mortality was higher in biofilm group than in non-biofilm group. The G. mellonella biofilm model was also used to evaluate efficacy of several antibiotics in the case of P. aeruginosa or K. pneumoniae infections. Together, these information support that the G. mellonella model constitute a suitable alternative to study bacterial biofilm and to test antibacterial compounds.

23For the first time, the G. mellonella model was successfully used to demonstrate the pathogenicity of the M. tuberculosis complex. Unlike the murine model that does not produce normally granulomas, a pathognomonic sign of tuberculosis infection, these aggregate structures were observed both in the hemolymph and the fat body of G. mellonella. Authors also provided that bacteria were phagocytosed by hemocytes, multiplied inside them, and then formed granuloma during the infection. These preliminary results pave the way to others studies such as testing the responsiveness of antituberculosis drugs. Indeed, the most common classes of anti-tuberculosis drugs have recently been screened, showing a convincing efficacy of most of the drugs tested. Rifampicin and isoniazid were the most efficient molecules both in terms of larval mortality and reduction of bioluminescence.

24Currently, antimicrobial resistance (AMR) is a major public health issue, and some predictions are not favorable, with a number of deaths related to AMR expected to be around 10 million per year in 2050 if the situation remains as it is. World Health Organization has listed the pathogens to be closely monitored grouped into the ESKAPE clad including Enterococcus faecium, S. aureus, K. pneumoniae, Acinetobacter baumannii, P. aeruginosa and Enterobacter spp.. Therefore, it is crucial to develop alternative solutions to conventional antibiotic treatments because few antibiotic compounds have been developed in recent years. Alternative therapies such as antibiotics in combination or with adjuvants, bacteriophages, AMPs, phytochemicals, are increasingly studied. The major advantage of this invertebrate model is to allow fast screening tests of new therapies, which will be validated subsequently in a mammalian model. The aim is to limit the untimely use of mouse models. In this respect, the G. mellonella model has been widely used to test the potential effectiveness of such alternative therapies ( Table 2 ).

25Association of linezolid plus fosfomycin was statistically more effective with an enhanced survival rate than linezolid alone against vancomycin-resistant E. faecium (VRE)-infected larvae. However, there were no difference between the combination treatment group and the use of fosfomycin alone. Oritavancin, a new glycopeptide vancomycin-derivative antibiotic, with an extended half-life span, has been successfully tested in the G. mellonella model after infection of both vancomycin-susceptible and vancomycin-resistant strains. Results with oritavancin were superior to other antibiotics alone or in combination (ceftriaxone, gentamicin and daptomycin), and no two-drug regimens including oritavancin has shown superiority over oritavancin alone.

26Antibiotic combinations were also tested on S. aureus infected larvae. The double-therapy linezolid plus fosfomycin even at low-dose was shown to act synergistically, reduce mortality and hemolymph bacterial burden. The fact that it can be used in low doses is of real benefit in limiting side effects (hematological cytotoxicity) and the appearance of resistant mutants.

27Extended-spectrum β-lactamase (ESBL)- or carbapenemase-producing Enterobacterales are one of the most important challenges in global health due to multidrug-resistance and worldwide spread. Ceftazidime/avibactam alone at high doses, or combined with other antibiotics (polymyxin B, amikacin or meropenem) reduced significantly larvae mortality after infection by KPC-producing K. pneumoniae strains resistant to ceftazidime/avibactam. Combination of ceftazidime/avibactam and meropenem seems to be the most promising based on toxicity evidence and low potential of resistance emergence.

28After establishing that the G. mellonella model was an accurate model to investigate E. cloacae virulence, Yang et al. shown that the association imipenem-colistin provided some interesting results in this invertebrate model, and could be a potential alternative when bacteria were pan-resistant.

29For multidrug-resistant (MDR) opportunistic pathogens such as A. baumannii or P. aeruginosa, the G. mellonella model was successfully used as a screening model for alternative therapies. It was demonstrated that a urea-derived compound used as an adjuvant potentiated colistin action both in vitro and in vivo in a G. mellonella model infected by a MDR A. baumannii strain. Indeed, combination of colistin plus adjuvant compared to colistin alone or other antibiotics enhanced larvae survival. These preliminary results imply a lower colistin dosage and consequently associated renal and neurological toxicities, and suggest a colistin activity restauration on resistant strains by the adjuvant which would block resistance mechanisms.

30The new multifunctional adjuvant, the non-ribosomal tobramycin-cyclam conjugate, could change therapeutic deadlock situations since it significantly restored both meropenem or aztreonam or ceftazidim/avibactam activity, and resulted in prolonged larvae survival infected with MDR P. aeruginosa strain.

31As previously explained, AMP role is crucial in larvae immune humoral response. Eisenhardt et al. highlighted that IMPI from G. mellonella inhibited the in vitro elastase and the secretome activity of P. aeruginosa strains, and confirmed these results in an in vivo wound model. Insect AMP such as IMPI could be another strategy to treat bacterial infections. In that sense, cecropin A2 from the insect Aedes aegypti was shown to attenuate P. aeruginosa infected larvae mortality when in combination with tetracycline.

32Using a cocktail of specific bacteriophages against VRE-infected larvae El Haddad et al. demonstrated an improving survival rate of infected larvae compared to the phage-free group.

33Two types of bacteriophages, Staphylococcal bacteriophage and PYO bacteriophage respectively could both treat and prevention Methicillin resistant S. aureus (MRSA) infection. Indeed, at therapeutic doses, phages presented no larval toxicity, reduced larvae mortality rates when administrated before and after infection. The effect was dose-dependent and increased survival rates were observed with phage cocktails.

34Thiry et al. determined that new bacteriophages at low doses were effective against KPC-producing K. pneumonia. However, authors conclude that although the G. mellonella host-model is a reliable tool for fast results, it would be premature to predict the effect on mammals requiring higher doses of bacteriophages. Others found the same results by examining phage effects on several carbapenem-resistant Enterobacterales: phage cocktails reduced larval mortality by 90% with a single dose, a figure that reached 100% with three successive doses.

35Similar results were also obtained on MDR or extensively drug-resistant P. aeruginosa strains with the use of bacteriophages or adjuvants respectively. Interestingly, the two novel bacteriophage effects were validated first using the G. mellonella host and then in a mouse acute pneumonia model thereby exhibiting the robustness of this invertebrate model.

36Phytochemical therapies have emerged in recent years as a potential alternative to conventional antibiotics. These anti-virulence therapies would reduce bacterial adaptation by acting on multiple targets. For example, Eugeria brejoensis essential oil inhibits Staphyloxanthin production in S. aureus and acts synergistically with conventional antibiotics and Myricetin downregulates multiple S. aureus virulence factors involved in adhesion, biofilm formation or hemolysis. Numerous plant-derived products have been evaluated in the G. mellonella model, in particular to target S. aureus, whether they are phenolic or flavonoid compounds (myricetin, epigallocatechin gallate) or molecules derived from plant oils (cinnamon oil, Eugeria brejoensis essential oil). All showed reduced larvae mortality with no toxicological effect when infected by S. aureus, so they might be promising substances to treat S. aureus infection including Methicillin-resistant strains.

37Combination of two polyphenols, theaflavin and epicatechin was effective on MDR A. baumannii infected larvae as recently shown. These two bioactive compounds act synergistically and have a bactericidal effect resulting in increased larvae survival when compared to their use alone.

38Traditionally, two other invertebrate models are widely described to study the virulence of bacteria or to test the effect of antimicrobial compounds: C. elegans and D. melanogaster. The G. mellonella model has several advantages over these two ( Table 3 ). It is established that the temperature influences the behavior of bacteria because the expression of many genes is under the control of temperature. For example, the L. monocytogenes prfA regulon regulates many virulence factors thus allowing bacteria adaptation to this host. prfA expression is a function of temperature: at 20°C, it is not activated and becomes so at 37°C, i.e. the temperature of the human body. By comparing the expression of the prfA regulon and associated virulence determinants at both 37°C and 30°C in the Galleria model, it was shown that these factors were more weakly expressed at 30°C than at 37°C. C. elegans and D. melanogaster models cannot withstand temperature of 37°C whereas G. mellonella does, a considerable benefit for investigating major human pathogens. Furthermore, it was demonstrated that incubation temperature induced different G. mellonella larval mortality profiles by studying virulence group A streptococci and Shigella spp.. These results indicated that temperature requirements are an unavoidable prerequisite to consider host-pathogen interactions rigorously. Another important consideration is the immune cell inability of C. elegans to phagocytize microorganisms unlike G. mellonella that exhibits professional phagocytes conferring it a major advantage for the study of pyogenic bacteria such as S. aureus, P. aeruginosa or S. pneumoniae. In the C. elegans model, bacteria proliferate in the intestinal lumen but do not penetrate intestinal epithelial cells, making it difficult to study the virulence of intracellular bacteria such as L. pneumophila.

39Furthermore, a technical aspect facilitate the use of the G. mellonella model. Its length varies between 3-30 mm, thus significantly higher than those D. melanogaster and C. elegans, (3 mm and 1 mm, respectively), which facilitates the operation and handling of the larvae. Conversely, improvements are needed so that G. mellonella larvae will be widely accepted as a leading infection model. In particular, the lack of stock centers and databases such as Flybase or WormBase as well as standardized procedures or mutant strains are significant hindrance that can be explained because this model is still in its infancy.

40No animal model is perfect. Ideally, to study host-pathogen relationships, one would need to work directly in humans, which is not feasible for both ethical and technical reasons. It is therefore necessary to accept a compromise by choosing the most suitable animal model on the basis of ethical, economic and scientific arguments. In this sense, the G. mellonella model has proven its usefulness as a screening model for the study of the virulence of many microorganisms, with results confirmed in mammalian models. However, this robust model has shortcomings that do not allow an optimal use making complex comparisons between scientists.

41To date, the main limitation to the use of this model is the lack of standardized procedures, especially in the obtaining and storage conditions of the larvae. For example, previous works using G. mellonella as a host model for study S. aureus virulence or testing antibacterial agents showed many differences. Three labs purchase larvae in fishing bait stores, one used internal rearing protocol with artificial diet whereas for two another studies there is no data available about origin of larvae. These different processes are sources of variability since it was shown that larvae from fishing bait stores or pet shops contain antibiotics or hormones that lead to variable and inconsistent results. It was demonstrated that different artificial diets influenced the survival of larvae after infection caused by S. aureus, E. coli or C. albicans. Diet influences immune system activation, responsible for fluctuations in hemolymph volume and hemocyte concentration. Commonly, larvae are not fed during infection experiments, and it is advised to starve larvae before infection. However, since diet and immune activation are linked, it would be interesting to monitor larvae mortality with and without added food during infection process in order to observe any potential effect.

42It would be useful to create stock centers for research on G. mellonella larvae as it exists for the D. melanogaster. Recently, Biosystems technology Ltd developed standardized G. mellonella larvae called TruLarv™ that are age and weight defined and whose breeding conditions are normalized without hormones and antimicrobials (https://biosystemstechnology.com/products/trularv). Experiments with those standardized larvae delivered more suitable and reproducible results with several pathogens. In addition, the use of calibrated reference populations would facilitate comparison of works from different labs.

43Larvae storage conditions differ from one study to another with an impact on results. Larvae are often stored in the dark but at various temperatures and for longer or shorter periods after receipt. The pre-incubation stage is crucial because when the duration is extended, the larvae were more susceptible to pathogens, with an altered immune response.

44Another critical aspect is the larvae weight used in the experiments. In six previous publications, the weight ranged from 150 to 700 mg. In 2019, a method was proposed to standardize the intra larval concentrations of bacteria after injection. Authors showed a strong correlation between larvae weight and liquid volume allowing then to estimate with precision the in vivo concentration of bacteria and drugs administered.

45G. mellonella infections can be proceeding either by ingestion or by intra-haemocolic injection in the last proleg, this last method allowing a tight control of the inoculum using a microinjector with a calibrated syringe ( Figure 4 ). Methods for injection are not the same between laboratories, and at least, three methods are described. The differences are not related to the number of bacteria injected but rather the time it takes to infect one larva and the risk of injury for the operator. A recent study showed that with simple restraint devices, there was a reduction of larvae manual handling with a rapid rate of injection, combined with increased protection against needle stick injury.

46Considerable progresses have been made to improve interpretation of tests with this model. First studies calculated larvae mortality at the endpoint to objective the LD50. Larvae were considered dead if they were immobile, no longer responding to stimuli and were melanized. This approach can be considered as subjective because melanization process may differ from one larva to another. A variant is to calculate a larval virulence index allowing an inter-laboratory easier comparison. A health index scoring system dedicated to the G. mellonella model was also introduced to monitor severity and mortality. It includes 4 grades (activity, cocoon formation, melanization, and survival) and is rated from 0 to 10: the higher the score, the healthier the larvae ( Table 4 ). There is still need for improvement since the weighting of the health index scoring system is sometimes imperfect. For example, larvae transformed in the pupal stage are excluded from the monitoring because it is impossible to attest melanization once the chrysalis is formed, just as when a complete cocoon surrounds the larva.

47Over the past decade, G. mellonella entered the omics era, and both trancriptomic and proteomic data provided a more suitable comprehension of this model host. In 2011, the first transcriptional analysis of immune system genes was reported and numerous genes encoding immune proteins as well as effector soluble molecules were identified. From this transcriptomic project, several genes were associated with immune functions after infection with L. monocytogenes. Moreover, miRNAs were also discovered and associated with the immune response after contact with human pathogens such as E. coli and L. monocytogenes. However, numerous immune proteins were still not characterized and have unknown functions. Majority of studies dedicated to observe humoral response after bacterial exposure focused on already known AMPs such as lysozyme, galliomycin, gallerimycin, cecropin. Recently, a quantitative shotgun proteomics was successfully realized in response to a S. aureus infection demonstrating a rapid and coordinated humoral immune response. Still using a proteomic approach, these authors have mapped the proteome of both infected larvae and a fungal pathogen. By this dual approach, they were able to describe the genesis of the infection, from host-pathogen recognition to the process of melanization and encapsulation. In 2020, the Antimicrobial Peptide Database reported about 310 AMPS in insects thus suggesting that the 20 or so peptides identified in G. mellonella constitute only a tiny part of the humoral repertory, and that further studies are required to get more information about this humoral response.

48At last, G. mellonella genome is now sequenced, but it is still not entirely analyzed and annotated (accession number: NTHM00000000). The genome sequenced was obtained from the isolate FT-Tue with a total of 2,141,900 reads and 20,638,932,410 bases. Moreover, a G. mellonella transcriptome database is now available (https://www.uni-giessen.de/fbz/fb08/Inst/bioinformatik/Research/Supplements/galleria). These preliminary data might allow the implementation of a shared-file database from genomic, transcriptomic and proteomic projects.

49The G. mellonella infection model is now well demonstrated as a screening model, enabling to study bacterial virulence by monitoring larval survival, bacterial load, immune response of infected larvae or histological data. So, the remaining question to be asked is can we go further in the study of virulence and host-pathogen interactions? High-throughput technologies (genomics, transcriptomics, proteomics or metabolomics), have revolutionized biomedical research. To properly establish, survive and grow in a host, bacterial pathogens must counter hostile conditions related to host immune response and environmental conditions, and also compete with other bacteria for nutrient deprivation. This adaptation leads to an alteration in patterns of gene expression. In the field of infectious diseases, omics methods such as transcriptomics not only allow to notify the presence of a gene but also to know if it is expressed through a quantification of host and/or bacterial cellular mRNA transcripts. These high-throughput technologies are crucial from both diagnostic and therapeutic points of view, and have been extensively applied to human pathogenic bacteria both in vitro and in vivo, including humans or mammalian models.

50To date, few studies are available about bacterial pathogen gene expression within the G. mellonella host whether by bioluminescence or RT-qPCR. The main reason is probably related to the lack of standardized procedures as previously explained. Nevertheless, the implementation of stock centers with referenced, standardized and sequenced larvae could solve this problematic, and therefore could herald the development of bacterial transcriptomic studies in this host-model.

51A key step in transcriptomic studies is the quality and quantity of bacterial RNA. In most organisms, the vast majority of RNA corresponds to ribosomal RNA (rRNA) in bacteria as well as in mammals or in vertebrates, and must be depleted to enable a suitable analyze of transcriptomic data which corresponds predominantly of mRNA. The major issue is that several commercial kits are available to deplete rRNA from mammalian samples but none are validated to invertebrate models such as G. mellonella. Nonetheless, the in vivo transcriptome of two microorganisms (Coxiella Burnetii and Yersinia entomophaga) within G. mellonella system have been realized. In these two studies, the authors succeeded to extract intact bacterial RNA in sufficient amount from the infected pool hemolymph. In vivo analysis of the transcriptome of C. burnetti as well as Y. entomophaga revealed that a significant number of genes were expressed differently, either increased or decreased compared to in vitro media. Pattern expression profile genes were then compared to mammalian cell lines or a mouse model, highlighting many similarities. These two pioneering studies demonstrate the G. mellonella larvae model relevance, no longer as a simple screening model, but also as a genuine infection model allowing to explore the virulence regulation at a transcriptomic level, and thus to better define the complex systems that are bacterial regulation networks. This novel approach deserves to be expanded for the study of major pathogenic bacteria, not only in the hemolymph but also in other larval anatomical sites, with the prerequisite of selective cell lysis to obtain bacteria-enrich samples. By combining high throughput technologies with the advantages of the G. mellonella model would make it possible to get more easily comprehensive bacterial transcriptomic data. Finally, to concretize these new developments, conducting dual-RNAseq would provide the opportunity to simultaneously obtain details about complex interactions host and bacteria.

52GM and P-YD contributed to conception, design and writing the first drive of the manuscript. All authors contributed to manuscript revision, read, and approved the submitted version.

53The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

54All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.