Galleria mellonella larvae exhibit a weight-dependent lethal median dose when infected with methicillin-resistant Staphylococcus aureus

11 Galleria mellonella is a recognised model to study antimicrobial efficacy; however, 12 standardisation across the scientific field and investigations of methodological 13 components are needed. Here we investigate the impact of weight on mortality 14 following infection with Methicillin-resistant Staphylococcus aureus (MRSA). Larvae 15 were separated into six weight groups (180-300 mg at 20 mg intervals) and infected 16 with a range of doses of MRSA to determine the 50% lethal dose (LD 50 ) , and the 17 ‘lipid weight’ of larvae post-infection was quantified. A model of LD 50 values 18 correlated with weight was developed. The LD 50 values, as estimated by our model, 19 were further tested in vivo to prove our model. 20 We establish a weight-dependent LD 50 in larvae against MRSA and demonstrate that 21 G. mellonella is a stable model within 180-260 mg. We present multiple linear 22 models correlating weight with: LD 50 , lipid weight, and larval length. We demonstrate 23 that the lipid weight is reduced as a result of MRSA infection, identifying a potentially 24 new measure in which to understand the immune response. Finally, we demonstrate


Introduction
Galleria mellonella (Greater wax moth) larvae are widely utilised for toxicity screening (Desbois and Coote 2012;Maguire, Duggan and Kavanagh 2016;Coates et al. 2019) and to study host-pathogen interactions (Peleg et al. 2009;Olsen et al. 2011;Junqueira 2012;Wojda and Taszłow 2013).Unlike many insect models, G. mellonella can be incubated at 37°C, which facilitates the investigation of human pathogens.This has included most of the ESKAPE pathogens: Enterococcus faecium (Chibebe Junior et al. 2013;Luther et al. 2014); Staphylococcus aureus (Brackman et al. 2011;Ramarao, Nielsen-Leroux and Lereclus 2012;Sheehan, Dixon and Kavanagh 2019); Klebsiella pneumoniae (Wand et al. 2013;Diago-Navarro et al. 2014); Acinetobacter baumannii (Peleg et al. 2009); and Pseudomonas aeruginosa (Jander, Rahme and Ausubel 2000;Seed and Dennis 2008).Additionally, Escherichia coli (Leuko and Raivio 2012;Alghoribi et al. 2014;Jønsson et al. 2017;Guerrieri et al. 2019), Bulkholderia mallei (Schell, Lipscomb and DeShazer 2008) and several fungi (Cotter, Doyle and Kavanagh 2000;Reeves et al. 2004;Mylonakis et al. 2005) have also been studied using G. mellonella.Crucially, a positive correlation between the virulence and immune responses between mammalian models and G. mellonella has been established for P. aeruginosa (Jander, Rahme and Ausubel 2000), Cryptococcus neoformans (Mylonakis et al. 2005), and S. aureus (Sheehan, Dixon and Kavanagh 2019), demonstrating the powerful potential of this invertebrate model.Antibiotic efficacy at dosages recommended for human use can be tested in G. mellonella, in addition to their toxicity correlating with toxicity observed in murine models (Ignasiak and Maxwell 2017).This has been shown with both natural and synthetic compounds (Gibreel and Upton 2013;Smitten et al. 2019), opening up the possibility of a rapid and cheap model for the early stages of discovery and development of natural and synthetic products, without the challenges of ethical approval, specialist training and the difficulties of using mice-models in early-stage drug development.Infections caused by antibiotic-resistant S. aureus are of global concern and it is listed as a high priority pathogen for which new antibiotics are urgently needed (The World Health Organisation 2017).Methicillin-resistant S. aureus (MRSA) has been utilised with G. mellonella for the study of virulence (Mannala et al. 2018), pathogenicity (Ebner et al. 2016), antimicrobial efficacy of existing antimicrobials (Ba et al. 2015;Ferro et al. 2016), and for novel candidates (Gibreel and Upton 2013;Jacobs et al. 2013;Dong et al. 2017) (Table S1).
Despite the increased popularity of G. mellonella, there is much variability in method application (Andrea, Krogfelt and Jenssen 2019).This includes differences in larval size, storage, infective dose, and injection intervals.In this study, we address larval size and its potential impact in experimental design.In antibiotic efficacy studies, typically the model is infected with a pathogen shortly before the candidate treatment is presented.This has not been standardised with respect to the parameters previously mentioned for G. mellonella.In our preliminary experimentation in determining a 50% lethal dose (LD50) for MRSA in G. mellonella, it was noted that smaller larvae were more susceptible to infection than larger larvae.This was when using a broad range of larval weights (~200-300 mg), as previously reported (Jacobs et al. 2013).Furthermore, the larval weight has been demonstrated to positively correlate with the larval liquid volume, leading to recommendations on how in vivo concentrations of injected compounds and pathogens should be calculated (Andrea, Krogfelt and Jenssen 2019).This led us to hypothesise that the larvae LD50 for a pathogen, in our case here MRSA, is directly proportional to the larvae weight and that larvae weight is an essential parameter in experimental design that must be tightly controlled.
When physical and anatomical barriers are breached, the wax moth larvae have an innate immune response relying on germline-encoded factors for the detection and clearance of microbial pathogens (Trevijano-Contador and Zaragoza 2019).There are two branches, cellular and humoral immunity.Cellular immunity is conducted by haemocytes, which are present in an open circulatory system called the haemolymph, which is analogous to vertebrate blood.There are at least six subpopulations of haemocytes which perform similar roles to those of the myeloid lineage in vertebrates (Boman and Hultmark 1987;Lavine and Strand 2002), and they are also associated with digestive system, trachea and fat body (Ratcliffe 1985).
Five types of haemocytes were identified in fifth larval instar of G. mellonella; prohaemocytes, plasmatocytes, granulocytes, oenocytoids and spherulocytes (Salem et al. 2014).The main immune processes include coagulation, phagocytosis and encapsulation (Tojo et al. 2000).Circulating haemocyte density increases during pathogenesis due to the release of suspended cells from the fat body (Tojo et al. 2000).Haemocyte density and subpopulation variations changes with time of exposure to pathogen and pathogen virulence (Arteaga Blanco et al. 2017).
Melanisation additionally occurs in the haemolymph, the process of melanin production resulting in the darkened appearance of the larvae (Tojo et al. 2000).The humoral branch is involved in the production of lytic enzymes (Vogel et al. 2011), and antimicrobial peptides (AMPs) that are active against bacterial pathogens (Cytryńska et al. 2007;Tsai, Loh and Proft 2016).These molecules are mostly produced by the larval 'fat body', analogous to the mammalian liver, and are released into the haemolymph (Zasloff 2002).
A proteomic investigation has shown S. aureus infections lead to an increase in production of proteins such as AMPs and peptidoglycan recognition proteins (Sheehan, Dixon and Kavanagh 2019).Critically, the same study identified similarities between G. mellonella and mammal immune response to S. aureus infections.What has not been investigated is the physiological change in G. mellonella lipid as a result of S. aureus infections.For this investigation, we were motivated to quantify the lipid weight, a proxy for the fat body, of the larvae to observe how the fat body might have been affected as a result of MRSA infection.
The aim of the work here is to investigate methodological adjustments which may improve the reproducibility and reliability of using pet-food grade G. mellonella as an experimental model.This was achieved by (i) examining the effect of larval weight on the LD50 to MRSA infection, and (ii) characterising physiological changes occurring to lipid weight as a result of the larval immune response to MRSA.

Cultivation of MRSA
A single colony of Methicillin-resistant Staphylococcus aureus (MRSA) NTCT 12493 was streaked onto fresh Luria broth (LB, Fischer Scientific, UK; Tryptone 10 g/L, yeast extract 5 g/L, and sodium chloride 10 g/L) solidified with 1.5% agar (Acros Organics, UK) 24 h before experimentation.Single colonies were suspended in Dulbecco A Phosphate buffered saline (PBS, Oxoid UK) to a range of optical density (OD) read at 600 nm (Eppendorf BioPhotometer, Netherlands).These dilutions were OD600 = 0.1 -1.0 in 0.1 increments.Viable cell counts were made of each dilution.
Determining a weight-based LD50 for Galleria mellonella larvae Larvae were purchased commercially from Livefoods UK Ltd. (Somerset, UK; www.livefoods.co.uk).On receipt, larvae were individually weighed using an accurate scale and grouped into the following weight bands: 180-200, 201-220, 221-240, 241-260, 261-280, and 281-300 mg.Larvae were stored at 4°C for up to 7 days in the dark with no food and water.Healthy larvae were identified by a uniform cream colour, with no indications of melanisation such as spots or markings (Fig. 1A) (Li et al. 2018).Larvae were euthanised by chilling them at 4°C for 1 h, before freezing them at -20°C for a minimum of 24 h.Larvae (n = 10) from each weight band were infected by injecting 10 µl of one of the 10 dilutions of MRSA into the left penultimate pro-leg (Fig. 1C), using a 50 µl Hamilton 750 syringe (Hamilton Company, UK) with a removable needle.Injected larvae were placed into Petri dishes lined with tissue paper (KIMTECH, UK).Three independent replicates of this experiment were carried out.Syringes were cleaned before and after each bacterial dilution.Cleaning consisted of taking up and discarding of each wash solution thrice before progressing to the next wash solution.
After infection, the larvae were maintained at 37°C in the dark without food or water.
A placebo control of sterile PBS was used to account for the effect of the physical trauma of injection, along with a non-manipulation (NM) control.After 24 h the live/dead counts were recorded.Larvae were recorded as dead when they met the following: (i) complete melanisation (Fig. 1A), (ii) did not respond to touch, and (iii) could not correct itself when rolled onto its back.
Determining the weight-dependent LD50, live/dead counts were converted into percentage mortality at 24 h for each group.For this investigation we have defined LD50 as CFU of MRSA per mg of organism resulting in 50% mortality.To model the dose-response and describe the relationship between increasing the infection dose on survival for each weight group, a non-linear sigmoidal regression curve was plotted.The infection dose, represented as CFU/ mg of total weight of larva, was logtransformed.A non-linear regression curve was calculated to fit best the data generated from three independent replicas.From the equation generated from this curve, the theoretical LD50 was calculated along with the standard deviation (SD).Estimated LD50 from each weight groups were plotted against the mean larvae weight.A regression line was drawn, and the coefficient of determinant R 2 was calculated.

Correlating larval size with rate of pupation
On the day of receipt, larvae were placed into weight groups in Petri dishes.They were immediately placed at 37°C, in the dark with no food or water and permitted to pupate over 15 days.Larvae were observed daily and pupation events recorded.

Quantifying lipid weight of G. mellonella
Following investigation of the LD50 for MRSA, the lipid weight for all living and dead larvae was quantified.Live larvae from treatments, the NM and PBS controls were ethically euthanised.Dead larvae were stored at -20°C until needed.Larvae were left to thaw at room temperature for 24 h and were weighed and individually placed in Eppendorf tubes to be dried over 7 days at 55°C, and re-weighed to reveal their dry weight.Larvae were then submerged in ≥99.9% diethyl ether (Sigma-Aldrich, UK) and left for 3 days at 4°C to dissolve lipid.Diethyl ether was utilised as the lipid extraction solvent (Tzompa-Sosa et al. 2014).After, ether was left to evaporate in a fume hood for 24 h.Once dried, larvae were weighed again to acquire the post-ether weight.Quantities are then presented as followed: 'total weight' is the weight of the larvae pre-experimentation; 'water weight' ( ℎ =   ℎ −  ℎ); 'lipid weight' ( ℎ =  ℎ −  ℎ ℎ).

Larval weight affects LD50
To begin testing our hypothesis, LD50 values were determined for each weight group.
A sigmoidal non-linear model best fit the dose-dependent response of the data, resulting in an LD50 calculated for each weight group (Fig. 2).When adjusted to the number of cells injected into each larvae per one unit of body weight (CFU/mg), the resulting LD50 ranged from 1.19 x10 7 CFU/mg, for the 180-200 mg group, to the highest LD50 which was 8.97 x10 7 CFU/mg for the 261-280 mg group (Table 1).The LD50 increased across weight groups except for the 281-300 mg group, which had a lower LD50 than the 261-280 mg weight group.Throughout this experiment, we encountered some difficulties when handling larvae from the two higher weightbands (261-280 and 281-300 mg), such as high variation in mortality at the lowest infective dosages (0-40% mortality) and highest dosages (60-100% mortality).
Nevertheless, we were able to calculate an LD50 with the final data.Example of how this is calculated can be found in Table S3 We observed a positive correlation between weight of the larvae and LD50, as calculated by Pearson correlation test (r = 0.87, p = 0.025, n = 18).A linear regression model arriving at an equation ( = 0.007966 + 5.548) was used to estimate LD50 (Fig. 3A).The LD50 values, as estimated by our model, were tested in vivo, demonstrating an approximate 50-56% (± 5.7 -10%) survival for four of the weight groups (Fig. 3B).Survival at 24 h for the weight groups 261-280 and 281-300 mg was 30% (± 0%) and 43% (±15.3%),respectively.

MRSA infection leads to a reduction in lipid weight
With the non-manipulated (NM) group, we assessed the overall relationship between total weight, dry weight, and lipid weight and length of the larvae (Fig. 4).Determined by Pearson's correlation test, we found a positive correlation between the total and dry weight, (r = 0.972, p < 0.0001, n = 83) (Fig. 4A), and total and water weight (r = 0.989, p < 0.0001, n = 83) (Fig. 4B).These two results support the findings of previous research (Andrea, Krogfelt and Jenssen 2019).Two additional positive correlations were observed between total weight and lipid (r = 0.788, p < 0.0001, n = We also investigated the effect of infection on the lipid weight of all the larvae used in determining the LD50 for MRSA (Fig. 5).As calculated by one-way ANOVA, injection with MRSA resulted in an overall decrease in the lipid weight for both dead (18.7 mg  S2).Finally, we observed that at a high infective dosage of MRSA, the larvae had a lipid weight close to the mean of the NM and PBS control compared to the lower dosages (Fig. 5C-D).This was supported by a positive correlation between lipid weight and infective dose for both live (r = 0.778, p = 0.008) (Fig. 5C) and dead larvae (r = 0.669, p = 0.035) (Fig. 5D).

Pupation is unaffected by weight
To explore whether larger larvae were closer to the final instar stage (pupae) in which they begin to pupate into adult moths, an observational experiment was performed.NM larvae were left to pupate at 37°C, and it was observed that 80-100% of larvae pupated within the 15 day incubation period, independent on their weight grouping, as calculated by Log-rank (Mantel-Cox) test (X 2 (5, N = 60) = 4.004, p = 0.549) (Fig. 6).

MRSA exhibits a weight-dependent LD50
In this study, we have demonstrated it is possible to develop a model in which a LD50 can be predicted based on the weight of the larvae, and that the prediction can be experimentally validated (Fig. 3).The linear model correlating total and water weight (Fig. 4B) imply that in increasingly larger larvae, the in vivo dilution of MRSA increases requiring a greater density of pathogen to reach the LD50.Likewise for the positive correlation confirmed with total and lipid weight (Fig. 4C), the presence of a larger fat body that can be degraded for the production of immune factors, may well be why we observe the weight-dependent effect on LD50.The LD50s (1.19 -8.97 x10 7 CFU/mg) for the MRSA strain was not within range of infective dosages utilised in previously investigated MRSA and Methicillin-sensitive S. aureus (MSSA) strains (0.8 -5.0 x10 6 CFU) (Table S1).However, a direct comparison may not be appropriate given the variation in reporting densities as in our study the LD50 was adjusted to account for in vivo dilution in the larvae as described in Andrea, Krogfelt and Jenssen ( 2019), but this is not always done.
During the process of this investigation, we found two of the largest weight groups (261-280 and 281-300 mg) to be unreliable, which hindered progress.This was consistent across multiple batches of larvae orders.LD50, as calculated by our model for 261-280 and 281-300 mg larvae, resulted in less than 50% survival at 24 h (Fig. 3B), indicating that our model for a weight-dependent LD50 had overestimated the LD50.Our first assumptions were that larger larvae were older and closer to pupation than the smaller weight groups, as larvae increase in size until pupation (Jorjão et al. 2018), which might somehow impact on survival.Given the difficulty in identifying an age for each larva, it is a difficult hypothesis to test beyond quantifying the number of days it took for NM larvae from each weight group to pupate.
When this was conducted, we found that larval size did not influence the probability of pupation (Fig. 6), and we conclude that the larvae received from the supplier had an 80-100% probability of pupating within 15 days if kept at 37°C, regardless of weight.It would appear that larger larvae were not likely to be closer to pupation than smaller ones, so the reason for our observed decrease in LD50 for large larvae remains unknown.Since larvae were kept without food, this may be a reason for the observed similar pupation times across all weight groups as lack of food source may be forcing the larvae into pupation.Feeding regimes are not the standard protocol when investing antibiotic efficacy, as such we feel this best represented the conditions larvae would be exposed to at the start of experimentation.
Using G. mellonella does have drawbacks, one such being the functional equivalent of adaptive immunity termed 'immune priming' (Little and Kraaijeveld 2004;Sadd and Schmid-Hempel 2006).Individual larvae that survive infection or exposure to a particular pathogen may exhibit increased immune resistance against the same or similar pathogens.Priming with heat-killed pathogens was observed to result in increased larval survival (Wu et al. 2014).Ultimately there will be no control over the immune history of the larvae and this should always be recognised when working with pet-food grade G. mellonella.Across the literature, a wide range of weight bands have been utilised: 150-200 mg (Mannala et al. 2018); 300-700 mg (Ebner et al. 2016); 200-300 mg (Jacobs et al. 2013); and in other studies this is not declared (Ba et al. 2015;Jorjão et al. 2018).Our results suggest that choosing weight ranges as wide as 300-700 mg and 200-300 mg could result in inconsistent data.While a weight range of only 20 mg is likely a conservative approach, ranges such as 100 mg or greater in our weight-dependent LD50 model for MRSA indicates that there would be significant differences in survival (Fig. 3A).
Weighing individual larvae is a time-consuming procedure.This study also demonstrated that larvae length is reasonable proxy for the weight (Fig. 4D).Larval length has been previously used to characterise larvae for experimentation where larvae of 15-25 mm were utilised (Bazaid et al. 2018).Like total weight, a large length grouping may also encounter similar challenges.A 20 mg weight grouping would equate to roughly 1 mm, for example, 180-200 mg would be 20-21 mm.
Measuring length may be a preferred alternative to accurately weighing all larval.
When sourcing larvae from our supplier, we frequently found that larvae belonging to the weight groups 201-220 and 221-240 mg were most abundant, which will inevitably be the practical determining factor in weight group selection.Our findings would support selection of larvae in this range.

Lipid metabolism occurs in response to MRSA infection
MRSA infection leads to a decreased lipid weight in the larvae after 24 h, whether they died or survived the infection (Fig. 5A).The reduction in lipid weight is likely the result of lipolysis during an immune response.This is to be expected, the fat body of the larvae produce many defence compounds essential to the larvae's immune response (Cytryńska et al. 2007;Tsai, Loh and Proft 2016).This reaction can be rapid, in some models showing production of AMPs within the first 4 to 6 h postinfection (Sheehan, Dixon and Kavanagh 2019;Trevijano-Contador and Zaragoza 2019).This is supported by proteomic work, which demonstrated that at 6 and 24 h post-S.aureus infection larvae had increased expression of AMPs (Sheehan, Dixon and Kavanagh 2019).
On exposure to the infecting pathogen, there may be a rapid metabolism of the fat body to provide the required energy to fight the infection.Larvae with larger lipid weight before infection might be more likely to survive, as seen with the surviving larvae having a greater lipid weight than dead larvae (Fig. 5A).Within this experimental design, the larvae are not fed before or during the experiment, and therefore they cannot be acquiring more lipid.Where lipid weight was seen as closer to the NM and PBS control baseline, as observed in the trend of lipid weight positively correlating with infective dose (Fig. 5C-D), it is more likely that lipid metabolism has been compromised.
What could reasonably be expected is that lipolysis of the fat body occurs to increase the production of AMPs and additional defence compounds.When Drosophila are stimulated by a systemic infection with S. aureus, signalling from the Toll receptor increases, which leads to increased production of AMPs and reduced accumulation of lipids (Liu et al. 2016;Lee and Lee 2018).This could suggest that for larvae surviving high infective dosages, there are additional immune responses that do not deplete the fat body.
We intended to quantify the larval lipid weight to aid in understanding the weightdependent LD50 effect and the observed unreliability of the two largest weight groups (261-280 and 281-300 mg).We report several observations regarding the lipid weight and MRSA infection; however, none can fully explain the irregularity we encountered for the largest weight groups.Analysing larval lipid weight has proved some insight, but would benefit from further investigation, though alternative methods to estimate lipid mass would be required.

Overall assessment of G. mellonella as a model
There remains a lack of widely available and cheap standardised stocks of larvae reared under controlled conditions.Temperature (Mowlds and Kavanagh 2008), diet (Banville, Browne and Kavanagh 2012;Jorjão et al. 2018), past infections (Fallon, Kelly and Kavanagh 2012), and antibiotics and hormones in the feed (Büyükgüzel and Kalender 2008) are all reported to influence laboratory experimentation.Most larvae currently used are acquired from commercial insect food providers (Andrea, Krogfelt and Jenssen 2019), where it is understood that use of antibiotics and hormones in the culture medium is common practice, and acquiring accurate information regarding the conditions in which the larvae are reared is challenging.All of which may vary between larvae suppliers, which is a challenge that warrants further investigation.
Ultimately from our investigation, it would appear that lipid deposits are essential in However for pet-food grade larvae to be reliably used in research, more significant consideration should be taken over the parameters that can be controlled, and in this study, we emphasise that such experiments can be reproducible and reliable.We recommend that investigators consider the potential variability associated with using different larval weight as we have shown herein.We would recommend using weight groupings as a means to control this. .Our data suggests that all larvae used should be within 10 mg of the mean weight of all larvae to provide consistency.Additionally, larvae of >260 mg should not be avoided.
In this work, we present several linear regression curves that could be used as tools to aid in experimental design, such as the linear model for LD50 (Fig. 3A), weight and lipid content (Fig. 4C), and length (Fig. 4D).Finally, we demonstrate that the lipid weight is reduced as a result of MRSA infection, identifying a potentially new measure in which to understand the immune response.Similarities between G.
mellonella and mammals in response to S. aureus infections can be used to study the efficacy and interactions of novel antimicrobials, even at early development stages.By refining and standardising methodologies in which to handle and select G. mellonella for study, we can improve the reliability of this powerful model for multiple purposes.This is the total density of MRSA 12493 that must be injected into larva of the selected weight-group in order to kill 50% of the population

Figure 1 :
Figure 1: Galleria mellonella larvae.(A) Melanisation is a visual indication of the health of the larvae, as larvae progress from none to complete melanisation as a result of stress and/or infection.(B) Larvae pupation.(C) Route of infection for larvae is by intrahaemocoelic injection at the penultimate pro-leg (arrow).Larvae diagram was adapted from Singkum et al. (2019).(D) Larvae are divided up into six weight groups.

Figure 2 .
Figure 2. Sigmoidal non-linear logistic regressions best fit the dose-dependent response observed when calculating an LD50 for MRSA.LD50 was calculated for each weight group with 10 larvae/group.Data are shown as mean ± SD (n = 10) of three independent replicas.

Figure 3 .
Figure 3. LD50 as calculated by non-linear regression models positively correlated with weight and was validated in vivo for all but the two highest weight groups.(A) Calculated LD50 by non-linear models correlation positively with total weight.(B) LD50 value as calculated by the model was validated by injecting into larvae and observing mortality.Data are shown as mean ± SD (n=10) of three independent replicas.

Figure 4 .
Figure 4. Multiple correlations observed between larvae total weight and dry weight, water weight, lipid weight, and larvae length.Non-manipulated (NM) larvae were used to analyse the relationships between (A) total weight and dry weight, (B) total weight and the lipid weight after here presented as lipid weight, (C) total weight and lipid weight as proportional to the total weight, water weight, and (D) total weight and larvae length where data is presented as mean ± SD (n = 252).

Figure 5 .
Figure 5. Injection of the larvae with MRSA results in an overall decreased in the lipid weight of the larvae.(A) Statistical results from a one-way ANOVA are illustrated above the bars as compared to the NM control.Summary of multiple analysis can be found in Table S2.(B) Box-plots above and to the right of the scatter plot are to illustrate the distribution of the data.Colours are as follows: black, NM control; purple, PBS control; blue, live larvae; and red, dead larvae 24 h post-MRSA infection.Correlations of infective dose and lipid weight for (C) living and (D) dead larvae.Data is presented as Log[CFU], as the infective doses are not adjusted for larvae weight.Data presented as mean ± SD (n = 10) of three independent replicas.

Figure 6 .
Figure 6.The weight did not influence the probability of pupation of NM larvae.NM larvae were incubated at 37ºC for 15 days and observed daily for pupation events.No significant difference was found between the weight group and the probability of pupation as calculated by a Log-rank Mantel-Cox test (p = 0.5489).Data are shown as mean ± SD (n = 10) repeated twice.

G
. mellonella response to MRSA.Prior investigation has evaluated the effect of nutrient deprivation on larvae(Banville, Browne and Kavanagh 2012), and the selection of diet(Jorjão et al. 2018), which both influence susceptibility to S. aureus infection.This emphasises the issues associated with a having lack of knowledge of rearing conditions used by suppliers and how they will influence experimental results.TruLarv™ (BioSystems Technology, UK) currently provide the only standardised G. mellonella in the UK.While cheap compared to murine models, it is considerably costlier (£1.20 per larvae) than purchasing larvae from commercial pet food providers.
Calculating the LD50 for MRSA NCTC 12493 Equation as given by linear model: y = 0.007966x + 5.548 Larvae weight (mg): x = 180 Insert here the median weight for the selected weight group LD 50 (Log[CFU/mg]): y = 9.61E+06This is the LD 50 for your selected weight group LD 50 (CFU): 2.88E+09

Table 1 .
Summary of the LD50s as calculated by non-linear models for each weight group (N, number of replicas; R 2 , coefficient of determination).

Table S2 .
Multiple comparison results for change in lipid weight 24 h post-MRSA infection described in Figure 5. (PBS; phosphate-buffered saline injection; NM, no manipulation control)

Table S3 .
Editable excel table that can be used in order to calculate the LD50 for MRSA based on weight-grouping.