G257, a carbapenem-resistant A. baumannii clinical isolate from our laboratory collection, was used in all studies. Two ATCC reference strains (A. baumannii ATCC19606 and A. lwoffii ATCC15309) were used as controls. In fitness assays, G422, a multidrug-resistant (MDR) carbapenem-resistant A. baumannii clinical isolate from our laboratory collection, was shown to have low fitness in a previous study (Nutman et al., 2020) (Table 1).

For various experiments, bacteria were cultured at 37°C on blood agar (tryptic soy agar supplemented with 5% sheep blood; Hylabs, Rehovot, Israel), MDR Acinetobacter agar (Hylabs, Rehovot, Israel), or brain–heart infusion (BHI) broth. Bacteria were grown at 37°C to the logarithmic phase (OD₆₀₀ of 0.6). Cells were spun down, washed with phosphate-buffered saline (PBS), and then diluted to indicate cell density. The infecting inoculum was freshly prepared before the experiment and was confirmed by CFU quantification with serial dilutions.

Zophobas morio was grown and maintained on sterilized sawdust (Meital Laboratories, Shekhanya, West Galilee, Israel). The larvae, in the final instar larval stage, were shipped and maintained in a plastic box with ventilation openings, at room temperature. Freshly washed vegetables were provided every 2 days for hydration. Larvae with an average weight of 640–800 mg were used within a month after arrival.

G. mellonella larvae in the 3-day instar larval stage (Biosystems Technology, Brierley Hill, UK) were stored at 25°C in a plastic box with ventilation openings. Larvae with an average weight of 200–300 mg were used within a week after arrival.

Larvae were randomly divided into treatment groups of 30 larvae each. A 0.5 mL, 31 G syringe was used to inject 25 μL of aliquots of bacteria into the proleg of each larva. After injection, larvae were incubated in plastic containers at 25°C or at 37°C as specified, and the number of dead larvae was scored daily until the end of the experiment (days 5–7). Larvae were considered dead when they displayed no movement in response to touch.

Groups of 5 G. mellonella larvae were infected with 10⁵ CFU bacteria/larvae. Groups of 4 Z. morio larvae were infected with 10⁷ CFU/larva. Larvae were collected at 0 h and 4 h post-infection and homogenized individually in 1 mL of PBS using the gentleMACS Dissociator (Miltenyi Biotec, Bergisch Gladbach, Germany). Serial dilutions of homogenate in PBS were plated on MDR Acinetobacter plates and incubated at 37°C for 24 h, followed by CFU quantification. Non-manipulated larvae of each species (n = 4) and PBS-injected larvae of each species (n = 4) served as controls.

We tested in vivo growth using a neutropenic murine thigh infection model. Fitness was defined based on the CFU count 24 h after the injection of an inoculum of 10⁵ CFU. A detailed protocol was described previously (Nutman et al., 2020). The protocol was approved by the Institutional Animal Care and Use Committee at Tel-Aviv Sourasky Medical Center.

Larvae were injected with PBS (control) or with 10⁶ CFU of G257 (infected). At time points of 0 and 16 h post-infection, larvae were washed with PBS, fixed with 70% ethanol, and stored at 4°C. Paraffin-embedded slides were prepared (Hepatho Lab Ltd., Rehovot, Israel), and H&E staining was performed as described (Barnoy et al., 2017).

Zophobas morio larvae (30 per group) were injected into a proleg with 10⁷ CFU/larva of the bacterial strain. At the 30 min post-infection time point, a single dose of antibiotics, either 200 μg/mg ampicillin or 60 μg/mg meropenem, was injected into a different proleg. Z. morio larvae injected with PBS (30 per group) were used as controls. Survival and fitness assays were performed as described above. At least three independent experiments were performed with similar results.

Statistical analysis was performed using GraphPad Prism 9 software (version 9.3.1, GraphPad Software Inc., San Diego, CA, United States). The Kaplan–Meier survival curves were generated. For two-group comparisons, a two-tailed unpaired t-test was applied. For multigroup comparisons, an analysis of variance (ANOVA) with Turkey–Kramer’s multiple-comparison test was used. A value of p of <0.05 was considered statistically significant. To build a phylogenetic tree, the cytochrome C amino acid sequences of each organism were retrieved from NCBI and aligned using MAFFT (Katoh et al., 2002). A phylogenetic tree was constructed using the NJ algorithm. Accession numbers were as follows: Homo sapiens AAP36314.1; Mus musculus NP_031834; Gallus NP_001072946.1; Drosophila melanogaster NP_001260509.1; Danio rerio NP_001002068.1; Caenorhabditis elegans P19974.2; Rattus norvegicus NP_036971.1; Galleria mellonella XP_052756993.1; Zophobas morio GCA_027724725.1 (Cytochrome C amino acid sequence was extracted from the full genome).

We constructed a phylogenetic tree to measure the distance between Z. morio and other invertebrate and vertebrate animals used as infection models (Figure 1A). The phylogenetic distance from humans to both Z. morio and G. mellonella is comparable and shorter than the distance from C. elegans to humans. We then compared key phenotypic features of the Z. morio larvae to those of G. mellonella. Z. morio is larger and darker than G. mellonella (Figure 1B). The average weight of the Z. morio larvae was nearly 2.5-fold higher than the average weight of the G. mellonella (758 ± 14.2 mg vs. 308 ± 8.4 mg) (Figure 1C). The differences in size and weight allowed for easier handling and injection in the Z. morio model. Like G. mellonella, dead Z. morio larvae are easily distinguishable from live ones—they appear straight and rigid and do not respond to tactile stimuli (Figure 1D). The differences between the two models and their comparison to C. elegance are presented in Supplementary Table S1.

We used two laboratory reference strains as controls: a non-virulent Acinetobacter lwoffii ATCC15309 and Acinetobacter baumannii ATCC19606 (Table 1). As a clinically relevant strain, we used a previously described (Nutman et al., 2020; Frenk et al., 2022) A. baumannii G257 strain isolated from a patient with severe infection. This strain was previously found to harbor various virulence factors (Supplementary Table S2) and to be highly virulent in a mouse model, with a fitness of 10.08 CFU/thigh.

We first established a survival assay. Infection was induced in Z. morio by injecting with the three study strains—A. lwoffii ATCC15309, A. baumannii ATCC19606, and the clinical A. baumannii isolate G257. Groups of 30 larvae were injected with 10⁵ CFU–10⁸ CFU of bacteria and incubated at 25°C or at 37°C for the indicated number of times. A group of 30 larvae injected with PBS and a group of 20 larvae without any manipulation were used as controls. Figure 2 shows the lethal dose effect of the three strains in a dose-dependent and temperature-dependent manner. A clear dose–response was observed for all three strains, as well as differences in the lethality between the strains (Figures 2A–C). At an inoculum of 10⁵ CFU/larvae, lethality was observed only among larvae infected with A. baumannii G257 (80% survival on days 5–7). At an inoculum of 10⁶ CFU/larvae, all larvae infected with A. lwoffii ATCC15309, 90% of those infected with A. baumannii ATCC19606, and only 40% of those infected with A. baumannii G257 survived at day 7. At an inoculum of 10⁷ CFU, higher lethality was observed in all three groups. The effect of temperature varied for different inoculums. At an inoculum of 10⁶ CFU, no significant differences in the survival of larvae at 25°C or 37°C were observed (value of p >0.05) (Figures 2D–F). At an inoculum of 10⁷ CFU/larvae, ATCC19606 was significantly more lethal at 37°C than at 25°C (p value <0.05) but not for the two other strains used (value of p = 0.29 for ATCC15309 and p value = 0.406 for G257) (Figures 2G–I).

We next determined the in vivo fitness of A. baumannii in Z. morio by quantifying bacterial CFU 4 h after injection. As the larvae are not sterile, to specifically quantify the injected A. baumannii, we plated the bacteria on a selective medium (containing meropenem 4.5 mg/mL). As both reference strains used in this study are sensitive to meropenem, we used a clinical MDR isolate G422 as a control. G257 has high fitness in a mouse model (10.08 CFU/thigh), while G422 has lower fitness in a mouse model (5.23 CFU/thigh) (Figure 3C). The fitness of the two isolates in Z. morio larvae was compared with their fitness in G. mellonella larvae and was consistent with that of the two isolates in the mouse model (Figure 3). G422 was less fit, and both models did not show an in vivo increase in bacterial count. In contrast, strain G257 showed a significant increase in cell counts in both models (there was a 0.5-log increase in G. mellonella and a one-log increase in the Z. morio model, p < 0.05).

To visualize structural alterations, cellular abnormalities, or tissue damage caused by the infection, we visualized whole body sections of infected Z. morio larvae with H&E staining. At 16 h post-infection with the A. baumannii G257 strain, H&E stained cross-sections of larvae revealed clusters of hemocytes exhibiting phagocytosis of bacteria (blue dots throughout the section), with a small area of melanization around hemocyte aggregates (Figures 4A,B). Moreover, we observed hemocytes present in nodules, with some attached to organ structures with evidence of melanization (Figures 4C,D,G). In contrast, cross-sections of H&E-stained PBS-injected larva looked healthy (mock infection), with individually distributed hemocytes showing no noticeable aggregates of melanization (Figures 4E,F,H).

We used this novel infection model to test the efficacy of antimicrobials against Acinetobacter spp. We tested the efficacy of ampicillin and of meropenem against A. baumannii ATCC19606 (ampicillin-resistant and meropenem-susceptible, MIC = 1 mg/L for meropenem and MIC >32 mg/L for ampicillin) and against A. baumannii G257 (ampicillin and meropenem resistant, MIC =128 mg/L for meropenem and MIC >32 mg/L for ampicillin). We used a lethal inoculum (5×10⁷ CFU) and compared Z. morio survival after a single dose of ampicillin (200 mg/kg) or meropenem (60 mg/kg).

By day 3, there were no surviving larvae of Z. morio infected with A. baumannii ATCC19606. Treatment with ampicillin, to which this strain is resistant, had no effect on larvae survival. On the other hand, treatment with meropenem, to which this strain is sensitive, significantly increased larvae survival (p < 0.001), with 45% surviving at day 5 (Figure 5A). All Z. morio infected with MDR A. baumannii G257 died by day 3, and treatment with either antibiotic had no effect on their survival (Figure 5B).