The established mycobacterial virulence factor EccB1 is required for Mmar virulence in Drosophila.

For Drosophila to be a relevant host model to study mycobacterial virulence, established mycobacterial virulence factors should be required for Mmar infection in the fly. Hence, we infected Drosophila with an Mmar (strain E11) mutant in EccB1. EccB1 (encoded by eccB1) is a core component of the 6 kDa early secretory antigenic target (ESAT-6) Type VII secretion system 1 (ESX-1) [36]. ESX-1 is a hallmark of Mtb virulence through secretion of specific protein substrates, such as CFP-10 and ESAT-6, across the complex mycobacterial cell wall [37]. Indeed, we showed that disruption of Mmar eccB1 (transposon insertion mutation; eccB1::tn) clearly increased the survival of flies upon infection, compared to Mmar wild type (wt) infection (Fig 1A). Correspondingly, the eccB1 mutant showed attenuated growth within the flies (Fig 1B), most probably explaining the prolonged survival of the mutant-infected flies. The wt and eccB1::tn mutant strains grew comparably in vitro in 7H9 medium (Fig 1C). Hence, we confirmed the attenuated phenotype of an ESX-1-disrupted mutant during Mmar Drosophila infection.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. EccB1 is required for Mmar virulence in Drosophila.

Male Drosophila infected with 5000 CFU/fly of Mmar E11 strain wt (dsRed), eccB1::tn, a 1:1 mix of wt (dsRed) and eccB1::tn, or PBS. Fly survival (A) and CFU per fly (B) was recorded over the course of infection. For fly survival (A), data represent the percent survival of initial 30 flies per condition and were recorded daily. For CFU measurements (B), wt (dsRed) and eccB1::tn were selected on hygromycin or kanamycin, respectively, and data represent means ± standard deviation (SD) for three individual flies per condition and time point. (C) Mmar E11 wt, wt (dsRed), and eccB1::tn growth in vitro (7H9 medium) as measured by OD₆₀₀. Data represent means ± SD of three technical replica recorded every 24 hours.

https://doi.org/10.1371/journal.ppat.1013474.g001

Drosophila is suitable for genome-wide screening for mycobacterial virulence genes

Next, we sought to determine whether Drosophila is suitable to identify novel mycobacterial virulence genes. By TnSeq, we and others have previously identified mycobacterial virulence genes in a genome-wide manner using mice [38–41], cattle [42], or amoeba [43–45], as host infection models. Discovery of virulence genes by TnSeq is based on infecting the host with a bacterial high-density transposon (tn) mutant library, followed by harvesting the library for analysis after infection, comparing its mutant constituents to the input library using massive parallel sequencing [46]. Genes with tn mutations that cause bacterial growth-impairment within the host but not on standard agar medium, are here defined as virulence genes. Hypothetically, a mutant deficient in for instance ESX-1 secretion could be rescued by mutants sufficient in ESX-1 secretion residing within the same plasmatocyte. To investigate whether this could obscure our screen for virulence genes when we pass a high number of mutants through one single fly, we infected Drosophila with 5000 colony forming units (CFU) per fly of either Mmar wt, eccB1::tn mutant or a 1:1 mix of Mmar wt and eccB1::tn mutant. By using an Mmar wt strain carrying a plasmid conferring hygromycin resistance and dsRed expression, wt (dsRed), in combination with the eccB::tn mutant carrying kanamycin resistance encoded within the tn insert, we could by antibiotic selection separate the growth of the mutant from wt growth during co-infection. As seen in Fig 1B, the eccB1 mutant was still growth-attenuated within the flies when co-infected with Mmar wt, while the co-infected flies died similarly to those infected by wt only (Fig 1A). All strains grew comparably in 7H9 liquid medium (Fig 1C). From this, we take that virulence mutants are not necessarily rescued by co-infection with other mutants intact in their version of the virulence gene in question, although this may indeed vary depending on the mechanism of the gene. By day seven post infection the wt bacteria had doubled approximately seven times, enabling us to differentiate between mutants in virulence and non-virulence genes during TnSeq screening (Fig 1A). Taken together, our results suggest that Drosophila is a suitable host model to screen for mycobacterial virulence genes.

Genome-wide identification of Mmar virulence genes in Drosophila

We aimed to identify Mmar virulence genes in Drosophila. Hence, we constructed and sequenced a high-density tn insertion library in the Mmar E11 strain using ϕMycoMarT7 [47]. The obtained library contained mutants in 80% of TA sites (ϕMycoMarT7 insertion sites) covering 96.5% of the genes. We uncovered 430 essential (ES) genes, 130 genes conferring growth defect (GD) when disrupted, 4314 non-essential (NE) genes, and 79 genes conferring growth-advantage (GA) when disrupted (S1A Dataset). Combining ES and GD categories, the number of 560 genes that are essential or cause a growth-defect when disrupted, is in line with what has been observed in Mtb [48,49]. In a previous TnSeq study of the Mmar E11 strain [44], 300 genes were identified as being essential in vitro, of which 82% (247) were also in the ES or GD category in our data (S1A Dataset).

To specifically identify Mmar virulence genes during Drosophila infection, we passed the generated library through Drosophila (250 male or female flies, 5000 CFU/fly), covering the mutant library 17 times across the 250 flies. We subjected the input and output libraries to TnSeq, the latter harvested from the flies seven days post infection. To show reproducibility when spreading the tn library across several hosts, we created three scatter plot matrices showing high correlation of the gene-level mean tn insertion counts among the three biological replicates of each condition (in vitro input library and in vivo output libraries passed through male or female flies). The correlations in these plots demonstrate similar estimates of gene requirement across experimental replicates, suggesting that to spread the library over several hosts has no negative impact on the downstream analysis and result (S1 Fig).

We found that 181 genes were required for optimal growth within the flies, based on a permutation test (“resampling analysis” [50]) of the difference in mean tn insertion counts per gene in libraries that had undergone fly infection (output) versus libraries grown under in vitro condition (input) (Log₂ fold change <0 and adjusted P-value <0.05) (S1B Dataset and Fig 2A–E). Among the virulence genes identified were those encoding established mycobacterial virulence factors, such as by phthiocerol dimycoceroserate (PDIM, a cell wall lipid, mmar_1767, _1770–1771), components of the ESX-1 secretion system (10 genes between mmar_5399–5459), and the LytR-CpsA-Psr domain-containing protein CpsA (mmar_4966) (Fig 2C). Moreover, all the four genes encoding the succinate dehydrogenase Sdh2 (mmar_1200–1203) as well as seven genes involved in cobalamin (vitamin B12) biosynthesis were hits in our screen (Fig 2C). We also identified genes (15) that conferred Mmar growth advantage within Drosophila when disrupted (Log₂ fold change >0 and adjusted P value <0.05), making them the mere opposite of virulence genes (S1B Dataset and Fig 2A, 2C–2E). Examples of these are ESX-1’s eccA1 (mmar_5443) and espH (mmar_5442), and the two-component signal transduction system trcS/trcR (mmar_4455–4456) (Fig 2C). When we compared the virulence gene set of male versus female flies, we found three genes that were required for virulence in males only (S1C Dataset). Of these, two did not have Mtb orthologs, while the third encodes the NADH-quinone oxidoreductase subunit NuoF. Taken together, we identified 181 genes as required for full Mmar virulence during Drosophila infection, and 15 genes that conferred a growth advantage in the fly when disrupted.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Genome-wide identification of Mmar virulence genes in Drosophila.

(A) Volcano plot of Mmar E11 virulence genes in male Drosophila based on three independent experiments of 250 flies infected with 5000 CFU/fly of the Mmar E11 tn mutant library. The Log₂ fold change indicates fold change of the mean normalized tn insertion counts per gene of Mmar selected in vitro (7H9) versus in vivo (after Drosophila infection). Each dot represents one gene. Dots in color represent genes with an adjusted P-value (using Benjamini-Hochberg correction, FDR < 0.05) <0.05. Red dots represent genes that are in vivo attenuated when disrupted (e.g., virulence genes), while green dots represent genes that confer in vivo growth advantage when disrupted. (B) Venn diagram illustrating Mmar virulence genes in Drosophila (red) overlayed the entire pool of Mmar (E11 strain) and Mtb (H37Rv strain) genes and their mutual orthologs. (C) List of Mmar E11 genes with adjusted P-value <0.05 that have Mtb mutual orthologs. Selected pathways are highlighted with black frames, IRON; genes involved in low iron growth, ESX-1; genes involved in ESX-1 secretion, B12; genes involved in cobalamin biosynthesis, PDIM; genes involved in PDIM synthesis, SUCC DEHYDR; genes encoding succinate dehydrogenase, TWO COMP; two-component signal transduction system. (D) List of Mmar E11 genes with adjusted P-value <0.05 but no Mtb mutual orthologs. (E) Plot of each Mmar E11 gene’s Log₂ fold change respective to genomic position (from mmare11_00010 to mmare11_53170). Dots in red represent Mmar E11 genes with an adjusted P-value <0.05 and Mtb orthologs, while blue dots represent Mmar E11 genes with an adjusted P-value <0.05 but no Mtb mutual ortholog. Genes that are either discussed in this study or have particularly high/low Log₂ fold ratios are highlighted.

https://doi.org/10.1371/journal.ppat.1013474.g002

The relevance of Mmar Drosophila virulence to Mtb infection

To investigate the potential relevance of the Mmar virulence genes in Drosophila to Mtb infection, we compared them to the entire Mtb gene pool (Fig 2B). There are 2981 mutual orthologs shared between Mmar E11 and Mtb H37Rv, where a mutual ortholog here means that the gene in one organism is the best match for the ortholog in the other organism and vice versa, with a BLAST E value of <10⁻¹⁰. An unbiased proportion of Mmar virulence genes with Mtb mutual orthologs would be ~ 60%, while we found that that 91% (163/181 genes) had mutual orthologs in Mtb (Fig 2B–E). Moreover, when we compared the Mmar virulence genes to those previously identified by TnSeq for Mtb, 48% had Mtb orthologs that were defined as required for full in vivo growth in mice [39,41] (S1B Dataset). Together, our findings suggest that mechanisms of the identified virulence genes could be conserved across Mmar and Mtb and across the host model species in question.

mmar_1660 is a novel mycobacterial virulence gene

To validate our screen, we aimed to create a targeted knockout mutation of one of the virulence genes identified. Due to our previous experience and interest in mycobacterial low iron growth [51–53], we chose mmar_1660 which encodes a putative conserved ATP-binding protein ABC transporter whose Mtb ortholog is suggested by homology to be involved in metal transport [54,55] and specifically transport of iron [56–58]. For reasons unknown we were unable to mutagenize the Mmar E11 strain in a targeted manner using bacteriophage-mediated allelic exchange, while we succeeded in the Mmar M strain. Hence, we infected Drosophila with 500 CFU/fly Mmar M wt or Δ1660, and found that Δ1660-infected flies survived on average one day longer than wt-infected flies (Fig 3A). The strains grew comparably in vitro in standard liquid medium (Fig 3B). The Mmar M strain has previously been reported to grow faster than the E11 strain under hypoxic conditions [59]. In line with this, we observed that the M strain killed Drosophila faster than the E11 strain by an average death after 6 days of 500 CFU/fly M strain infection compared to average death after 8 days of 5000 CFU/fly E11 infection (Figs 1A, 3A). Our screen showed that mmar_1660 is among the genes giving rise to a moderate virulence phenotype when disrupted, with a Log₂ fold change of -1.54 (Figs 2B, 2E). Thus, the modest but strongly significant one-day in vivo attenuation of the M strain Δ1660 mutant’s growth (Fig 3A) in comparison to the 13-day attenuation of the E11 eccB1 mutant (Fig 1A) may be due to an inherent faster growth rate of the M strain – shortening the window to detect virulence phenotypes – in combination with a less prominent role of mmar_1660 during infection. Even so, these results validate mmar_1660 as a virulence gene during Drosophila infection, a trait that transfers from the E11 to the M strain.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. mmar_1660 is required for full Mmar virulence in Drosophila and human macrophages.

(A) Male Drosophila infected with 500 CFU/fly of Mmar M strain wt, Δ1660 or PBS, with their survival recorded over the course of infection. Data represent the percent survival of initial 30 flies per condition and were recorded daily. The four asterisks represent a statistically significant difference with a P value <0.0001 between wt and Δ1660-infected flies as calculated by Log-rank Mantel-Cox testing using GraphPad Prism 9. (B) Mmar M wt and Δ1660 growth in vitro (7H9 medium). Data represent means ± SD of three technical replicates recorded every 24 hours. Human iPSDMs were infected with Mmar M strain wt or Δ1660 at a multiple of infection of 4:1 (macrophage:bacterium) under normal cell culture conditions (C) or with 100 μM FeAC added to the cell culture medium (D). C and D represent the combined averages of three technical triplicates from three independent experiments. Statistically significant difference was calculated using the Mann Whitney U-test on the datasets of day one post infection (unpaired, one-tailed). ns, not significant. *, P-value ≤ 0.05.

https://doi.org/10.1371/journal.ppat.1013474.g003

Furthermore, we wanted to investigate whether mmar_1660 contributed to virulence in human macrophages. We infected human macrophages derived from induced pluripotent stem cell-derived macrophages (iPSDMs) with the Mmar M wt and Δ1660 strain, and found that at day one post infection the wt but not the Δ1660-mutant had proliferated within the macrophages, indicating that the mutant is growth-impaired within the cells (Fig 3C). In summary, we show that mmar_1660 is required for full virulence in Drosophila as well as in human macrophages.

mmar_1660 and its Mtb ortholog rv3041c are required for low iron growth

mmar_1660’s Mtb ortholog rv3041c encodes a conserved ATP-binding ABC transporter of unknown function suggested to be involved iron transport [54–58]. To test whether these genes are indeed required for growth under iron limitations, we created a frameshift (fs) mutant causing a premature stop codon within Mtb rv3041c and subjected the Δ1660 and the rv3041c fs mutants to growth in low iron in vitro. Both mutants showed impaired growth when iron was low, compared to their respective wt strains, while under high iron conditions (i.e., same level as in 7H9) they grew similarly to their respective wts (Fig 4A–D). The initial rise in OD₆₀₀ values within the first 3 days in low iron medium is likely due to bacteria sedimenting at the bottom of the 96- or 100-well (honeycomb) plates used for Mtb and Mmar incubation, respectively, rather than actual bacterial growth.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. The orthologs mmar_1660 and rv3041c are required for low iron growth.

Mmar M (A, B) and Mtb (C, D) strains were grown in high (150 μM or 300 μM FeCl₃, respectively) or low iron media as indicated. Mtb low iron media contained 1 μM DFO. Data represent means ± SD of three technical replicates of which OD₆₀₀ was recorded regularly at the time points indicated in the figure.

https://doi.org/10.1371/journal.ppat.1013474.g004

To support the notion that the observed phenotypes were caused by the introduced mutations, we whole genome sequenced the strains. We successfully identified the introduced deletions in both mmar_1660 and rv3041c (S2 Dataset). In addition to the introduced rv3041c fs mutation, we identified in this strain two single nucleotide polymorphisms (SNP) that distinguished the strain from the wt, creating an amino acid change in rv2217 (lipB, amino acid 51 of 230 mutated from alanine to glutamic acid) and rv3452 (cut4, amino acid 80 of 226 mutated from asparagine to threonine). Similar mutations were not found in the orthologous genes (or any other genes) in the Mmar Δ1660 strain, strongly reducing the probability that one or both of the additional mutations contributed to the observed low iron phenotype in Mtb. Together, our results show that mmar_1660 and its Mtb ortholog rv3041c are required for mycobacterial growth when iron availability is low.

mmar_1660 and rv3041c are required for virulence in human macrophages in an iron-dependent manner

Iron acquisition is critical for survival and replication of Mtb in phagosomal containment of macrophages during infection [34,35]. Hence, to investigate whether rv3041c is required for Mtb virulence, we infected iPSDMs with the rv3041c fs and wt strains and found that the fs mutant was growth-impaired within the cells compared to the wt (Fig 5A-B). The role of iron in mycobacterial infections has been studied by others by supplementing iron to in vitro cell culture models [60–62]. When we applied the same principle to our Mmar and Mtb iPSDM infection models, we found that the Mmar Δ1660 mutant’s intracellularly attenuated phenotype was fully rescued (Fig 3D) while the orthologous rv3041c fs mutant´s phenotype was partly rescued in macrophages cultivated in the presence of additional iron supplement (ferric ammonium citrate, FeAC) (Fig 5A). For rv3041c, this was evident when comparing the fold difference in CFU/ml lysate between normal and high iron conditions at day 6 post infection, where under standard conditions the fold difference between wt and fs mutant was 8.3 compared to only 2.3 when FeAC was added (Fig 5B). These results demonstrate that mmar_1660 and rv3041c are less required for intra-macrophagic growth when iron is plentiful. In summary, mmar_1660 and rv3041c are required for full virulence in human macrophages in a manner that depends on the host cell’s iron status.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. rv3041c is required for full virulence in human macrophages in an iron-dependent manner.

Human iPSDMs were infected with Mtb wt or rv3041c fs strain at a multiple of infection of 2:1 (macrophage:bacterium). After 2-hour infection medium was supplemented with either 100 µM FeAC or no supplementation. (A) Intracellular bacterial burden was assessed by CFU enumeration of cell lysates at the given time points post infection. Data represents means ± SD from three technical replicates per condition. The data is representative of two (for FeAC) or three (for untreated) independent experiments (B) The relative growth differences between wt and rv3041c fs in the different conditions were determined on day 6 post infection. 8.3x and 2.3x indicate the fold difference between wt and rv3041c fs in unsupplemented (normal iron) and 100 μM FeAC supplemented (high iron), respectively. Statistically significant difference was calculated using the Mann Whitney U-test on the datasets of day 6 post infection (unpaired, one-tailed). *, P-value ≤ 0.05.

https://doi.org/10.1371/journal.ppat.1013474.g005

Future perspectives

We used Drosophila to genome-widely identify mycobacterial virulence genes during host infection, complementing its role in studying innate immune responses towards mycobacterial infections. As a whole animal model, Drosophila hence provides a valuable platform for studying HPIs that affect mycobacterial disease outcomes, for instance by pinpointing specific genetic interactions between mycobacterial virulence factors and Drosophila host response mechanisms using TnSeq in combination with Drosophila host mutants. This capability was exemplified by Moule et al. in their Transposon Site Hybridization analysis of Francisella novicida genetic interaction with the Drosophila Imd signaling pathway [84]. Apparent limitations of the host model, such as the lack of adaptive immunity, may rather create an advantageous opportunity to specifically study HPIs relevant to innate immunity in an isolated yet in vivo setting. Of course, careful consideration in light of current available knowledge must be taken to assess the aptness of Drosophila to answer the individual research questions towards better understanding mycobacterial HPIs.

Bacterial strains and growth conditions

The Mycobacterium marinum (Mmar) wt strains used in this study were Mmar E11 and Mmar M (NCBI GenBank accession numbers CP000854.1 and HG917972.2, respectively). The Mycobacterium tuberculosis (Mtb) strain was Mtb H37Rv (NCBI GenBank accession number NC_000962). The Mmar wt strains, in addition to the mutant strains Mmar E11 eccB1::tn (kanamycin resistant, [44]) and Mmar E11 dsRED (pSMT3dsRed, hygromycin resistant), were kind gifts from Wilbert Bitter at Vrije Universiteit Amsterdam. Mmar strains were cultured in Middlebrook 7H9 (Becton Dickinson, BD) supplemented with 0.2% glycerol, 0.05% Tween 80 and 10% ADC (50 g BSA fraction V, 20 g dextrose, 8.5 g NaCl, 0.03 g catalase, dH₂O up to 1 L) for liquid growth. For solid growth, Middlebrook 7H10 (Becton Dickinson, BD) supplemented with 0.5% glycerol and 10% ADC or OADC (Becton Dickinson, BD), for Mmar or Mtb, respectively, was used. 20 μg/ml kanamycin, 50 μg/ml hygromycin, and 100 ng/ml anhydrotetracycline were added where required. In vitro Mmar growth curves were performed as previously described for M. avium [38], using the Bioscreen growth curve reader (Oy Growth Curves Ab Ltd.), with the exemption of temperature set at 30°C instead of 37°C. Mtb strains were cultivated as Mmar in Middlebrook 7H9 except for OADC (Becton Dickinson, BD) replacing ADC. in vitro Mtb growth curves were performed using Infinite M Plex microplate reader (TECAN). Experimental work using Mtb was carried out according to Norwegian regulation for organisms classified as Biosafely Level 3 (BSL3). The dedicated BSL3 laboratory is approved for BSL3 work with Mtb (Reference 12/8648–9 – Norwegian Directorate of Health) and is located within a restricted area for authorized personnel.

Low iron growth conditions

Mmar and Mtb strains were cultured in Middlebrook 7H9 with supplements as described above until mid log phase. The strains were washed twice in low iron medium prepared as following: 5 g L-asparagine, 5 g KH₂PO₄, 0.25% glycerol, 0.05% Tween 80, 0.5 g BSA, and 0.2 g dextrose was dissolved in 900 ml H₂O. The solution was adjusted to pH 6.8 with NaOH and H₂O was added to reach 1000 ml. The solution was chelated with magnetic stirring for 1–3 days with 20–50 g Chelex 100 resin (Bio-Rad Laboratories) before filtered (0.2 μM pores) into a plastic bottle. Finally, 50 μl 10 mg/ml ZnCl₂, 50 μl 2 mg/ml MnCl₂ and 200 μl 200 mg/ml MgSO₄, all prepared in iron free water in plastic containers and sterile filtered, was added. For Mtb growth curves, the strains were subcultured once (after wash) to mid log phase before back-diluted to OD₆₀₀ 0.02. Deferoxamine (DFO, Sigma Aldrich) was furthermore added to the low iron medium at a final concentration of 1 μM. For Mmar, the strains were directly back-diluted to OD₆₀₀ 0.02 for growth curve experiments. FeCl₃ was added to the high iron growth curves to a final concentration of 150 μM for Mmar and 300 μM for Mtb.

Mmar_1660 knockout strain

To create the mmar_1660 knockout strain (Δ1660), bacteriophage-mediated allelic exchange using pYUB1471 and phAE159 was used (kind gift from William R Jacobs at Albert Einstein College of Medicine), following published protocols [85]. In short, around 1000 bp upstream and downstream of mmar_1660 were amplified, and cloned into pYUB1471, creating pYUB_1660. pYUB_1660 was further cloned into phAE159, which was used to transduce Mmar M to induce allelic exchange in order to replace the mmar_1660 gene with genes encoding hygromycin resistance (hygR) and levansucrase (sacB) genes. Positive clones were selected on 7H10 agar containing hygromycin and further validated by PCR.

Mtb rv3041c fs knockout strain

To create the rv3041c knockout strain (rv3041c fs), sgRNA targeting the gene was cloned into pCRISPRx-Sth1Cas9-L5 (addgene plasmid #140993) using BsmBI restriction sites and oligonucleotides aaacaacgtgtccctgcgccgtaat and gggaattacggcgcagggacacgtt according to published protocols [86] creating pCRISPR_3041c. pCRISPR_3041c was subsequently transformed into Mtb H37Rv wt and selected on 7H10 agar plates containing 20 μg/ml kanamycin. After incubation at 37°C for 3–4 weeks, colonies were re-streaked on 7H10 agar plates containing 100 ng/ml anhydrotetracycline and incubated 37°C for 3–4 weeks. Arising colonies were analyzed for rv3041c fs mutations by PCR against rv3041c using oligonucleotides gttgggcgtgactcgcgg and acgctgatccggcgtccg. PCR products were purified using AMPure XP beads (Beckman Coulter) and sent for Sanger sequencing using the gttgggcgtgactcgcgg oligonucleotide. Mutations were identified using Synthego ICE (Synthego Performance Analysis, ICE Analysis. 2019. v3.0. Synthego). The selected mutant had a 5 base pair deletion within the rv3041c gene, creating a fs and a subsequent early stop codon after 30 amino acids (the full length rv3041c gene product is 288 amino acids).

Whole genome sequencing

Genomic DNA was prepared using Epicentre MasterPure Complete DNA Purification kit. Sequencing libraries were prepared using the Illumina DNA prep kit and sequenced on an Illumina MiSeq instrument using the MiSeq Reagent Kit v3 with 300 bp paired end configuration. Raw data were quality controlled using FastQC v0.11.9 [87]. Reads were mapped, filtered and variants called relative to the respective reference genomes (CP000854.1 for Mmar M strain and NC_000962.3 for Mtb H37Rv) using Snippy v4.4.3 [88]. Average coverage for each sample was calculated using Samtools v1.9 [89]. Large deletions were additionally identifed by defining regions with zero coverage using Geneious Prime v2023.0.4 (https://www.geneious.com). For Mtb strains, variants were filtered using tb_variant_filter v0.4.0, excluding PE/PPE genes [90], UVP repetitive loci and Refined Low Confidence (RLC) plus low mappability regions. For Mmar strains, all CDS annotated as PE/PPE-family proteins were excluded. For Mtb and Mmar knockout strains, variants that were also detected in wt or in very low coverage regions (<10x) of the wt strain were disregarded. WGS sequence data are available from the NCBI Sequence Read Archive under BioProject PRJNA1068332.

Mmar tn mutant library

The high-density tn mutant Mmar E11 library was prepared as previously described for M. avium [38], using ϕMycoMarT7 (kind gift from Eric Rubin at Harvard T.H. Chan School of Public Health [47]), except for heating medium and bacterial cultures to 30°C as opposed to 37°C prior to transduction, and incubating the library at 30°C as opposed to 37°C on 7H10 plates with Tween 80 (0.05%) and kanamycin (20 μg/ml) for 2 weeks.

Bacterial single-cell suspension for infection

Mmar strains and the Mmar E11 tn mutant library were prepared to obtain single-cell suspensions prior to Drosophila infections. Strains were cultured to stationary phase, pelleted down by centrifugation, and resuspended in 7H9 medium with 0.2% Tween 80 to resolve larger clumps. Cultures were then pelleted again and resuspended in 7H9 media with 15% glycerol. The resuspension sat at room temperature for 30 minutes to let clumps fall to the bottom before the supernatant was transferred and aliquoted to cryotubes for storage at -80˚C. To calculate CFU/mL of the infection stocks, they were diluted and spotted on 7H10 agar plates to determine the CFU/mL for Drosophila infection. For iPSDM infection, Mtb and Mmar M strains were prepared to obtain singe-cell suspension at the day of macrophage infection. Mid-log cultures, pelleted down and resuspended in RPMI 1640 with 10% FCS, underwent vortexing (2 times for 30 s), sonication at the power of 70% (2 times for 5 s), and vortexing 2 times for 30 s again. The cultures were centrifuged at 300 g for 4 min as a final clump-removing step. The CFU/mL of Mtb-containing supernatants used to infect cells were calculated by an OD_600-conversion of 1 = 4.5 x 10⁸ CFU/mL). The CFU/mL of Mmar-containing supernatants used to infect cells were calculated by an OD_600-conversion of OD₆₀₀ 1 = 2 x 10⁷ CFU/mL.

Drosophila strains

The Drosophila melanogaster (Drosophila) strain used for TnSeq was a cross between the Transgenic RNAi Project (TRiP) control fly AttP40 (y v; attP40, y + , stock #36304 at the Bloomington Drosophila Stock Center, BDSC) and a tubulin-Gal4 driver fly. For all other Drosophila infections, the Oregon R-C (Flybase ID FBsn0000276, stock #5 at the BDSC) was used. Flies were bread at 25ºC with constant light:dark cycles of 12 hours each and a humidity of 70%. The Bloomington standard cornmeal formulation containing yellow cornmeal, corn syrup solids, inactive nutritional yeast, agar and soy flour was used to feed flies (pre-mixed dry version available at Genesee Scientific).

Drosophila infection, survival and CFU assay

Infections were performed in 3–5-days-old flies. Anaesthetized flies were infected into the abdomen using a Nanoject II (Drummond Scientific Company) set to inject 13.8 nl using glass needles prepared using a PB-7 needle puller (NARISHIGE) and were not exposed to CO₂ anesthesia for more than 15 minutes during the process. Bacterial infection stocks were prepared as previously described [25] and diluted to 500 or 5000 CFU/13.8 nl in 1:1 ratio of phosphate-buffered saline (PBS) to Brilliant Blue (Sigma Aldrich). Flies were incubated at 29°C after infection. For CFU assays, flies were put briefly into 70% ethanol before washed once with PBS and homogenized in 100 μl PBS using a pestle. The homogenates were spotted in a 10-fold dilution series on 7H10 agar plates containing 20 μg/ml kanamycin and 1.25 μg/mL Amphotericin B (to eliminate the fly’s microbiota). For each CFU count, three flies per condition were harvested, homogenized and treated separately during dilutions and spotting. Two technical replica per fly were spotted. The agar plates were grown for around 10 days at 30°C before counting CFUs. For TnSeq, 250 flies were infected with 5000 CFU of Mmar E11 tn library per fly in three biological replica (3 x 250 infected flies). 10 and 10 flies were put briefly into 70% ethanol before washed once with PBS and homogenized in 200 ul PBS using a pestle. The homogenates from 10 and 10 flies were plated onto 15 cm diameter 7H10 agar plates containing 20 μg/ml kanamycin, 1.25 μg/mL Amphotericin B and 0.05% Tween 80, ending up with around 25 plates per library, and grown for 14 days at 30°C. The three biological replicas were treated separately throughout the experiment. For survival assays, 30 flies were infected per condition and the number of dead and live flies was noted every morning.

Transposon insertion sequencing, TnSeq

The tn library was harvested and pooled by scraping agar plates with colonies. Total DNA was purified using Masterpure DNA purification kit (Epicentre) and prepared for TnSeq by PCR amplification of tn-genome junctions and adapter ligation as previously described [91]. The samples were sequenced on an Illumina NextSeq 2000, generating around 12–15 million 150 + 150 bp paired-end reads per sample.

Bioinformatic analysis of TnSeq datasets

The reads were processed using TPP in TRANSIT [92], which counts reads mapping to each TA dinucleotide site. Beta-Geometric correction was applied to the datasets to adjust for skewness [93]. Essential genes were identified using a hidden Markov model (HMM), incorporated into TRANSIT [92], as described in more detail previously for M. avium [38]. Virulence genes (comparative analysis between input and output tn libraries; determining statistical differences in sum of tn insertion counts in genes within library selected in vitro versus after infection) were identified using the “resampling” algorithm incorporated into TRANSIT [92].

Macrophage infection and CFU assay

Human induced pluripotent stem cell (iPSC) were obtained from European Bank for induced pluripotent Stem Cells (EBiSC, https://ebisc.org/about/bank), distributed by the European Cell Culture Collection of Public Health England (Department of Health, UK) and produced into monocytes as previously described for our laboratory [94], based on the protocol of Armesilla-Dias et al. [95]. Monocytes were seeded in 96-well plates (40 000/well) and differentiated into macrophages in RPMI 1640 with 10% fetal calf serum (FCS) and 100 ng/mL M-CSF (Prepotech, 300–25). At day five of differentiation, the cells’ medium was changed to RPMI 1640 with 10% FCS (normal iron) or RPMI 1640 with 10% FCS with 100 μM ferric ammonium citrate (FeAC) (high iron). At day 6 of differentiation the cells were infected with the respective Mmar strains at a multiplicity of infection of 4:1 macrophage:bacterium followed by incubation at 33˚C and 5% CO₂. After one hour incubation, the cells were washed once with phosphate-buffered saline (PBS) to remove extracellular bacteria before adding RPMI 1640 with 10% FCS or RPMI 1640 with 10% FCS with FeAC. At day 0 and 1 post infection cells were washed twice with Hanks balanced salt solution before being lysed in 100 μl PBS with 0.1% Triton X-100 (Sigma-Aldrich). The cell lysate was then spotted in a 10-fold dilution series on 7H10 agar plates containing 10% ADC and 0.2% glycerol. The plates were incubated at 30˚C and taken out when CFUs appeared in dilutions with optimal countable range. Macrophage infection with Mtb strains were conducted in a similar manner to Mmar strains except for a few key differences. Mtb-infected macrophages and bacteria-containing agar plates, the latter supplemented with OADC instead of ADC, were kept at 37˚C instead of 30˚C. Mtb proliferates at a slower rate than Mmar, hence, cells were lysed at day 0, 3 and 6 post infection to determine intracellular bacterial growth. Due to the extended time-period of the experiment, half-volume media change was carried out on day 3 post infection.