Adult workers of R. metallica used for transcriptome sequencing and our main venom analyses were collected from a 2-m2 area of lawn (and were assumed to be from a single colony) at The University of Queensland, St Lucia, Queensland, Australia. Venom was collected from 100 R. metallica workers by holding the ant with a pair of forceps while allowing it to sting through a layer of parafilm covering the opening of a 1.5-mL tube (Supplemental Video 1). Ants aggressively and repeatedly stung the parafilm layer presented, depositing, on most occasions, a small drop of clear colorless venom. Collection was continued until no more venom was deposited. Venom from the stings of multiple ants was collected from the bottom of the parafilm layer by rinsing with 10 μL of ultrapure water. All collected venom was pooled then stored at −20°C until further analysis. The total amount of venom collected from 100 individuals, as estimated from A280 measured using a NanoDrop spectrophotometer (Thermo Fisher, Waltham, MA, USA), was 370 μg (in a final volume of 200 μL). To compare the venoms of these ants with those of other nearby colonies of R. metallica, we also used the same method to obtain venom from four additional colonies across the UQ St Lucia campus, collecting eight workers from the entrances from each colony.
Transcriptome sequencing and assembly
Venom apparatuses (venom gland filaments, venom reservoir, and venom duct) were dissected from the ants three days after venom collection (using the same individuals from which venom was collected). Using TRIzol (Life Technologies, Carlsbad, CA, USA), total RNA was extracted from the pooled venom apparatuses of 30 individuals. Complementary DNA library preparation and sequencing was performed by the UQ Institute for Molecular Bioscience Sequencing Facility. A dual-indexed library was constructed with the TruSeq-3 Stranded mRNA Sample Prep Kit (Illumina, San Diego, CA, USA) with oligo (dT) selection and an average insert size of 180 base pairs. Samples were pooled in a batch of 20 samples, and 150-cycle paired-end sequencing was performed on an Illumina NextSeq 500 instrument. Adapter trimming of demultiplexed raw reads was performed using fqtrim v0.9.7 , followed by quality trimming and filtering using prinseq-lite v0.20.4 . Error correction was performed using BBnorm tadpole, part of the BBtools package. Trimmed and error-corrected reads were assembled using Trinity v2.4.0  with a k-mer length of 31 and a minimum k-mer coverage of 2. Assembled transcripts were annotated using a BLASTX  search (E value setting of 1e−3) against the UniRef90 database. Estimates of transcript abundance were performed using the RSEM  plugin of Trinity (align_and_estimate_abundance). Using TransDecoder, Transcripts were translated and filtered to open-reading frames (> 50 amino acid residues). This was used as a search database for ProteinPilot.
A combination of top-down of native and reduced and alkylated venom, and bottom-up proteomics of reduced, alkylated and trypsin-digested venom was used to examine the venom composition of R. metallica. Two aliquots of venom (10 μg each) were dried by vacuum centrifugation. Gas phase reduction and alkylation was performed according to the protocol described by Hale et al. . 100 μL of reduction/alkylation reagent (50% (v/v) ammonium carbonate, 48.75% acetonitrile (ACN), 1% 2-iodoethanol, 0.25% triethylphosphine) was added to the lid of each 1.5 mL tube containing dried venom, which was then inverted, closed, and incubated at 37°C for 90 min. One aliquot of reduced and alkylated venom was then digested by incubating with trypsin (20 ng/μL) overnight at 37°C, according to the manufacturer’s instructions (Sigma-Aldrich, St. Louis, MO, USA).
Three venom samples (10 μg each)—native venom, reduced and alkylated venom, and reduced, alkylated and trypsin-digested venom—were analyzed by LC-MS/MS. Samples were separated on a Nexera uHPLC (Shimadzu, Kyoto, Japan) with a Zorbax stable-bond C18 column (2.1 × 100 mm; particle size, 1.8 μm; pore size, 300 Å; Agilent, Santa Clara, CA, USA), using a flow rate of 180 μL/min and a gradient of 1 to 40% solvent B (90% ACN and 0.1% formic acid [FA]) in 0.1% FA over 25 min, 40 to 80% solvent B over 4 min, and analyzed on an AB Sciex 5600 TripleTOF (SCIEX, Framingham, MA, USA; operated with Analyst TTF v1.8) mass spectrometer equipped with a Turbo-V source heated to 550°C. MS survey scans were acquired at 300 to 1800 mass/charge ratio (m/z) over 250 ms, and the 20 most intense ions with a charge of +2 to +5 and an intensity of at least 120 counts were selected for MS/MS. The unit mass precursor ion inclusion window mass within 0.7 Da and isotopes within 2 Da were excluded from MS/MS, with scans acquired at 80 to 1400 m/z over 100 ms and optimized for high resolution. Using ProteinPilot v5.0 (SCIEX), MS/MS spectra were searched against the translated venom apparatus transcriptome.
Transcripts identified as encoding venom components in the first ProteinPilot search were used to generate a BLAST database, which, in an effort to identify all venom component homologs in the assembled transcriptome, was aligned using BLASTn (e value 1e−6) back to the complete transcriptome. Transcripts encoding venom components and identified homologs were then manually examined using the Map-to-Reference tool of Geneious v10.2.6 , where numerous “masked” homologs were extricated from assembled transcripts and erroneous transcripts discarded. These were then reincorporated back into the complete transcriptome, estimation of transcript abundance repeated, and a second, final ProteinPilot search performed. Peptides identified by ProteinPilot were validated by comparison of experimentally derived MS/MS peaks against a theoretical peak list generated using MS-Product in ProteinProspector v5.22.1 (http://prospector.ucsf.edu/prospector/cgi-bin/msform.cgi?form=msproduct).
We also examined the composition of R. metallica venom using MALDI MS. 1 μL of diluted venom was mixed with 2 μL diluted a-cyano-4-hydroxycinnamic acid (CHCA) solution (stored as an acetone-saturated solution and diluted 1 in 10 with a solution of ethanol:acetone:0.1% TFA (6:3:1) as a working solution) and spotted onto a polished steel target. MALDI spectra were acquired using a Bruker Autoflex Speed MALDI time-of-flight (TOF)/TOF system (Bruker Daltonics Inc, MA, USA; operated with Flex Control v3.4) in linear reflectron mode at 2000 Hz, with a m/z range of 1,000–7,000. Ten spectra of 500 shots each were saved. The group of 10 spectra were loaded into ClinProt Tools v3.0 (Bruker) as a separate group and observed as a gel view, with the averaged spectrum for the entire dataset produced after recalibration of the entire loaded sample cohort. This dataset was then analyzed by PCA and clustering analysis in ClinProt Tools 3.0 using default settings.
Gel electrophoresis and in-gel digestion
R. metallica venom (10 μg) was denatured (70°C, 10 min) in sample buffer and separated on 4-12% Bis-Tris Plus gels (150 V, 22 min; Life Technologies, CA). Gels were stained with Coomassie Brilliant Blue R-250. Four protein bands > 20 kDa were excised and reduced and alkylated, in gel, according to the protocol described by Hale et al. . Excised bands were destained for 1 h with 50% ACN in 40 mM ammonium bicarbonate, incubated for 90 min at 37°C in reduction/alkylation reagent (50% (v/v) ammonium carbonate, 48.75% ACN, 1% 2-iodoethanol, 0.25% triethylphosphine), rinsed with water then dehydrated with 100% ACN. Trypsin (20 ng/L), prepared according to the manufacturer’s instructions (Sigma-Aldrich, St. Louis, MO), was added to the gel pieces and then incubated for 16 h at 37°C. Reactions were quenched with FA (1% final concentration) and the samples were analyzed by LC-MS/MS as described above.
Peptides were produced using Fmoc solid-phase peptide synthesis at 0.1 mmol scale. Protecting groups used were Lys/Trp/His(Boc), Ser/Thr/Tyr(tBu), Asp/Glu(OtBu), Asn/Gln/Cys/His(Trt), and Arg(Pbf). Rm1a, Rm4a, Rm20a, Rm52d, Rm53a, Rm54a and Rm55a/b were assembled on Rink-amide ProTide resin (CEM, Matthews, NC) to produce a C-terminal amide, whereas Rm9a and Rm34a were assembled on preloaded Wang-polystyrene resin (ChemImpex, Wood Dale, IL), Leu- and Asn, respectively to achieve an acid C-terminal. For Rm52d and Rm53a, amino acids were coupled manually for 10 min using 4 eq of 0.5 M O-(1H-6-chlorobenzotriazole-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate and 8 eq of N,N-diisopropylethylamine, and Fmoc was removed with 20% piperidine (2 × 5 min). Rm1a, Rm4a, Rm9a, Rm20a, Rm34a, Rm53c, Rm54a and Rm55a/b were assembled on a CEM Liberty Prime HT24 microwave synthesizer (CEM Corp, Matthews, NC, USA) using N,N′-diisopropylcarbodiimide/oxyma and Fmoc groups were removed with 20% pyrrolidine, as per manufacturers protocols.
Peptides were released from resin by treatment with 95% TFA/2.5% water/2.5% triisopropylsilane for 50 min at 40 °C on a CEM Razor (CEM, Matthews, NC, USA). Rm54a and Rm53c were cleaved at room temperature while stirring for 2 h using the same cleavage cocktail. Peptides were precipitated with 15 mL ice-cold ether, extracted in A/B 50/50 (A: 0.05% TFA, B: 90% ACN, 0.045% TFA) and lyophilized prior to purification. Due to the hydrophobic nature of the peptides, both ether- and aqueous phases were saved after cleavage. Following lyophilization, Rm1a, Rm4a, Rm53c Rm54a, and Rm55b were not found in the aqueous phase and their respective ether-phases were evaporated and dissolved in A/B for purification. Peptides were purified on a Shimadzu Prominence LC-20AT RP-HPLC system equipped with a SPD-20AV UV detector and a FRC-10A fraction collector using a Agilent C18 column (30 × 250 mm; particle size, 5 μm; pore size, 100 Å; Agilent Technologies, CA, USA) at 8 mL/min. Gradients used were 10–70% B over 60 min (Rm34a, Rm52d, and Rm53a), 20–80% B over 60 min (Rm9a and Rm20a) and 40–90% B over 50 min (Rm1a, Rm4a, Rm53c Rm54a, and Rm55a/b). Fractions of interest were lyophilized and purity assessed using ESI MS and analytical RP-HPLC. Rm9a was oxidized at 0.1 mg/mL in 1 M NH4OAc at pH 8.0 over night at room temperature and purified and analyzed as described above.
Insect incapacitation assays
House crickets (A. domesticus; Pisces Live Food, QLD, Australia) (average mass 50 mg), were injected intra-abdominally with 40 nmol/g of each synthetic peptide (in 2 μL water). Crickets were observed for 30 min following injection and compared with negative control crickets injected with 2 μL water. Lethality was assessed at 24 h. For Rm1a, which showed an effect at the highest dose tested (40 nmol/g), a dose-response experiment was performed. To calculate dose-response curves, a four-parameter Hill equation with variable Hill slope was fitted to the data (GraphPad Prism 8.02). All data are expressed as the mean ± standard error of the mean (SEM) and are representative of at least three independent data points. The percentage of crickets incapacitated by a dose of 40 nmol/g, was compared with negative control (water injection) by one-way ANOVA Dunnett’s multiple comparisons test (GraphPad Prism8).
Female fruit flies (D. melanogaster strain Canton-S) were injected four to six days post-emergence. Female flies were used because their larger size makes injection easier. Groups of eight female flies were assayed in triplicate for each dose tested. Injection needles were formed from glass capillary tubes (#3-000-203-G/X, Drummond, Birmingham, AL, USA) using a micropipette puller (Sutter Instrument Co. Model P-97), with the fine tip manually trimmed afterwards using scissors. Needles then were filled with mineral oil, connected to a Nanoliter 2000 microinjector (Kanetec, Bensenville, IL USA) equipped with a foot pedal, and ~1 μL of peptide solution or water aspirated. Each group of flies were immobilized by cooling for approximately 2 min on ice, and then decanted on the top of the ‘injection stage’, a petri dish filled with ice. Under a dissecting microscope, each fly was carefully impaled on the lateral thorax behind the wing, and 50.6 nL of venom peptide or water was injected. After injection, each group of flies were returned to their 5 mL plastic tube at room temperature and their behavior observed. Climbing behavior was assayed at 5 min by tapping the tube five times on the table to dislodge flies to the bottom and counting the number that failed to climb from the bottom of the tube after 5 s.
Calcium imaging of F11 cells
F11 (mouse neuroblastoma × rat dorsal root ganglion (DRG) neuron hybrid) cells were cultured as previously described . Cells were maintained on Ham’s F12 media supplemented with 10 % fetal bovine serum, 100 μM hypoxanthine, 0.4 μM aminopterin, and 16 μM thymidine (Hybri-MaxTM, Sigma Aldrich). 384-well imaging plates (Corning, Lowell, MA, USA) were seeded 48 h prior to imaging resulting in 90 – 95% confluence at imaging. Cells were incubated for 30 min with Calcium 4 assay component A, as per the manufacturer’s instructions (Molecular Devices, Sunnyvale, CA) in physiological salt solution (PSS; composition in mM: 140 NaCl, 11.5 D-glucose, 5.9 KCl, 1.4 MgCl2, 1.2 NaH2PO4, 5 NaHCO3, 1.8 CaCl2, 10 HEPES) at 37°C. Ca2+ responses were measured using a FLIPRTETRA fluorescent plate reader equipped with a CCD camera (Ex: 470–490 nm, Em: 515–575 nM) (Molecular Devices, Sunnyvale, CA). Signals were read every second for 10 s before, and 300 s after, the addition of peptides in PSS supplemented with 0.1% bovine serum albumin. All data represent mean ± SEM of a representative assay in triplicate. Fluorescence changes on peptide addition were compared to addition of negative control solution (0.1% BSA in PSS). The resulting maximum-minimum fluorescence in the 300 s period after peptide addition was recorded as the response. A four-parameter Hill equation (variable slope) was fitted (a two-site fit was used for the Rm4a data) using Graphpad Prism8.
Pain behavior experiments
Male adult (6 weeks old) C57BL/6J mice were used for behavioral experiments. Each peptide (200 pmol) diluted in saline containing 0.1% bovine serum albumin (BSA) was administered in a volume of 20 μL into the hind paw by shallow intraplantar injection. Negative control animals were injected with saline containing 0.1% BSA. Following injection, spontaneous nocifensive behavior events were counted from video recordings by a blinded observer. Nocifensive behavior events were summed over 30 min and compared with negative control (saline injection) by one-way ANOVA Dunnett’s multiple comparisons test (GraphPad Prism8).
Antimicrobial and cytotoxicity assays
Antimicrobial and cytotoxicity assays were performed by CO-ADD (The Community for Antimicrobial Drug Discovery, Institute for Molecular Bioscience, The University of Queensland). Growth inhibition of each bacterial strain was determined by measuring absorbance at 600 nm, whereas growth inhibition of C. albicans was determined by measuring absorbance at 530 nm, and growth inhibition of C. neoformans was determined by measuring the difference in absorbance between 600 and 570 nm, after the addition of resazurin (0.001% final concentration) and incubation at 35°C for an additional 2 h. Growth inhibition of HEK293T cells was determined by measuring fluorescence (excitation 530/10 nm, emission 590/10 nm) after the addition of resazurin (25 μg/mL final concentration) and incubation at 37°C and 5% CO2, for an additional 3 h. Percentage growth inhibition was calculated using negative controls (media only) and positive controls (no peptide). For the hemolysis assays, human whole blood was washed three times with 3 volumes of 0.9% NaCl and then resuspended in 0.9% NaCl to a concentration of 0.5 × 108 cells/mL. Cells were incubated for 1 h at 37°C with or without the peptide. After incubation, plates were centrifuged at 1000 g for 10 min to pellet cells and debris and hemolysis determined by measuring the supernatant absorbance at 405 nm. Minimum inhibitory concentration (MIC), CC50 (concentration at 50% cytotoxicity) and HC50 (concentration at 50% hemolytic activity) values were calculated by curve fitting the inhibition values versus log(concentration) using a sigmoidal dose-response function (variable slope), in Pipeline Pilot’s dose-response component.
In addition to the cytotoxicity assays performed by CO-ADD, we screened the synthetic peptides for toxicity in cell viability assays using HAP1, MM96L, and HEK293T cells. peptides were reconstituted in 100% DMSO then diluted in cell media (HAP1: IMDM (Sigma Aldrich) with 10% FBS and 1% Penicillin/Streptomycin (P/S); MM96L: RPMI (Gibco) with 10% FBS, 1% P/S and 1% GlutaMAX (Gibco); HEK293T: DMEM high glucose (Gibco), with 10% FBS and 1% P/S. Trypsinized cells (35 × 104) were seeded in each well of a 96-well plate. 0.01 to 1000 μM peptide was added to each well after 24 h, and the cells incubated for an additional 74 h. After incubation, the medium was aspirated from each well and 150 μL of fresh medium containing a 0.002% solution of resazurin (Sigma-Aldrich) was added to the wells and incubated for 4 h at 37 °C. The absorbance was measured at 570 nm using a microplate spectrophotometer (FLUOstar Omega, BMG Labtech, Germany). Percentage cell viability at concentrations of 10 and 100 μM peptide were compared with negative control by one-way ANOVA Dunnett’s multiple comparisons test (GraphPad Prism8).
Molecular evolution analyses
R. metallica aculeatoxin peptide sequences were BLAST searched against UniProtKB and aligned with the resulting homologs as well as previously identified aculeatoxins from ants using an iterative refinement method incorporating local pairwise alignment information (L-INS-i) in MAFFT v7.304b . We then used a molecular phylogenetic approach to examine the evolutionary history of the identified aculeatoxin peptides. The most appropriate evolutionary model was identified using ModelFinder  and reconstructed molecular phylogenies by maximum likelihood with IQ-TREE multicore v2.0.6 , estimating branch support values by ultrafast bootstrap using 10,000 replicates .
Due to the extreme divergence between the mature aculeatoxin peptide domains compared to the signal and propeptide domains, we also searched for evidence of recombination within each of the three main clades of R. metallica aculeatoxins. We used a sequence threader to thread nucleotide sequences to amino acid sequence alignments generated for each clade using L-INS-i in MAFFT v7.304b (Fig. 2) , and then looked for evidence of up to two recombination break points (between each domain type) using the Genetic Algorithm for Recombination Detection (GARD)  implemented in Datamonkey . Because this analysis revealed evidence of breakpoints between the pro- and mature peptide domains in all clades, with clades 2 and 3 possibly representing true recombination breakpoints as opposed to differences in evolutionary rates (Additional file 4: Table S4), we performed two phylogenetic analyses based on alignments of either the full coding regions or only the signal and N-terminal propeptide domains. These analyses returned near-identical topologies (Additional file 4: Fig. S5), suggesting any recombination events occur within clades and that the overall topology generated by our phylogenetic analyses is robust. Nevertheless, we used separate nucleotide alignments for each domain type—signal peptide, N-terminal propeptide, and mature peptide domains—to search for evidence of selection in clades 1–3. FEL tests , assuming synonymous rate variation, were used to look for evidence of site-specific pervasive positive and negative selection, and MEME tests  were used to look for signs of site-specific episodic positive selection.
DNA extraction and amplification via PCR
Ten adult workers of R. metallica were collected from the same 2 m2 area of lawn from which ants for venom collection and transcriptomic studies were collected. Each individual was euthanized by cooling in a freezer, the cuticle was removed via dissection in PBS, and the remaining soft tissue was used for TRIzol (Life Technologies, Carlsbad, CA, USA) DNA extraction. Primer-BLAST  was used to design target-specific primers (Additional file 4: Table S5) for the amplification of Rm1a, Rm20a, Rm34a, Rm53a, Rm55a and DPP-4. Primers were synthesized by Sigma-Aldrich (Sigma-Aldrich, St. Louis, MO, USA). For each Polymerase Chain Reaction (PCR) we used ca. 500 ng ant DNA as estimated using a NanoDrop spectrophotometer (Thermo Fisher, Waltham, MA, USA), 0.4 μM primer (forward and reverse), 2 μM dNTP mix (Thermo Fisher, Waltham, MA, USA), 2 μM MgCl2 (Fisher Biotec Australia, Wembley, WA, AUS), 1X reaction buffer (Fisher Biotec Australia, Wembley, WA, AUS), and 1 unit TAQ polymerase (Fisher Biotec Australia, Wembley, WA, AUS). PCR was performed using the following cycling conditions: 94°C for 5 min, followed by 30 cycles of 94°C for 2 min, 57–63°C for 2 min, 74°C for 1–2 min, and a final elongation step at 74°C for 10 min. PCR products were separated and visualized on 1% agarose gels with SYBR-safe DNA stain ((Thermo Fisher, Waltham, MA, USA)). Each well was loaded with 10 μL PCR product, either 2 μL 6X loading dye (Thermo Fisher, Waltham, MA, USA)) or 2 μL 1Kb Plus DNA Ladder (Thermo Fisher, Waltham, MA, USA)), and 2 μL loading dye, and separated at 110 V for 55 min.
All data were plotted and analyzed using Graphpad Prism (v9.0.0). Data were fitted to equations as indicated. Statistical significance was defined as P < 0.05 using tests as indicated. All data are presented as means ± SEM.