Dataset - Exaptation of an evolutionary constraint enables behavioural control over the composition of secreted venom in a giant centipede

Mass Spectrometry Imaging and Serial Block-Face Electron Microscopy raw data of centipede (Scolopendra morsitans) venom glands. Comparison of venom glands by mass spectrometry imaging (MSI): Complementing our analyses of secreted venom, we also examined the effect of secretion on the relative abundance and distribution of venom components within full three-dimensionally (3D) reconstructed venom glands imaged by mass spectrometry imaging. Venom glands were dissected from S. morsitans after four different experimental treatments: after a starving period of five weeks to ensure that the venom gland was fully replenished (“full”); after venom was collected using electrostimulation (“electro”); after venom was collected using the defensive milking technique described above (“defensive”); or after envenoming a cricket (“predatory”). Dissected venom glands were fixed in RCL2, dehydrated in a graded ethanol series and embedded in paraffin as described previously9,41. Importantly, this sample preparation protocol depletes lipids and other small molecules, making it likely that the species observed are peptides or small proteins. Each gland was sectioned at 7 µm, transferred to microscope slides and washed with xylene to remove paraffin. The sections were then sprayed with CHCA (0.7 mg/mL in 50% ACN [vol/vol] and 0.2 % TFA [vol/vol]), using a Bruker ImagePrep automated matrix sprayer. Samples were analysed using a Bruker Autoflex Speed MALDI-TOF/TOF MS with Flex Imaging v4.1 and Flex Control v3.4 software. Spectra were acquired in linear positive mode over a mass range of 1000–20000 m/z using a medium laser spot size for a spatial resolution of 50 µm, and acquiring 600 shots at 2000 Hz. The data was analyzed, normalised, aligned, visualised, and 3D reconstructed using SCiLS Lab MVS premium (Bruker Daltonics) In total, seven venom glands were used to create three-dimensional reconstructions of MSI data. In addition, we collected MSI spectra from at least 4 glands and 100 slices per treatment (full: 8 glands, 166 slices, 152079 spectra, area 380.197 mm2; electro: 6 glands, 104 slices, 84115 spectra, area 210.287 mm2; defensive: 4 glands, 155 slices, 87081 spectra, area 217.702 mm2; predatory: 4 glands, 112 slices, 557701 spectra; all regions combined: 380976 spectra, area 952.44 mm2) and used these to generate venom gland-wide mean spectra (with one standard deviation) in SCiLS Lab MVS premium. The in SCiLS Lab integrated pLSA component analysis13 was used to look for distribution patterns. Serial Block-Face Electron Microscopy: For serial block-face EM (SBEM), samples were fixed overnight at 4°C in a solution containing 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer. The next day, samples were postfixed in a solution comprising 2% osmium tetroxide and 1% potassium ferricyanide in 0.1 M cacodylate buffer. Samples were then microwave processed (BioWave, Pelco, Clovis, USA) using a 2 min on–2 min off–2 min on cycle repeated twice at 80W under vacuum. Following this, samples were washed three times in ddH20; once on the bench and twice in the microwave at two cycles of 40 seconds at 80W under vacuum. Sequentially, samples were then immersed in 1% thiocarbohydrazide in ddH20, 2% osmium tetroxide in ddH20, 1% uranyl acetate in ddH20, and then in lead aspartate solution following the same microwave incubation and rinse cycle as described above, except for uranyl acetate incubation, during which the microwave temperature was raised from 22°C to 45°C. The lead aspartate solution was made by adding 0.066 g lead nitrate to 10 ml 0.03 M aspartic acid then adjusting pH to 5 by adding 1 N KOH. Subsequently, samples were dehydrated in ethanol at increasing concentrations (50%, 60%, 70%, 80%, 90%, 100%, 100%), employing microwave processing in cycles of 1 minute on, 1 minute off, and 1 minute on at 150W, without vacuum for each concentration. Samples were then embedded in Durcupan resin (Sigma Aldrich) at increasing concentrations (1:3, 1:2, 1:1, 2:1, 3:1, 100%, 100%, 100%) using a microwave cycle of 3 minutes at 150W with vacuum application for each concentration and left to polymerise in a 60°C oven for 48 hours. Following polymerisation, hardened samples were trimmed and polished using an ultramicrotome (Leica UC6) and mounted on a SBEM stub using conductive epoxy. A 50 µm x 100 µm x 65 µm image volume comprising several venom gland subunits was acquired using a VolumeScope SBEM (Thermo Fisher). During acquisition, the sample was cut at 50 nm thickness and imaged in low vacuum (10 Pa) at a pixel scale of 9 nm x 9 nm with a landing beam energy of 2 kV, 0.1 nA current, and pixel dwell time of 2 µs. The resulting image stack was then contrast adjusted, denoised, and aligned using the TrakEM2 plugin in FIJI47,48. To reconstruct the venom gland subunit, the aligned image stack was down-sampled to 1 µm x 1 µm x 1 µm voxel size using Amira (2020.2, ThermoFischer Scientific) and manually segmented using the segmentation editor. The resulting labels were then converted into surface meshes and visualised in different transparencies. To reconstruct neurons, annotate synapses, and generate connectivity graphs, we used CATMAID, an open source and web-based software49. Synapses were identified by the presence of a presynaptic cloud of vesicles adjacent to a synaptic cleft (Fig. 4H-I). Only connections that contained greater than a single synapse were included in our connectivity analysis (SI Fig. 25C-D).