Mapping Enzyme Activity on Tissue by Functional Mass Spectrometry Imaging

Abstract Enzymes are central components of most physiological processes, and are consequently implicated in various pathologies. High‐resolution maps of enzyme activity within tissues therefore represent powerful tools for elucidating enzymatic functions in health and disease. Here, we present a novel mass spectrometry imaging (MSI) method for assaying the spatial distribution of enzymatic activity directly from tissue. MSI analysis of tissue sections exposed to phospholipid substrates produced high‐resolution maps of phospholipase activity and specificity, which could subsequently be compared to histological images of the same section. Functional MSI thus represents a new and generalisable method for imaging biological activity in situ.


fMSI sample preparation
The brown forest cobra (N. subfulva; note that this species was until recently known as the forest cobra, N. melanoleuca) was a wild caught specimen of Cameroon origin housed in the herpetarium of the Liverpool School of Tropical Medicine -a UK Home Office approved and inspected experimental animal facility. Three days prior to euthanasian and venom gland dissection, venom was extracted and lyophilized prior to use. One venom gland from N. subfulva was fixed using RCL2 (Alphelys, France), and dehydrated through 70%, 90% and 100% ethanol (vol/vol), cleared with xylene, and embedded in paraffin. Tissue sections were cut and prepared as described previously. [1,2] Briefly, sections were cut at 7 µm thickness from cold paraffin block using a microtome, placed directly onto an indium-tin oxide (ITO) glass slide, heated to 57 °C to attach the section, and the paraffin removed by repetitive xylene washes in order to improve sensitivity. This sample process avoids crosslinking of proteins, does not result in significant delocalisation of proteins [1,2,3] or loss of phospholipase activity (Fig 2; Supporting Information Fig. S5, S6). However, it removes endogenous lipids from the tissue sections, which is desirable when examining the enzymatic depletion of exogenous lipid substrates dosed onto the tissue.
For the application of PC substrates, stock solutions were first prepared by diluting 50 µL of 100 µM PC in 4 mL 50% methanol. A slide with two sections of the snake venom gland was placed into a vibrational vaporization-deposition system (Bruker ImagePrep), one covered with a glass coverslip and the other exposed. The PC stock solution was sprayed onto the tissue for 60 cycles, where each cycle included 2 s spray, 30 s incubation and 20 s drying time. At the completion of the PC application phase, the ImagePrep was thoroughly cleaned with methanol and a CHCA solution (105 mg CHCA, 8 mL acetonitrile, 7 mL water, 30 µL trifluoroacetic acid) was applied using the standard Bruker ImagePrep CHCA application method to the control and PC-exposed snake venom gland sections.

In vitro mass spectrometry-based PLA 2 assays
An aqueous solution of ppPLA 2 (15 µL, 1 unit/mL) was added to 60 µL PC (3 µM in methanol, containing 5 mM ammonium acetate) in a 96-well PCR plate. 10 µL of the resulting solution was immediately sampled by a chip-based nano-electrospray ionization source (TriVersa NanoMate, Advion, Ithaca, NY, USA), which was used to eliminate carryover between isomeric samples, and infused into an LTQ Orbitrap Elite mass spectrometer (Thermo Fisher Scientific, Bremen, Germany) with a gas pressure of 0.5 psi Acquired spectra were analysed using Flex Analysis 3.4 (build 76).
For confirming the enzymatic generation of LPC from PC substrates, PC16:0/18:1 was analysed as above with either milked venom, liquid droplet extraction (water), or laser microdissected (LMD) pieces from venom gland tissue sections, but with and without mixing venom or extract with 1 mM Varespladib (1:1 vol/vol). For micro-dissection of venom gland tissue, 10 µm sections were placed on PET membrane LMD slides (Leica), heated to 57 °C to attach the section, paraffin removed using xylene, and dissection carried out with an LMD-7 laser dissection microscope (Leica).

Isomer purity determination
The relative abundance of regioisomeric impurities in each synthetic PC standard was determined using a combination of collision-and ozone-induced dissociation to identify the fatty acyl moiety esterified at the sn-1 position of the glycerol backbone. [4] To facilitate this experiment, the LTQ Orbitrap Elite mass spectrometer is modified to enable the introduction of ozone into the ion trap via the helium bath gas, as previously described. [5,6]

Liquid Extraction Surface Analysis
To determine if PLA 2 activity could be measured directly from tissue sections, deparaffinised tissue sections were loaded onto the sample stage of a robotically controlled micro-extraction and nESI system (Advion TriVersa NanoMate) operating in LESA mode. Using the robotic dispenser, a 1 µM solution of PC 16:1/16:1 in methanol was applied to a localised region of the tissue. After 5 s, the solvent was then re-aspirated and infused directly into a highresolution MS (Thermo Fisher Scientific LTQ-Orbitrap Elite) via a chip-based nanoESI interface. Spectra were acquired at 120,000 resolving power (at m/z 400). For acquisition of Varespladib-inhibited PLA 2 activity from tissue sections, we performed fMSI using an Atmospheric Pressure MALDI (AP-MALDI UHR, Masstech) coupled to a MicroTOF Q II QqTOf mass spectrometer (Bruker), which we found minimized MALDIinduced LPC formation. Tissue section of N. subfulva were prepared for fMSI as per above, except that we applied 500 µM Varespladib in 50 % methanol (vol/vol) prior to PC substrate deposition. Varespladib solution was applied using the same protocol as during the PC substrate deposition by ImagePrep, with a total of 60 spray cycles, and partial deposition was achieved by covering parts of the section with a cover slip. Following the exposure to inhibitor, a solution of PC 16:0/18:1 (Avanti 850457C-200mg) was applied to the whole tissue using the same protocol as for the inhibitor for 60 cycles. The PC solution was 125 ug/mL in 50% methanol (aq). Following PC deposition, CHCA matrix was applied using ImagePrep as described above.
Acquisition on the AP-MALDI UHR MicroTOF Q II QqTOf was performed using Target ng 8.8.3 to control the AP-MALDI UHR and microTOF control 2.3 (build 40). The AP-MALDI was set to acquire in a vertical line movement, 80 µm pixel size, and 0.5 pixels per second.
The microTOF Q II acquired 360 -1000 Da at 0.5 seconds per scan, collecting only centroid data. Overall the data was acquired at 2 pixels per second. After the acquisition, the xml files with target ng raster information and the raw BAF file from the microTOF Q II were converted to an imzML file using MT imzML convertor (ng) version 1.0.1, which uses the MSconvert feature from Proteowizard 3.0.19228 (64 bit). The resultant imzML file was then imported into SCILS LAB MVS 2019c for visualisation. The normalisation applied was root mean squared (RMS). The PC substrate and LPC product masses were selected manually.

Histology
After fMSI or MSI experiments, CHCA matrix was removed using methanol, and then brought to an aqueous environment stepwise through 75%, 50%, and 25% methanol. The slides were then stained with Hematoxylin and Eosin using standard protocols described elsewhere. [1]

Bottom-up proteomics
To confirm the presence of PLA 2 on tissue and tissue micro-dissections, sections were treated with xylene to remove paraffin before they were either cut into equal parts or micro-dissected using a Leica LMD-7. Tissue was then dissolved in 50 mM ammonium bicarbonate 15 % 6 acetonitrile (vol/vol) pH 8, and cystines reduced by incubation in 5 mM dithiothreitol at 70 °C for 5 min before they were alkylated with 10 mM iodoacetamide at 37 °C for 90 min. The reduced and alkylated samples were then digested by incubating with 30 µg/µL trypsin overnight at 37 °C. The resulting tryptic peptides were desalted using a C18 ZipTip     while their distributions are shown as heat maps below. C) To further confirm that the LPC formation was primarily due to PLA 2 enzyme activity, we used a Leica LMD-7 laser dissection microscope to collect approximately equal halves of contents within the ductules of a subsequent venom gland section (red or blue shapes in micrograph on left), which ensured equal content between the two groups. Each of the two groups of micro-sections were pooled and then incubated with PC 18:1/16:0, with (top) or without (bottom) 500 µM Varespladib. This revealed complete inhibition of LPC product formation by Varespladib, suggesting that incomplete suppression of LPC product formation observed in B) was likely due to problems with inhibitor application rather than an inability by Varespladib to prevent LPC formation. m/z of peaks corresponding to PC substrate and LPC product are indicated in blue and red, respectively, while the colours of the spectra correspond to the colour-coded micro-dissections in C). E) To confirm the presence of PLA 2 in the micro-dissection sample inhibited by Varespladib, we analysed the remaining solution by bottom-up proteomics, which indeed identified known enzymatically active venom PLA 2 from Naja melanoleuca, which is a closely related sister species to N. subfulva. UniprotKB accession numbers are given for each PLA 2 , while amino acid sequence regions corresponding to high-confidence matches in our analysed tryptic digest are shown in bold green. P00599 is Basic phospholipase A 2 , P00600 is Acidic phospholipase A 2 DE-II, and P00601 is Acidic phospholipase A 2 DE-III.
13 Figure S7. Venom PLA 2 is found along the whole length of the venom gland of N. subfulva.
A section of N. subfulva venom gland (below) was dissected into four parts, soluble protein dissolved in water, and analysed by shotgun LC-MS/MS. The graph (top) shows the number of unique trypsin-digested peptide fragments confidently assigned by Protein Pilot (>95 %) to three PLA 2 characterised from the venom of the closely related N. melanoleuca, normalized to the total number of high-confidence peptides assigned to all proteins from Naja spp. The absolute number of high-confidence peptides is shown in colours corresponding to each PLA 2 . Orange indicates basic PLA 2 (UniProt accession P00599), grey and blue indicate acidic PLA 2 's (UniProt accessions P00600 and P00601, respectively).