Investigation of Microbial Cooperation via Imaging Mass Spectrometry Analysis of Bacterial Colonies Grown on Agar and in Tissue During Infection

Summary

A novel sample preparation method is demonstrated for the analysis of agar-based, bacterial macrocolonies via matrix-assisted laser desorption/ionization imaging mass spectrometry.

Abstract

Understanding the metabolic consequences of microbial interactions that occur during infection presents a unique challenge to the field of biomedical imaging. Matrix-assisted laser desorption/ionization (MALDI) imaging mass spectrometry represents a label-free, in situ imaging modality capable of generating spatial maps for a wide variety of metabolites. While thinly sectioned tissue samples are now routinely analyzed via this technology, imaging mass spectrometry analyses of non-traditional substrates, such as bacterial colonies commonly grown on agar in microbiology research, remain challenging due to the high water content and uneven topography of these samples. This paper demonstrates a sample preparation workflow to allow for imaging mass spectrometry analyses of these sample types. This process is exemplified using bacterial co-culture macrocolonies of two gastrointestinal pathogens: Clostridioides difficile and Enterococcus faecalis. Studying microbial interactions in this well-defined agar environment is also shown to complement tissue studies aimed at understanding microbial metabolic cooperation between these two pathogenic organisms in mouse models of infection. Imaging mass spectrometry analyses of the amino acid metabolites arginine and ornithine are presented as representative data. This method is broadly applicable to other analytes, microbial pathogens or diseases, and tissue types where a spatial measure of cellular or tissue biochemistry is desired.

Introduction

The human microbiome is a highly dynamic ecosystem involving molecular interactions of bacteria, viruses, archaea, and other microbial eukaryotes. While microbial relationships have been intensely studied in recent years, much remains to be understood about microbial processes at the chemical level^1,2. This is in part due to the unavailability of tools capable of accurately measuring complex microbial environments. Advances in the field of imaging mass spectrometry (IMS) over the past decade have enabled in situ and label-free spatial mapping of many metabolites, lipids, and proteins in biological substrates^3,4. Matrix-assisted laser desorption/ionization (MALDI) has emerged as the most common ionization technique used in imaging mass spectrometry, involving the use of a UV laser to ablate material from the surface of a thin tissue section for measurement by mass spectrometry⁴. This process is facilitated by the application of a chemical matrix applied homogeneously to the surface of the sample, allowing for sequential measurements to be made in a raster pattern across the sample surface. Heat maps of analyte ion intensities are then generated following data acquisition. Recent advances in ionization sources and sampling techniques have enabled the analysis of non-traditional substrates such as bacterial⁵ and mammalian^6,7,8 cellular specimens grown on nutrient agar. The molecular spatial information afforded by IMS can provide unique insight into the biochemical communication of microbe-microbe and host-microbe interactions during infection^{9,10,11,12,13,14}.

Upon Clostridioides difficile infection (CDI), C. difficile is exposed to a rapidly changing microbial environment in the gastrointestinal tract, where polymicrobial interactions are likely to impact infection outcomes^15,16. Surprisingly, little is known about the molecular mechanisms of interactions between C. difficile and resident microbiota during infection. For example, enterococci are a class of opportunistic commensal pathogens in the gut-microbiome and have been associated with increased susceptibility to and severity of CDI^17,18,19,20. However, little is known about the molecular mechanisms of the interactions between these pathogens. To visualize small-molecule communication between these members of the gut microbiome, bacterial macrocolonies were grown herein on agar to simulate microbe-microbe interactions and bacterial biofilm formation in a controlled environment. However, obtaining representative metabolic distributions upon MALDI imaging mass spectrometry analysis of bacterial culture specimens is challenging due to the high water content and uneven surface topography of these samples. This is largely caused by the highly hydrophilic nature of agar and the non-uniform agar surface response during moisture removal.

The high water content of agar can also make it challenging to achieve homogeneous MALDI matrix coating and can interfere with subsequent MALDI analysis performed in vacuo²¹. For example, many MALDI sources operate at pressures of 0.1-10 Torr, which is a sufficient vacuum to remove moisture from the agar and can cause deformation of the sample. These morphological changes in the agar induced by the vacuum environment cause bubbling and cracking in the dried agar material. These artifacts reduce the adherence of the agar to the slide and can cause dismounting or flaking of the sample into the instrument vacuum system. The thickness of the agar samples can be up to 5 mm off the slide, which can create insufficient clearance from ion optics inside the instrument, causing contamination and/or damage to instrument ion optics. These cumulative effects can result in reductions of ion signal reflective of the surface topography, rather than the underlying microbial biochemical interactions. Agar samples must be homogeneously dried and strongly adhered to a microscope slide prior to analysis in vacuo.

This paper demonstrates a sample preparation workflow for the controlled drying of bacterial culture macrocolonies grown on agar media. This multi-step, slower drying process (relative to those previously reported) ensures that the agar will dehydrate uniformly while minimizing the effects of bubbling or cracking of agar samples mounted on microscope slides. By using this gradual drying method, samples are strongly adhered to the microscope slide and amenable for subsequent matrix application and MALDI analysis. This is exemplified using model bacterial colonies of C. difficile grown on agar and murine tissue models harboring CDI with and without the presence of commensal and opportunistic pathogen, Enterococcus faecalis. MALDI imaging mass spectrometry analyses of both bacterial and tissue models allow for the spatial mapping of amino acid metabolite profiles, providing novel insight into bioenergetic microbial metabolism and communication.

Protocol

NOTE: Animal experiments were approved by the Animal Care and Use Committees of the Children's Hospital of Philadelphia and the University of Pennsylvania Perelman School of Medicine (protocols IAC 18-001316 and 806279).

CAUTION: Clostridium difficile (C. difficile) and Enterococcus faecalis (E. faecalis) are BSLII pathogens and should be handled with extreme caution. Utilize proper decontamination protocols when necessary.

1. Growth of bacterial culture macrocolonies and preparation for overnight shipment

1. Prepare C. difficile and E. faecalis overnight cultures and grow them individually at 37 °C in an anaerobic chamber (85% nitrogen, 10% hydrogen, 5% carbon dioxide) in brain-heart-infusion broth supplemented with 0.5% yeast extract and 0.1% l-cysteine (BHIS). Use 1.5% agar for all plated samples.
2. Normalize the bacterial culture media to optical density at 600 nm (OD₆₀₀). Plate 5 µL of each macrocolony onto BHIS + l-cysteine agar plates and grow anaerobically for 7 days at 37 °C. For mixed species macrocolonies, mix the bacterial culture media at a 1:1 ratio prior to plating.
3. Prepare indium tin oxide (ITO)-coated microscope slides by using an ohmmeter to identify the side of the slide with the conductive coating. Use a diamond-tipped scribe to etch and label the ITO-coated side of the microscope slide.
4. Excise the entire culture from the agar growth plate and then mount onto an ITO-coated microscope slide while ensuring that air bubbles are not trapped between the agar and slide.
   NOTE: The water content in the agar media should allow the culture to adhere naturally to the microscope slide surface. A video exemplifying this process is provided in the supplementary documentation of Yang et al.²².
5. Place the colonies in microscope slide boxes for protection. Store the slide boxes in 8 in x 8 in biohazard specimen transport bags with a handful of desiccant pellets and seal for overnight shipping for further processing and analysis.
   ​NOTE: It is important to maintain a dry environment for the bacterial cultures when shipped under ambient conditions to slow bacterial growth and metabolism and maintain fixation of the bacterial specimens until analysis. This shipping method has been optimized through extensive experiments and is preferred over shipping the colonies frozen in dry ice. Major temperature changes prior to drying tend to cause dismounting and bubbling underneath the mounted agar sample.

2. Ambient drying of  C. difficile + E. faecalis bacterial macrocolonies

1. Remove the agarose bacterial colonies mounted on ITO-coated microscope slides from the packaging material. Place the samples in a dry box with desiccant for 48-72 h at room temperature.
   NOTE: This is a mild and slow drying process that minimizes bubbling, cracking, or detachment of the agar media from the microscope slide surface.
2. Properly dispose of all contaminated packaging material and decontaminate the workspace with an appropriate bactericidal/sporicidal disinfectant.
3. Visually inspect the colonies for deformations in the agar surface (e.g., bubbling, cracking, dismounting).
   ​NOTE: The height of the agarose surface should visibly decrease and lie flat across the slide.

3. Vacuum and heat-mediated drying of  C. difficile + E. faecalis bacterial macrocolonies

NOTE: A custom-built vacuum drying apparatus (Figure 1) was built to facilitate the removal of excess moisture from the agar samples. This apparatus utilizes a rotary vane vacuum pump connected in line to a HEPA biofilter, a cold trap, and a stainless-steel chamber, where the bacterial samples are placed. A variable voltage transformer is connected to an insulated wire filament, which allows the user to heat the chamber to expedite the drying process.

1. Close the vacuum line to the sample chamber and turn on the power switch to the rotary vane pump to allow the vacuum pump to warm up and achieve proper vacuum pressure.
2. Turn on the variable voltage transformer to heat up the wire filament wrapped around the chamber. Adjust the variable power supply until the internal temperature of the vacuum chamber reaches ~50 °C. While the pump is warming up, place a slurry of dry ice and 100% ethanol into the condenser of the cold trap.
   NOTE: The cold trap condenses any vapors or spores from the sample and avoids contamination of the rotary vane pump system and vacuum pump oil.
3. Open the vacuum chamber by using a wrench to loosen the 16 mm double claw flange clamps. Insert the agarose samples into the chamber and seal the chamber tightly with the 16 mm double claw flange clamps.
4. Open the pump valve to evacuate the chamber. Allow the samples to dry for 60-120 min (e.g., at ~150 mTorr).
   NOTE: This drying time is sufficient to remove most of the moisture in small agar sections. Empirical determination of the optimal drying time for moisture removal and decreasing the agar height may be necessary depending on the individual setup and samples. Substantially longer drying times can cause the dried agar to become brittle and prone to cracking.
5. When complete, slowly vent the vacuum chamber to ambient pressure by closing the valve on the rotary vane pump and opening the external valve to ambient air. Open the chamber using the previously mentioned protocol and remove the dried agarose sample from the chamber.
6. Store the sample in a dry box with desiccant until matrix application.
   ​NOTE: Figure 2 shows images of the agar surface prior to and after drying.

4. MALDI matrix application  via robotic spraying

NOTE: After the agarose samples have been thoroughly dried and the height of the culture sections have noticeably decreased, use a robotic matrix sprayer to homogeneously apply a thin coating of a chemical MALDI matrix compound. This procedure should be performed in a chemical fume hood and with proper personal protective equipment, including gloves, laboratory glasses, and a lab coat.

1. Select the appropriate MALDI matrix. To follow this protocol, use 1,5-diaminonaphthalene (DAN) MALDI matrix for its favorable desorption and ionization of amino acids in negative ion mode.
2. Prepare 10 mL of a 10 mg/mL DAN MALDI matrix solution in 90/10 (v/v) acetonitrile/water. Use HPLC-grade solvents, sonicate for 30 min, and filter the solution through 0.2 µm nylon syringe filters prior to introduction to the robotic sprayer syringe pump. Additionally, prepare wash solutions to ensure the sprayer line is clean between each use.
   NOTE: Wash solutions are chosen to increase the solubility of the matrix and other contaminants in the sprayer line and facilitate their removal from the system. The wash solutions used herein are 90/10 (v/v) acetonitrile/water, 50/50 (v/v) water/methanol, 99/1 (v/v) acetonitrile/acetic acid, and 95/5 (v/v) water/ammonium hydroxide.
3. Attach the sample to the sprayer tray and load the prepared solutions into the sprayer line (Figure 3). Using the computer software, specify the necessary parameters to allow for a uniform coating of the matrix compound: 30 °C nozzle temperature, eight passes, 0.1 mL/min flowrate, CC pattern, 0 s drying time, 10 psi.
   NOTE: It is generally accepted that most pathogens will be inactivated by application of the MALDI matrix, which is typically a small organic acid or base.
4. After the spraying sequence has finished, remove the sample from the sprayer tray and store in a desiccation cabinet until analysis.

5. Preparing bacterial macrocolonies for MALDI imaging mass spectrometry data acquisition

NOTE: All imaging mass spectrometry analyses were performed using a Fourier transform ion cyclotron resonance (FTICR) mass spectrometer. This instrument is equipped with a Nd:YAG MALDI laser system (2 kHz, 355 nm) .

1. Insert the matrix-coated sample into the MALDI target plate microscope slide adapter and scribe at least three fiducial markers that encompass the sample area using permanent markers (Figure 4). Use a flatbed scanner to acquire an optical image of the microscope slide including the fiducial markers.
   NOTE: The fiducial markers will allow the registration of the optical image to the MALDI camera view observed in the instrument.
2. Define an MS instrument method optimized for the desired mass range, sensitivity, and resolution, which includes mass calibration, MALDI laser settings, ion optics parameters, and ICR cell conditions. For this method, select a 100 Da mass window from m/z 100 to 200 for gas-phase signal enrichment in negative ion mode using a continuous accumulation of selected ions (CASI) approach,²³ which encompasses the m/z values of the metabolites of interest.
3. Open the instrument's image acquisition software and use the setup wizard to define the file name and location, the MS acquisition method, regions of interest to be sampled, and the spatial resolution of the image.
   NOTE: Spatial resolutions of 100-300 µm are typically employed for bacterial macrocolony imaging.
4. Once all the parameters have been defined, start the acquisition sequence to serially acquire a mass spectrum at each pixel across the defined region(s) of interest.
   ​NOTE: Image acquisition time depends on the instrument settings, but typically ranges from 2 to 6 h for images containing 5,000-10,000 pixels.

6. Imaging mass spectrometry data analysis and compound identification

1. Following image acquisition, save the data with a ".mis" file extension, which is a vendor-specific file format for imaging mass spectrometry platforms. Open the data file in flexImaging or SCiLS software packages or export to a non-proprietary data format such as .mzml and visualize using vendor-neutral software (e.g., Cardinal²⁴ or MSIreader^25,26).
   NOTE: An average mass spectrum is displayed upon opening the imaging data in flexImaging, which represents the average intensities of all ions detected in the sampled regions. Peaks of interest can be tentatively identified based on accurate mass measurements. Typically, mass accuracies of better than 5 parts per million (ppm) are sufficient for metabolite identification on FTICR mass spectrometers.
2. Using the referenced software, define mass windows for peaks of interest to generate false-color heat maps of ion distributions across the sampled regions.
   1. On the average mass spectrum display, zoom in to the peak of interest and select an appropriate mass window that encompasses the area of the peak profile.
   2. Right click on the highlighted mass window and select Add Mass Filter.... Label the selected m/z value using the tentative identification for the suspected metabolite as determined by the high resolution accurate mass measurement.
   3. Select the intensity threshold to adjust the dynamic range of the analyte image displayed by the false-color heat map. Further adjust the mass filtering parameters so that the mass window encompasses the area of the peak profile, and then add the mass filter.
      ​NOTE: Intensity normalization can be performed as appropriate to improve relative quantification across measured regions. The experiments herein utilized CASI data acquisition (vide infra), which may render total ion count (TIC) and root mean square (RMS) normalization methods inaccurate. As such, all ion images shown herein are displayed without normalization.

7. Preparation and shipment of noninfected control and C. difficile -infected mouse cecal tissues

1. Innoculate 4-8-week-old C57BL/6 male mice with antibiotics (0.5 mg/mL cefoperazone or 0.5 mg/mL cefoperazone + 1 mg/mL vancomycin) in drinking water ad libitum for 5 days, followed by a 2 day recovery period and subsequent infection.
2. Euthanize the animals by CO₂ asphixiation and harvest the mouse cecum organ immediately. Embed in a 20% mixture of optimal cutting temperature (OCT) compound in distilled water.
3. Place the samples on dry ice and then package and ship for analysis; store at -80 °C until analysis.

8. Cryosectioning of noninfected control verus  C. difficile -infected mouse cecal tissues

1. Clean all cryosectioning equipment by rinsing with 100% ethanol and allow to dry before placing samples into the cryostat chamber. Prepare ITO-coated microscope slides by using an ohmmeter to identify the side of the slide with the conductive coating. Use a diamond-tipped scribe to etch and label the ITO-coated side of the microscope slide.
   NOTE: Proper personal protective equipment, including gloves, laboratory glasses, a lab coat, and cryostat sleeves, should be worn.
2. Perform cryosectioning on a research cryomicrotome. Transfer the OCT-embedded mouse ceca tissues from a -80 °C freezer to the cryomicrotome chamber (30 °C chamber temperature, -28 °C object temperature) on dry ice. Mount the tissue samples to be compared using imaging mass spectrometry (e.g., noninfected control vs. C. difficile-infected) on the same microscope slide to ensure identical sample preparation of both tissue types and enable accurate metabolite comparisons.
3. Inside the cryomicrotome chamber, mount the OCT-embedded tissue on a sample chuck using additional OCT solution. After the OCT solution has solidified, fix the chuck to the specimen head. Begin cryosectioning at 10-50 µm increments to reach the desired tissue depth/plane of the organ.
4. Once an optimal cross section of the tissue is reached, begin collecting sections at 12 µm thickness. Gently manipulate the slice using artist paintbrushes and place it onto a Teflon-coated microscope slide.
5. Roll the ITO-coated microscope slide on top of the tissue section to pick up the tissue from the Teflon-coated slide. Thaw mount the tissue section to the microscope slide by pressing the palm to the back of the ITO-coated slide. Continue to thaw mount until the tissue section transitions from translucent to an opaque/matte texture.
6. Transfer the tissue-mounted microscope slides on dry ice to a dry box with desiccant for storage for same day analysis, or store at -80 °C for longer-term storage.

9. Preparation of noninfected control versus C. difficile -infected mouse cecal tissues for matrix application and MALDI imaging mass spectrometry

1. Perform MALDI matrix application using the same protocol described in step 4 (30 ° C nozzle temperature, eight passes, 0.1 mL/min flowrate, CC pattern, 0 s drying time, 10 psi).
2. Perform MALDI imaging mass spectrometry using the same protocols described in steps 5 and 6.
3. Analyze the ion images from multiple biological replicates and compare the results for significance via intensity box plot comparisons using the SCiLS statistical software referenced above.
   NOTE: The appropriate statistical tests and comparisons will depend on the application and can include spatial segmentation, classification models, and comparative analysis for determining discriminative and correlated spectral features. Comparitive analyses can include assigning p-values to significant features, generating principal component analyses for tissue comparisons, and visualizing box plots to identify variations. It is also useful to visualize the tissue using brightfield microscopy of the tissue section stained via hematoxylin and eosin (H&E) following imaging mass spectrometry. This enables clear identification of morphological features in the tissue and can be analyzed in consultation with a trained pathologist.

Representative Results

We have performed metabolite MALDI imaging mass spectrometry of model bacterial colonies and mice co-colonized with E. faecalis and C. difficile to study the role of amino acids in microbe-microbe interactions. Bacterial macrocolonies grown on agar serve as a well-defined model to analyze distinct biochemical changes in bacterial biofilm formation. It is important to ensure a controlled drying process for bacterial culture macrocolonies grown on agar media to minimize deformations and cracking in the agar surface. This was achieved via a two-step process, involving preliminary drying with desiccant under ambient temperature and pressure, followed by a more rapid vacuum-assisted drying step performed at an elevated temperature. A custom-built drying apparatus was used to facilitate agar dehydration while capturing the release of any bacterial contaminants or spores from the samples (Figure 1). The additional vacuum drying step is necessary to ensure that samples are completely dehydrated and will not undergo deformation upon exposure to the MS vacuum.

Agar sections were initially 2-4 mm in sample height (Figure 2A) and were reduced to ~0.5 mm in height following the drying protocol (Figure 2B). It is important to ensure uniform drying across the sample surface during this sample preparation protocol. Large variations in sample height will result in a defocused MALDI laser beam at the sample surface, which can detrimentally affect ionization efficiency and resulting analyte ion intensity. An example of an unsuccessful drying process is shown in Figure 2D, where the agar surface has bubbled and cracked, rendering the sample unusable for further analysis. Once the bacterial sample has been thoroughly dried, a DAN MALDI matrix layer is applied using a robotic sprayer (Figure 3). This instrument allows for the precise control of spraying parameters, enabling application of a uniform matrix coating (Figure 2C).

Following matrix application, the microscope slides containing the samples are inserted into a slide adapter for analysis on an FTICR MS (Figure 4). Coordinates on the slide are registered between the imaging mass spectrometry acquisition software and the instrument camera by first drawing fiducial marks on the slide around the regions to be measured (denoted by "+" symbols in Figure 4). The slide is then digitally scanned using a flatbed scanner to record an optical image, which is then used to define the acquisition sequence in the flexImaging software. We optimized the mass spectrometry instrument settings for the transmission and detection of metabolite anions. Using CASI gas-phase ion enrichment, an ion isolation window was defined to selectively accumulate ions at m/z 150 ± 50 Da (Figure 5A), which contains an array of compounds observed from the tissue surface. Specifically here for C. difficile infection, we found amino acid pathways to demonstrate interesting and changing roles during bacterial catabolism. Based on accurate mass measurements, the ion signals were identified at m/z 131.083 and m/z 173.104 to be ornithine (0.34 ppm mass error; Figure 5B) and arginine (0.43 ppm mass error; Figure 5C), respectively.

Imaging mass spectrometry of bacterial macrocolonies was performed on C. difficile monocultures and on C. difficile + E. faecalis co-culture colonies (Figure 6). An optical image of the samples following MALDI matrix application highlights the acquisition regions of interest, which extend to the outer areas of colony biofilms (Figure 6A,C). Imaging mass spectrometry of these samples allows for the spatial mapping of both ornithine and arginine amino acid metabolites, which are found to be altered and differentially localized in the presence of E. faecalis (Figure 6B,D). For example, arginine is significantly more abundant in the C. difficile monoculture compared to the C. difficile + E. faecalis co-culture (Figure 6D). We have also extended the imaging mass spectrometry analyses to mouse models of infection using 8-week-old germ-free C57BL/6 female mice that were infected with either C. difficile spores, both C. difficile spores + E. faecalis cells (wt) (5 × 10⁸/mL), or a transposon mutant containing both C. difficile spores + E. faecalis (ArcD mutant) cells (5 × 10⁸/mL) (Figure 7)²⁷. The mouse cecum was harvested and frozen in 25% OCT compound to maintain organ shape and fidelity. The OCT compound also helped support the thin tissue sections during cryosectioning. Similar to the protocol mentioned above, tissue sections are coated with a DAN MALDI matrix layer, scanned using a flatbed optical scanner (Figure 7B), and analyzed by MALDI imaging mass spectrometry (Figure 7C). Histological staining was performed on tissue sections following imaging mass spectrometry analysis, and high-resolution brightfield optical scanning was employed to evaluate tissue morphology (Figure 7A). The epithelial cell layer is clearly inflamed during infection compared to the control tissue.

Figure 1: Home-built vacuum drying apparatus. A custom-built vacuum drying apparatus is used to expedite moisture removal from agar samples. Samples are placed into the vacuum chamber, which is sealed and evacuated using a rotary vane pump, and then heated using a variable voltage transformer. Any contaminant vapors/spores from the sample are trapped in the cold trap and HEPA in-line filters. Pressures of 100-150 mTorr are typically employed. Please click here to view a larger version of this figure.

Figure 2: Drying and preparation of bacterial macrocolonies. Optical images of agar samples (A) prior to and (B) after moisture removal using desiccation and vacuum-assisted drying. (C) A 1,5-diaminonaphthalene MALDI matrix is applied to the bacterial co-culture following drying using a robotic sprayer. (D) Unsuccessful drying typically results in cracking and bubbling within the agar, rendering it unusable for further analysis. Abbreviations: DAN = 1,5-diaminonaphthalene; MALDI = matrix-assisted laser desorption/ionization. Please click here to view a larger version of this figure.

Figure 3: Robotic spraying of MALDI matrix on agarose samples. Images of (A) an HTX M5 TM sprayer and (B) agarose samples loaded into the robotic matrix sprayer for MALDI matrix application. Abbreviation: MALDI = matrix-assisted laser desorption/ionization. Please click here to view a larger version of this figure.

Figure 4: Mass spectrometer adapter with loaded samples for analysis. Matrix-coated agarose samples are loaded into an MTP slide adapter for MALDI analysis. Fiducial markers are visible as black plus signs ("+"). Abbreviation: MALDI = matrix-assisted laser desorption/ionization. Please click here to view a larger version of this figure.

Figure 5: Representative mass spectra of identified metabolite ion peaks. (A) Negative ion mode MALDI analysis of bacterial culture samples allows for the detection of dozens of metabolites. High-resolution accurate mass measurements allow for the tentative identification of (B) ornithine at m/z 131.083 (0.34 ppm mass error) and (C) arginine at m/z 173.104 (0.43 ppm mass error). Mass spectra is acquired using a mass resolving power (FWHM) of 75,000 at m/z 150. Abbreviations: MALDI = matrix-assisted laser desorption/ionization; FWHM = full width at half maximum. Please click here to view a larger version of this figure.

Figure 6: Imaging mass spectrometry analysis of bacterial macrocolonies. (A,C) Optical images of agarose samples shows the measurement regions for imaging mass spectrometry analysis, which are highlighted by white dashed lines. (B,D) MALDI ion images for ornithine (m/z 131.083, 0.038 ppm) and arginine (m/z 173.104, 0.60 ppm) display unique spatial distributions across the bacterial cultures. Note that the absolute intensity scales are different for the co-culture ion images shown in panels B and D. For example, arginine is significantly more abundant in the C. difficile monoculture relative to the co-culture in panels C and D, meaning this intensity scale is relative to arginine's abundance in the monoculture, and it appears that there is no arginine present in this co-culture. Conversely, the co-culture in A and B projects arginine's intensity onto a lower absolute intensity scale, since this is the only sample in the image. Abbreviations: MALDI = matrix-assisted laser desorption/ionization; CASI = continuous accumulation of selected ions. Please click here to view a larger version of this figure.

Figure 7: Imaging mass spectrometry analysis of infected murine cecal tissues. (A) Brightfield microscopy images of hematoxylin and eosin-stained tissues from infected mouse ceca allow for the visualization of tissue morphology. (B) Optical images of mouse ceca tissues show the samples following application of a DAN MALDI matrix layer. (C) MALDI ion images for ornithine (m/z 131.083, 0.88 ppm) and arginine (m/z 173.104, 0.15 ppm) display unique spatial distributions across the tissue samples. Abbreviations: MALDI = matrix-assisted laser desorption/ionization; DAN = 1,5-diaminonaphthalene. Please click here to view a larger version of this figure.

Discussion

During MALDI imaging mass spectrometry, it is important to have a flat sample surface to provide for a consistent focal diameter of the incident MALDI laser on the sample substrate. Deviations in sample height can cause the MALDI laser beam to shift out of focus, causing alterations in beam diameter and intensity, which can affect MALDI ionization efficiency. These alterations in ionization efficiency can result in differences in analyte intensity across the tissue surface that are not reflective of the underlying tissue biochemistry, but instead reflective of the surface topography. Additionally, most MALDI ionization sources operate at sub-atmospheric pressures, and hydrated agar samples are fragile samples when subjected to in vacuo conditions. The vacuum of these sources can cause dehydration of agarose samples, which can alter the sample surface during drying. Samples containing air pockets or large amounts of water are highly susceptible to these changes. This issue is commonly encountered during MALDI imaging mass spectrometry analysis of bacterial specimens grown on agar. Nutrient agar has a high water content to facilitate bacterial growth; however, it is essential to remove as much moisture as possible before applying a MALDI matrix coating and introducing the sample to the vacuum of the mass spectrometer.

Hydrated agar samples are therefore not conducive for in vacuo analysis. In addition to morphological changes induced by in vacuo conditions, there are also potential issues that exist with sample heights that protrude too far from the MALDI target plate slide adapter, which is used to hold the microscope slide mounted sample and bring it to the proper position for MALDI laser focal point. The MALDI stage mechanism is tightly inferfaced with other mechanical components of the stage, and while in position, there is a clearance of only a few millimeters between the surface of the sample target plate and the front end of the ion source optics. This can cause scraping of the sample, which can move or completely dismount the sample to be analyzed. Moreover, while hydrated agar specimens adhere naturally to a microscope slide surface, this tends to be much weaker than the adherence of a completely dried agar specimen. Therefore, the hydrated agar samples protrude further off of the surface of the sample target plate and have a weaker adherence, which causes a greater risk of the sample dismounting and flaking into the downstream instrument ion optics.

The identity of the MALDI matrix and the selection of the application method are also important variables to consider in an imaging mass spectrometry experiment. The selection of an organic MALDI matrix depends largely on the sample substrate and target compounds of interest. For example, many low-molecular weight metabolites (MW < 300 Da) are comprised of many organic acids, phosphates, and other moieties, which are more readily ionized in the negative polarity²⁸. For negative mode analysis, basic matrices are preferred, such that their chemical structures are more amenable to accepting a proton (e.g., 9-aminoacridine, 1,5-diaminonapthalene), therefore leaving a negative charge on many small-metabolite ions of interest. We performed a matrix screen for the targeted analytes of interest and found that 1,5-diaminonapthalene (DAN) matrix produced the highest signal for amino acid metabolites of interest for control tissue samples. During MALDI matrix application, analytes of interest are extracted from the sample substrate and co-crystallize with the MALDI matrix. In some instances, "wetter" matrix application conditions will facilitate increased extraction and co-crystallization, resulting in improved analyte detection upon MALDI analysis. However, if the matrix application process uses too much solvent, analytes may be delocalized within the tissue. It is important to strike a balance between effective analyte extraction and maintaining the spatial fidelity of analytes in the tissue when selecting the conditions for matrix application. For example, a custom-built sublimation apparatus can be used to apply a homogeneous matrix layer to sample surfaces and produces extremely small matrix crystal sizes^29,30. However, this process is quite dry and may not be optimal for analyte extraction³¹. The robotic sprayer used here provides for reproducible and precise control of the spray pressure, flow rate, nozzle temperature, and other parameters that can allow for optimized methods of analyte extraction and matrix co-crystallization³².

The instrumental parameters on the FTICR MS used here have been optimized for the transmission and detection of metabolite ions in negative ion mode. This includes tuning RF and DC elements in the ion transfer region to preferentially transmit ions of low molecular weight (<m/z 200). Parameters that have a large effect on ion transmission include the funnel RF amplitude (65 V), the Q1 mass setting for quadrupole mass filter ion isolation (m/z 150), and the time-of-flight delay between the hexapole accumulation cell and the ICR cell (0.400 ms). These selected parameter values more efficiently transmit ions of low m/z values at the expense of the transmission efficiencies of ions of larger m/z values. Moreover, we have utilized a continuous accumulation of selected ions (CASI) approach, which allows for the selective enrichment of ions in a defined mass window of interest. In this approach, the quadrupole mass filter is used to selectively transmit a small range of m/z values from the ionization source to the hexapole accumulation cell in the instrument, while ions with m/z values outside this mass window have unstable ion trajectories and are not transmitted through the quadrupole mass filter. This allows for the selective enrichment of lowly abundant analytes of interest to improve the limit of detection and dynamic range by up to three orders of magnitude through the elimination of chemical noise. The high mass resolving power and mass accuracy of the FTICR instrument platform allows for the facile separation of isobaric (i.e., same nominal mass) compounds and the tentative identification of metabolites.

Careful and reproducible control of sample preparation, MALDI matrix application, and imaging mass spectrometry analyses can allow for reproducible views of bacterial biochemistry in colonies and tissue samples. Bacterial macrocolonies serve as simplified models to characterize metabolic cross communication of pathogens during microbial infection, which can then be compared to more complex systems such as animal tissues. For example, arginine is found to be lowly abundant in the C. difficile + E. faecalis co-culture sample compared to the C. difficile culture alone (Figure 6), which is supported by E. faecalis's noted ability to consume arginine as an energy source and export ornithine from the cell through the ArcD antiporter³³. The macrocolony results led to the investigation of this regulation of amino acids in an animal model. Mice infected with C. difficile + E. faecalis (wt) showed significantly less arginine than mice infected with C. difficile alone (Figure 7C), recapitulating the colony image results.

To test the hypothesis of arginine metabolism and the involvement of the ArcD antiporter, we have used a transposon mutant that knocks out metabolite transport through E. faecalis's ArcD antiporter. Mice infected with C. difficile + E. faecalis (arcD::Tn) show an increase in the detection of arginine and a decrease in the detection of ornithine relative to the wild-type co-infection model (Figure 7C) and generally rescue the original amino acid levels observed in the model of C. difficile infection alone (Figure 7A). This inverse relationship of metabolite regulation corroborates the findings that E. faecalis cross feeds on amino acid nutrients produced from C. difficile. Furthermore, histological analysis showed extensive cellular damage from mice infected with both C. difficile and E. faecalis, which is primarily observed in the wild-type strain of E. faecalis opposed to the ArcD transposon mutant. These results support our recent findings that arginine and ornithine are key to C. difficile virulence, behavior, and
fitness²⁷. Overall, this workflow demonstrates an effective method for preparing agar-based bacterial macrocolonies for MALDI imaging mass spectrometry and shows that these systems can serve as useful models for examining microbial cooperation during infection.

Materials

Name	Company	Catalog Number	Comments
0.2 μm Titan3 nylon syringe filters	Thermo Scientific	42225-NN
1,5-diaminonaphthalene MALDI matrix	Sigma Aldrich	2243-62-1
20 mL Henke Ject luer lock syringes	Henke Sass Wolf	4200.000V0
275i series convection vacuum gauge	Kurt J. Lesker company	KJL275807LL
7T solariX FTICR mass spectrometer equipped with a Smartbeam II Nd:YAG MALDI laser system (2 kHz, 355 nm)	Bruker Daltonics
Acetic acid solution, suitable for HPLC	Sigma Aldrich	64-19-7
Acetonitrile, suitable for HPLC, gradient grade, ≥99.9%	Sigma Aldrich	75-05-8
Ammonium hydroxide solution, 28% NH3 in H2O, ≥99.99% trace metals basis	Sigma Aldrich	1336-21-6
Autoclavable biohazard bags: 55 gal	Grainger	45TV10
Biohazard specimen transport bags (8 x 8 in.)	Fisher Scientific	01-800-07
Brain heart infusion broth	BD Biosciences	90003-040
C57BL/6 male mice	Jackson Laboratories
CanoScan 9000F Mark II photo and document scanner	Canon
CM 3050S research cryomicrotome	Leica Biosystems
Desiccator cabinet	Sigma Aldrich	Z268135
Diamond tip scriber, Electron Microscopy Sciences	Fisher Scientific	50-254-51
Drierite desiccant pellets	Drierite	21005
Ethanol, 200 Proof	Decon Labs	2701
flexImaging software	Bruker Daltonics
ftmsControl software	Bruker Daltonics
Glass vacuum trap	Sigma Aldrich	Z549460
HTX M5 TM robotic sprayer	HTX Technologies
Indium Tin Oxide (ITO)-coated microscope slides	Delta Technologies	CG-81IN-S115
In-line HEPA filter to vacuum pump	LABCONCO	7386500
Methanol, HPLC Grade	Fisher Chemical	67-56-1
MTP slide-adapter II	Bruker Daltonics	235380
Optimal cutting temperature (OCT) compound	Fischer Scientific	23-730-571
Peridox RTU Sporicide, Disinfectant and Cleaner	CONTEC	CR85335
PTFE (Teflon) printed slides, Electron Microscopy Sciences	VWR	100488-874
Rotary vane vacuum pump RV8	Edwards	A65401903
Tissue-Tek Accu-Edge Disposable High Profile Microtome Blades	Electron Microscopy Sciences	63068-HP
Transparent vacuum tubing	Cole Palmer	EW-06414-30
Ultragrade 19 vacuum pump oil	Edwards	H11025011
Variable voltage transformer	Powerstat
Water, suitable for HPLC	Sigma Aldrich	7732-18-5
Wide-mouth dewar flask	Sigma Aldrich	Z120790