In situ detection of fatty acid C=C positional isomers by coupling on-tissue mCPBA epoxidation with infrared matrix-assisted laser desorption electrospray ionization mass spectrometry

Abstract

Rationale:

Unsaturated fatty acids (UFAs) play vital roles in regulating cellular functions. In-depth structural characterization of UFAs such as localizing carbon–carbon double bonds is fundamentally important but poses considerable challenges in mass spectrometry (MS) given that the most widely accessible ion activation method, low-energy collision-induced dissociation (CID), primarily generates uninformative fragments (e.g., neutral loss of CO₂) that are not suggestive of the double-bond positions.

Methods:

m-Chloroperoxybenzoic acid (mCPBA) was uniformly deposited onto the sample slides using a TM Sprayer, converting the carbon–carbon double bonds into epoxides under ambient conditions. The epoxidation product was ionized In situ by infrared matrix-assisted laser desorption electrospray ionization mass spectrometry (IR-MALDESI-MS), and subsequently cleaved via CID, generating a diagnostic ion pair associated with the double-bond position. The reaction efficiency, sensitivity and relative quantification capability of the method were validated with five UFA standards dried on glass slides, and then this strategy was demonstrated on thin tissue sections of rat liver and human bladder.

Results:

The mCPBA reaction yielded conversion rates in the range of 44–60% in 10 min with high specificity and sensitivity. Further tandem mass spectrometry (MS/MS) of the mono-epoxidized products generated informative fragment ions specific to the double-bond positions, and relative quantification of positional isomers in binary mixtures was performed across a wide mole fraction from 0 to 1. An innovative spiral scan pattern was utilized during data acquisition, elucidating the major isomeric compositions of multiple UFAs from a tissue section in a single run.

Conclusions:

The on-tissue mCPBA epoxidation was implemented into an ambient MS imaging workflow to offer a rapid and simple way for In situ identification and relative quantification of double-bond positional isomers without the requirement for instrument modification. The method can be readily implemented on many other MS platforms to reveal the role of double-bond positional isomers in lipid biology and to discover potential biomarkers.

1 |. INTRODUCTION

Fatty acids (FAs) carry out a broad variety of physiological functions such as energy sources, cell membrane components and precursors of eicosanoids serving regulatory roles.^1–3 FAs structurally differ in acyl chain length, degree of unsaturation, and position and stereochemistry of carbon–carbon double bonds. Unsaturated FAs (UFAs) comprise an important subclass of FAs. Some polyunsaturated FAs (PUFAs) are also known as essential FAs (EFAs) which cannot be completely synthesized in the human body and need to be supplied by dietary intake,⁴ e.g., linoleic acid, of which deficiency can cause health issues, such as scaly lesion in skin.⁵ In addition, the location of the C=C bond on an unsaturated acyl chain is a key structural feature due to its profound biological effects. For instance, studies suggested positive associations between high ratios of dietary ω−6:ω−3 PUFAs and production of inflammatory mediators, while increased consumption of ω−3 PUFAs appeared to suppress inflammation.^6,7 Moreover, changes in C18 (Δ9):(Δ11) were observed in breast cancer.^8–10 Mass spectrometry (MS) is a primary tool for lipid structure characterization. MS imaging (MSI) combines the molecular specificity of MS with spatial information to reveal the distributions of respective analytes without the need for sample homogenization, extraction or labeling, and thus has gained widespread interest in the field of lipidomics.^11,12

However, identifying C=C positions in unsaturated lipids by MS still faces a great challenge because traditional tandem mass spectrometry (MS/MS) methods, e.g., low-energy collision-induced dissociation (CID), do not result in informative fragment ions for locating C=C bonds due to the high dissociation energy.¹³ Several advanced fragmentation techniques are capable of providing richer fragmentation spectra from which more thorough structural information can be derived. Ultraviolet photodissociation (UVPD) at 193 nm of unsaturated lipids enables a high-energy photoactivation process, resulting in the cleavage of C-C bonds adjacent to a C=C bond and yielding a diagnostic pair of ions with 24.0000 Da difference.^9,14 Ozone-induced dissociation (OzID) involves trapping of isolated ions with ozone vapor in an ion trap mass spectrometer, allowing for gas-phase ozonolysis of unsaturated bonds.^15,16 A variant approach is termed ozone electrospray ionization-MS (OzESI-MS), where the ozonolysis takes place in the electrospray ionization source.¹⁷ Electron-induced dissociation (EID) leads to extensive cleavage along the acyl chain and has been successfully used to localize double bonds in phosphatidylcholines.^18,19 Alternatively, ion mobility has been integrated with MS workflows for gas-phase separation of lipid C=C positional isomers.²⁰ However, these techniques all require modifications to the current MS instruments which limits the availability for many researchers.

An alternative way is to chemically derivatize double bonds into more labile functional groups suitable for low-energy CID. One emerging derivatization method is the Paternò–Büchi (PB) reaction, a [2 + 2] cyclo-addition reaction commonly used in organic chemistry, which has been successfully applied for pinpointing lipid C=C bonds in combination with MS in recent years. The PB reaction converts a C=C bond into a four-membered oxetane ring with a carbonyl compound such as acetone upon UV irradiation, which is cleaved after collisional activation, producing a diagnostic ion pair associated with the location of the double bond.^21,22 Online PB derivatization has also been recently demonstrated by using acetone/water as the electrospray solvent and allowing lipids dissolved in the solvent to pass through a UV-transparent region to apply UV irradiation at 254 nm.^8,13,23

Another chemical derivatization method, called the meta-chloroperoxybenzoic acid (mCPBA) epoxidation reaction, has been implemented in the MS measurement of FAs and phospholipids to determine double-bond positions. In this reaction the C=C bond is epoxidized by mCPBA, yielding a lipid epoxide with 15.9949 Da mass increase. Cleavage occurs from both sides of the epoxide via CID, generating a pair of diagnostic ion fragments with 15.9949 Da mass difference (Scheme 1). In comparison with the PB reaction, the mCPBA reaction offers high specificity and complete conversion,²⁴ and occurs under ambient conditions without the need to be initiated by UV irradiation, which makes it a more accessible option. By spraying mCPBA on mouse kidney sections using an airbrush and allowing them to react for 10 min for subsequent desorption electrospray ionization (DESI) MS/MS, Kuo et al²⁵ showed that the FA 18:1 (Δ11) displayed an enriched level in the kidney medulla, while the summed FA 18:1 was found at lower abundance in the medulla.

SCHEME 1.

mCPBA epoxidation reaction and subsequent cleavage of the generated epoxide forming a pair of diagnostic fragment ions separated by 15.9949 Da that are indictive of the double-bond position

In this study, we aimed at In situ analysis of FA positional isomers in animal tissues by coupling the on-tissue mCPBA epoxidation derivatization with an ambient mass spectrometry imaging (MSI) technique termed infrared matrix-assisted laser desorption electrospray ionization (IR-MALDESI). IR-MALDESI is a hybrid and soft^26,27 ionization source based on IR-laser desorption followed by electrospray post-ionization,^28–30 requiring little to no sample preparation prior to analysis. The analytical performance of the epoxidation method was first evaluated with representative FA standards, and the practical utility to biological tissue analysis was further demonstrated on rat liver and human bladder cancer tissue.

2 |. EXPERIMENTAL

2.1 |. Materials

Oleic acid (FA 18:1 (Δ9)), cis-vaccenic acid (FA 18:1 (Δ11)), linoleic acid (FA 18:2 (Δ9, Δ12)), γ-linolenic acid (FA 18:3 [Δ6, Δ9, Δ12], GLA), linolenic acid (FA 18:3 (Δ9, Δ12, Δ15), ALA) and 3-chloroperbenzoic acid (mCPBA) were purchased from Sigma-Aldrich (St Louis, MO, USA). Acetonitrile, isopropanol, formic acid, and water in Optima grade were purchased from Fisher Chemical (Fair Lawn, NJ, USA). Nitrogen gas for purging the sample stage enclosure was purchased from Arc3 Gases (Raleigh, NC, USA).

2.2 |. Preparation of lipid standard samples

Stock solutions of FAs were prepared in ethanol at a concentration of 80 mM. For method validation, the lipid standard ampules were opened on the day of experiment and were handled carefully. Stock solutions of FA 18:1 (Δ9), FA 18:2 (Δ9, Δ12) and FA 18:3 (Δ9, Δ12, Δ15) were mixed and diluted to 2 mM with ethanol unless otherwise noted. Similarly, stock solutions of FA 18:1 (Δ11) and FA 18:3 (Δ6, Δ9, Δ12) were mixed and diluted to 2 mM. For relative quantification of positional isomers, a series of mixtures of FA 18:1 (Δ9) and FA 18:1 (Δ11) or FA 18:3 (Δ6, Δ9, Δ12) and FA 18:3 (Δ9, Δ12, Δ15) were prepared with molar ratios of 0:100, 10:90, 20:80, 30:70, 40:60, 50:50, 60:40, 70:30, 80:20, 90:10 and 100:0. The total lipids were kept at 2 mM in all mixtures. For all experiments, 50-nL droplets were spotted onto plain glass microscope slides (Fisherbrand™ Premium; Thermo Fisher Scientific, San José, CA, USA) using a 0.5-μL syringe (Hamilton, Reno, NV, USA), and allowed to air-dry. Each sample spot had a diameter of about 5 mm.

mCPBA was dissolved in acetonitrile/isopropanol/water (65:30:5, v/v/v) at the desired concentrations, and then the mCPBA solution was sprayed on the sample spots evenly with a TM Sprayer (Model 3; HTX Technologies, Chapel Hill, NC, USA) to epoxidize the double bonds. The following TM Sprayer settings were applied: nitrogen gas pressure, 10 psi; temperature, 30C; flow rate, 20 μL min⁻¹; number of passes, 4; velocity, 500 mm min⁻¹; line spacing, 3 mm; pattern, crisscross; nozzle height, 40 mm.

2.3 |. On-tissue derivatization by mCPBA epoxidation

The rat liver was provided by NCSU Department of Biological Sciences and experiments were performed in accordance with NCSU IACUC protocols. The liver was cut into 10-μm-thick sections at −15°C via a CM1950 cryostat (Leica, Buffalo Grove, IL, USA). The tissue sections were thaw-mounted onto microscope glass slides, and were analyzed right away or they underwent on-tissue derivatization. The deidentified human bladder cancer tissue sample with matching normal tissue were obtained from Wake Forest Baptist Medical Center Tumor Tissue and Pathology Shared Resources (WFBMCTTPST) under approved protocol (IRB00036014, P.I. Said). The bladder tissues embedded in optimal cutting temperature (OCT) media were cut into 20-μm sections at −20°C, thaw-mounted onto microscope slides, and gently washed with pure water to remove OCT media. On-tissue epoxidation derivatization was carried out by spraying mCPBA at 35 mM onto the tissue sections using the same TM sprayer parameters noted above. Then the tissue sections were incubated under ambient conditions for 10 min and subjected to IR-MALDESI MS analysis.

2.4 |. MS parameters

All MS experiments were performed by IR-MALDESI MS in negative ionization mode. A novel “burst-mode” 2970-nm laser at a pulse rate up to 10 kHz developed by JGM Associates, Inc. (Burlington, MA, USA) was used in the current platform to resonantly excite O-H stretching bands of water, facilitating the desorption of neutral species from the sample. Five pulses/burst with a total energy of approximately 1 mJ was applied, and the laser beam was focused through a 50-mm effective focal length aspheric lens (Thorlabs, Newton, NJ, USA). An ice layer was uniformly deposited on the sample surface as a mid-IR absorbing matrix by cooling the sample stage down to −8°C and was maintained throughout the experiment by keeping the relative humidity inside the sample enclosure at 10 ± 2%.³¹ The electrospray solvent of 50:50 v/v acetonitrile/water with 1 mM formic acid³² was injected at 1.0 μL min⁻¹ flow rate to ionize the ablated neutral materials via an electrospray and transmit them into the mass spectrometer.

The IR-MALDESI source is coupled to an Orbitrap Exploris™ 240 mass spectrometer (Thermo Fisher Scientific, Bremen, Germany). The mass spectrometer was operated in “Small Molecules” mode with a RF lens level of 70%. The spray voltage and capillary temperature were set to 3.0 kV and 350°C, respectively. MS1 and MS2 mass spectra were acquired over m/z range 200–400 and 100–350, respectively, with a resolving power of 120,000_FWHM at m/z 200 and 2.5 Hz scan rate. For MS2 experiments, the isolation window of precursor ions was 1.0 Da, and a normalized collision energy (NCE) of 50% and 30% was applied to FAs with ≤2 and ≥3 C=C double bonds, respectively. The automatic gain control (AGC) function was disabled, and the injection time remained constant at 15 ms to coordinate the laser desorption and ion acquisition events. A regionof-interest (ROI) of 15 × 15 was sampled on each standard sample spot for either MS1 or MS2.

2.5 |. Nine-step multiplex MS2 data acquisition

To acquire more comprehensive information in a single experiment and to reduce data acquisition time, a spiral scan pattern was implemented in the analysis of tissue sections. Using this setup, each rastir pixel was split into 3 × 3 spatially separated spiral pixels, and each spiral pixel represented an individual scan event that was defined in the Xcalibur software (Thermo Scientific). A spiral pattern tool has been added to RastirX,³³ an innovative imaging control platform developed in our lab. The spiral pattern is realized by permuting the normal row-major order of scan locations for an ROI using a simple indexing scheme. RastirX supports scanning the spiral pixels in any of the 16 possible sequences that can be formed starting at the center of a 3 × 3 rastir pixel and visiting the spiral pixels in a clockwise or counter-clockwise fashion (for more details, see Figure S1, supporting information). Here, nine targeted MS2 scans of epoxidized FAs were performed in each rastir pixel (Figure 1). The spiral step spacing was 100 μm in both the x and y coordinates, which resulted in an integrated rastir pixel size of 300 × 300 μm accordingly. For each tissue section, a ROI of 18 × 18 was sampled containing 36 scans for each MS2 target.

FIGURE 1.

(A) Diagram showing the multiplex data acquisition procedure where each rastir pixel is split into nine spiral steps for MS2 analysis of nine target precursor ions in a single run. The number indicates the sequence of the stage movement. (B) MS2 target for each spiral step and the corresponding NCE

2.6 |. Data analysis

The initial data files were stored in Xcalibur raw file format and converted into mzML files using MSConvert from the ProteoWizard toolkit, and then converted into imzML image files by imzMLConverter. Ion images presented in this work were generated with ± 2.5 ppm m/z tolerance using MSiReader v1.02 (https://msireader.wordpress.ncsu.edu/).^34,35 The MSiReader built-in the MSiExsport tool was used to export ion abundance values for each pixel. Identifying C=C positional isomers in tissue sections was done by searching diagnostic MS/MS fragments against a home-built database spreadsheet (see supplementary Table S1, supporting information) containing FA positional isomers possibly existing in mammalian tissues and their corresponding fragments. A KruskalWallis test was used to determine the significance of difference between cancerous and normal bladder tissue using the Matlab function kruskalwallis.

3 |. RESULTS AND DISCUSSION

3.1|. Method validation with FA standards

The performance of the mCPBA epoxidation method for In situ pinpointing C=C positions was evaluated using five FA standards containing 18 carbon atoms and 1–3 double bonds. The FA standards dissolved in ethanol were spotted and dried on glass slides, which were subsequently sprayed with mCPBA of varying concentrations to estimate the conversion rate defined as the ion abundance ratio of the lipid epoxide to the total lipid consisting of the underivatized FA and its epoxidation product ($\text{\%Conversion}\text{rate}=\frac{{\text{A}}_{\text{Lipid}\text{epoxide}}}{{\text{A}}_{\text{Total}\text{ipids}}}\times 100\text{\%}$). Figure 2A compares the conversion rates after spraying with mCPBA at 0, 10, and 20 mM and being incubated under ambient conditions for 10 and 60 min. It can be clearly seen that the compositions of the epoxidation products were greatly enhanced after being treated with mCPBA. A higher concentration of mCPBA and a longer incubation time led to a nearly complete conversion ranging from 83.7% to 98.3%. Nonetheless, in this case PUFAs tended to be epoxidized at multiple double-bond sites, while the amounts of mono-epoxidized products (i.e., one oxygen atom is added to each lipid molecule) were reduced. Given that the multiply epoxidized lipids would yield MS2 spectra that are difficult to interpret,³⁶ the mono-epoxies were targeted for the following MS/MS analysis. To maximize the proportion of the monoepoxidized products, 10 mM mCPBA and 10 min incubation time were determined to be sufficient and selected for the method validation experiments, achieving the conversion rates of UFAs into mono-epoxies in the range of 44–60% (Figure 2B).

FIGURE 2.

(A) Conversion rates (%) from the epoxidization of FA 18:1, 18:2 and 18:3 with mCPBA of different concentrations and different incubation times. (B) Conversion rates (%) of FAs into epoxides with 10 mM mCPBA and 10 min incubation time, expressed as mean ± standard deviation (n >100)

Figure 3 displays the MS2 spectra of mono-epoxidized FAs via higher-energy collisional dissociation (HCD), showing that the observed fragment features are in accordance with the theoretical fragmentation pattern of cleavages at the epoxide group, which can be used to unambiguously assign the double-bond positions. The results for FA 18:1 (Figure 3A) illustrate that the fragmentation pathway producing the diagnostic ion with an aldehyde group is more favorable than that producing the diagnostic ion containing an olefin group. For FA 18:2 (Figure 3B), the cleavage occurred predominantly at the side that was close to the other unsaturated bond, resulting in a single diagnostic ion corresponding to each C=C position, which could be attributed to the production of fragments with a conjugated π-bond that helps stabilize the products. This observation is in line with the previously reported study by Kuo et al.²⁵ With regard to FA 18:3 (Figure 3C), a similar conjugated π-bond phenomenon, i.e., only one diagnostic ion for each C=C position, could be observed for the Δ9- and Δ15-positions in FA 18:3 ((Δ9, Δ12, Δ15), and for the Δ12-position in FA 18:3 (Δ6, Δ9, Δ12), while it was not the case for a double bond in close proximity to the carboxyl group, e.g., the Δ6-position in FA 18:3 (Δ6, Δ9, Δ12), which produced both aldehyde and olefin fragments at relatively high abundance. Overall, the MS2 results demonstrate that the mCPBA-epoxidized FAs generate fragment ions that are highly informative and specific for the locations of the C=C double bonds.

FIGURE 3.

Fragmentation schemes and HCD MS2 spectra of mono-epoxidized (A) FA 18:1, (B) FA 18:2, and (C) FA 18:3. Diagnostic peaks are labeled in red and the double-ended arrows indicate the ion pairs with 15.9949 Da mass difference. (D) Calibration curves for relative quantification of FA 18:1 (Δ9) and FA 18:3 (Δ6, Δ9, Δ12) were constructed with standard mixtures in different proportions. The ion abundances at m/z 171.1026, 199.1339, 113.0607 and 129.0556 are denoted as A₁₇₁, A₁₉₉, A₁₁₃ and A₁₂₉, respectively. GLA and ALA represents FA 18:3 (Δ6, Δ9, Δ12) and FA 18:3 (Δ9, Δ12, Δ15), respectively. Error bars represent standard deviations (n >100)

The limit of detection (LOD) was further investigated by serially diluting the mixed standard solution with ethanol from 2 mM to 1, 0.5, 0.2 and 0.1 mM. Figure S2A (supporting information) shows that the diagnostic ion at m/z 171.1026 from FA 18:1 (Δ9) was still observable even when diluted to 0.1 mM (signal-to-noise ratio (S/N) >10), and that from FA 18:2 and 18:3 was detectable above 0.2 mM (S/N >5). Hence, we estimated that a LOD of 2–4 fmol per laser spot was achieved with this method (Figure S2B, supporting information).

3.2 |. Relative quantification of mono- and polyunsaturated FA positional isomers

Relative quantification measurements were conducted by analyzing a series of isomeric mixtures in varying mole fractions from 0 to 1 while the total lipid concentration was kept constant at 2 mM. Then the calibration curve for relatively quantifying FA 18:1 (Δ9) in a mixture of Δ9 and Δ11 isomers was established by plotting the abundance fractions of aldehyde diagnostic ions (m/z 171/(171 + 199)) against the mole fractions of FA 18:1 (Δ9) in mixtures. Likewise, the calibration curve for relatively quantifying FA 18:3 (Δ6, Δ9, Δ12) in a mixture of Δ6, Δ9, Δ12 and Δ9, Δ12, Δ15 isomers was constructed by plotting the abundance fractions of the proximal diagnostic ions (m/z (113 + 129)/(113 + 129 + 171)) against the mole fractions of FA 18:3 (Δ6, Δ9, Δ12). The calibration plots shown in Figure 3D exhibit good linear relationships with R² values of 0.99 and a wide dynamic range from 0 to 1. The slopes of the calibration curves are close to unity, implying the positional isomers possess similar desorption, ionization and dissociation efficiency.

3.3 |. FA C=C isomers in mammalian tissues by multiplex MS2 analysis

The mCPBA epoxidation approach was subsequently applied to rat liver tissue sections as a proof-of-concept experiment. Nine UFA species (Figure 1) were chosen to demonstrate the applicability of the mCPBA epoxidation method to animal tissues owing to their generally higher abundances. Due to the vast complexity of the biological matrix, the concentration of mCPBA solution sprayed on tissue sections was elevated to 35 mM to yield adequate production of epoxides. A concentration ≥100 mM was found to considerably suppress the lipid signals. The derivatized tissue section demonstrates an up to 42-fold increase in the composition of mono-epoxidized species compared with the intact tissue section (Figures 4A and 4B). The conversion rates of these UFAs to mono-epoxies ranged from 30–59% (Figure 4C).

FIGURE 4.

Average MS1 spectra of 100 scans from the rat liver section (A) without treatment and (B) after derivatization with 35 mM mCPBA. Inserts: Zoomed-in mass spectra of m/z 105–350. Non-epoxidized and epoxidized FA peaks are labeled in red and blue, respectively. (C) Compositions (%) of mono-epoxies of nine FAs before and after the mCPBA epoxidation derivatization, expressed as mean ± standard deviation (n = 100)

For an IR-laser-based MS system of which the penetration depth is typically on a micrometer scale,³⁷ thin tissue material is ablated completely from the probed area after one laser shot, making multiple scans at the same spot impractical. To obtain lipid structural profiles with MS/MS and subsequently realize unambiguous identification, a multiplex MS2 data acquisition strategy adapted from Korte and Lee,³⁸ Hansen and Lee³⁹ and Ouyang et al⁴⁰ was adopted, which enables multiple precursor ions to be analyzed from a single tissue section in the consideration of a shorter analysis time without the need for co-registration of datasets from serial tissue sections to recognize spatially colocalized species. The sample stage was moved in a spiral pattern as detailed in section 2. The spiral movement is especially useful for histologically heterogeneous tissue sections as each rastir pixel is contained in a compact area, making it less likely to cross multiple subregions. Figure 5A presents the MS2 spectra of FA 18:1 epoxide, and the MS2 spectra of five other epoxidized FAs and their identifications are displayed in Figure S3 (supporting information). It is worth noting that even though some of the diagnostic ion abundances were multiplied 5/10-fold for a clear view in the MS2 spectra, the assignments should be confident because the double-bond positions were assigned when multiple diagnostic ions were present. Additionally, analysis was performed on off-tissue regions to evaluate the impact of the background FAs on the identification results. No fragments corresponding to the diagnostic ions were observed from the blank glass substrate (Figure S4, supporting information), confirming that the background had no effect on the double-bond position assignment. Three out of nine FAs (FA 20:1, 20:2 and 22:4) did not produce informative fragment ions presumably due to their insufficient abundances for a MS2 experiment. Regarding mono-unsaturated FAs (MUFAs), FA16:1 exclusively contained the Δ9 isomer, while FA 18:1 was shown to consist of a mixture of Δ9 and Δ11 isomers. The results revealed that long-chain PUFAs in rat liver were mostly ω−6 with the exception of FA 18:3, where ω−3 were primarily detected, together with a minor portion of ω−7. Based on the calibration curve shown in Figure 3D, the relative abundance of FA 18:1 (Δ9) in rat liver was determined to be 0.77 ± 0.04 (Figure 5D), which was comparable with the ratio (0.73 ± 0.01) from the literature that used the online PB reaction coupled with MS using a liquid microjunction surface sampling probe.⁸

FIGURE 5.

MS2 spectra of FA 18:1 epoxide show the presence of Δ9 and Δ11 isomers in (A) rat liver and (B) cancerous and (C) normal human bladder tissue. Diagnostic ions from different positional isomers are shown in different colors. ×5 and ×10 indicate that the ion abundances are amplified by 5 and 10 times, respectively. (D) Mole fractions of FA 18:1 (Δ9) in rat liver and cancerous/normal human bladder (n = 36)

To further evaluate its potential application in elucidating lipid dysregulation linked to disease state, FA positional isomer compositions in normal and cancerous human bladder tissues were analyzed using this strategy. Similarly, in the human bladder samples, FA 16:1, 18:2, 18:3, 20:3, 20:4 and 22:4 were found in pure forms of FA 16:1 (Δ9), 18:2 (Δ9, Δ12), 18:3 (Δ9, Δ12, Δ15), 20:3 (Δ8, Δ11, Δ14), 20:4 (Δ5, Δ8, Δ11, Δ14) and 22:4 (Δ7, Δ10, Δ13, Δ16) (Figure S3, supporting information). FA 18:1 was composed of Δ9 and Δ11 isomers (Figures 5B and 5C). The bladder tumor sample appeared to contain a slightly lower content of the Δ9 isomer compared with its normal counterpart (0.88 ± 0.04 vs. 0.92 ± 0.09, cancer vs. normal, Kruskal-Wallis test p-value = 8.8 × 10⁻⁴) (Figure 5D), but this change was not as substantial as reported in breast cancer with a rat/mouse model, where the relative abundance of FA 18:1 (Δ9) was determined to be approximately 0.7 in cancer while 0.9 in normal.^8,13 The difference could originate from the varying elongase enzymatic activity.⁴¹

4 |. CONCLUSIONS

We have developed an on-tissue chemical derivatization method aiming at In situ identification and relative quantification of double-bond positional isomers of MUFAs and PUFAs from tissue sections. By spraying the oxidizing reagent mCPBA onto samples, an epoxidation reaction between alkenes and mCPBA occurs, converting UFAs into FA epoxides labile at the epoxide group upon collisional activation with conversion rates up to 60%. The method with a TM sprayer provides a rapid (~7 min per sample slide), automated and reproducible workflow with the merits of high efficiency, specificity and sensitivity. The mCPBA epoxidation approach outlined here can be readily adapted to other In situ MS and MSI platforms and tissue types, which in particular may offer a more sophisticated understanding of alterations in FA metabolism relevant to disease state. Further exploration of this strategy will be directed to assess its applicability to other lipid classes of more complex structures from animal tissues, such as phospholipids, although MS3^8,25 or CID²⁴ may be needed for yielding diagnostic fragments from the acyl chains.