Desorption Electrospray Ionization Mass Spectrometry Imaging (DESI-MS) allows spatial localization of changes in acetaminophen metabolism in the liver after intervention with 4-methylpyrazole

Abstract

Acetaminophen (APAP) overdose is the most common cause of acute liver failure in the US and hepatotoxicity is initiated by a reactive metabolite which induces characteristic centrilobular necrosis. The only clinically available antidote is N-acetylcysteine, which has limited efficacy and we have identified 4-methylpyrazole (4MP) as a strong alternate therapeutic option, protecting against generation and downstream effects of the cytotoxic reactive metabolite in the clinically relevant C57BL/6J mouse model and in humans. However, despite the regionally restricted necrosis after APAP, our earlier studies on APAP metabolites in biofluids or whole tissue homogenate lack the spatial information needed to understand region-specific consequences of reactive metabolite formation after APAP overdose. Thus, to gain insight into the regional variation in APAP metabolism and study the influence of 4MP, we established a Desorption Electrospray Ionization Mass Spectrometry Imaging (DESI-MSI) platform, for generation of ion images for APAP and its metabolites under ambient air, without chemical labeling or a prior coating of tissue which reduces chemical interference and perturbation of small molecule tissue localization. The spatial intensity and distribution of both oxidative and non-oxidative APAP metabolites were determined from mouse liver sections after a range of APAP overdoses. Importantly, exclusive differential signal intensities in metabolite abundance were noted in the tissue microenvironment, and 4MP treatment substantially influenced this topographical distribution.

Graphical Abstract

INTRODUCTION

Drug-induced liver injury (DILI) is a prominent health concern¹ and a major challenge for drug development worldwide². One of the most common causes of DILI in the developed world is that caused by an acetaminophen (APAP) overdose. APAP is an analgesic and antipyretic drug that is safe at therapeutic doses, because it is efficiently metabolized by glucuronidation and sulfation reactions in the liver to form the non-oxidative metabolites APAP-Glucuronide (APAP-GLUC) and APAP-Sulfate (APAP-SULF). Meanwhile, a minor component of APAP can form a reactive metabolite N-acetyl-p-benzoquinone imine (NAPQI), which is rapidly scavenged by hepatic glutathione stores^{3, 4} to form the oxidative metabolites, APAP-Glutathione (APAP-GSH), APAP-Cysteine (APAP-CYS) and APAP-N-acetylcysteine (APAP-NAC) preventing liver injury. However, the increased NAPQI generation after an overdose of APAP overwhelms the protective effect of GSH, resulting in the onset of acute liver injury, ultimately leading to centrilobular hepatocyte cell death. N-Acetylcysteine, which facilitates re-synthesis of the antioxidant glutathione, is the only clinical antidote against APAP overdose, despite serious concerns for its limited efficacy^5–8. We have recently demonstrated the safety and efficacy of 4-methylpyrazole (4MP, Fomepizole), an FDA-approved antidote for ethylene glycol and methanol poisoning, against APAP-induced liver injury in our studies in mice^{5, 9, 10} and human volunteers¹¹, with 4MP currently under a phase III clinical trial for this application. Since the proteins involved in APAP metabolism and the generation of NAPQI are zonated within the liver^{12, 13}, it is expected that a spatially distinct accumulation of APAP and its metabolites would occur in the liver after an overdose.

Biomarkers of drug metabolism have typically been assessed by liquid chromatography-tandem mass spectrometry (LC-MS/MS) in biofluids, such as plasma or urine, or whole tissue homogenates^14–17, as in our previous studies examining acetaminophen metabolism^{5, 9, 10}. Though this analysis provides some knowledge on the abundance of drugs or their metabolites, it gives no information on the local variations in drug metabolism within tissues or the regional concentrations of drug-derived metabolites that could be toxic to specific regions within individual organs. In fact, traditional methods used for gaining insight into the spatial distribution of drugs have generally relied upon the use of radiolabels or antibodies^{18, 19}. However, these methods are prone to miss secondary metabolites^20–23, which can accumulate in the tissue and cause toxic side effects²¹.

Mass spectrometry imaging (MSI) is a novel analytical technique that can spatially localize drugs and their metabolites in single cells, tissues, organs, and even within the whole body while maintaining spatial distinctions^24–31. Matrix-assisted laser desorption/ionization (MALDI)-MSI is the most widely used MSI technique that has been applied to study the distribution of abundant small molecule drugs^{32, 33}, and lipids³², and is very suitable for protein analysis^{34, 35}. However, there are intrinsic challenges to MALDI MSI, especially for small molecules (<600 Daltons) due to the predominance of ion suppression effects by matrix adducts^{36, 37}. Furthermore, the application of the MALDI matrices may also disturb the spatial localization of small molecules^{37, 38}.

Several studies have applied MALDI-MSI to study the spatial distribution of APAP metabolism in mouse liver sections³⁹ as well as liver⁴⁰ or whole-body sections⁴¹ from rats. However, no liver zonation was evident for APAP and its non-oxidative metabolites using the technique, which was unable to detect oxidative metabolites except APAP-GSH³⁹. Though the sensitivity of detection was improved for oxidative metabolites with on-tissue chemical derivatization, this was at the expense of a lower resolution (400 μm)^{40, 41} required to resolve the spatial distribution of analytes within the region of interest in the mouse liver³⁹. Thus, it is of utmost importance to have a label-free method for the direct tissue analysis of APAP and its metabolites to evaluate their spatial distribution. In contrast to MALDI-MSI, Desorption Electrospray Ionization (DESI) is an ambient ionization technique that allows the simultaneous analysis of small molecules under ambient air, without chemical labeling or a prior coating of tissue which may induce chemical interference or perturbation of drugs and their metabolites⁴². In fact, DESI-MSI-based metabolomics is a label and matrix-free tool with superior sensitivity for small molecules⁴³ and lipid analysis^{42, 44}, unlike MALDI-MS that requires analysis under vacuum.

Here we report on the spatial visualization of APAP metabolism within the liver after APAP overdose in the human-relevant C57BL/6J mouse model⁴⁵ by DESI-MSI analysis and demonstrate alterations induced by treatment with 4MP. The analyses were performed on the SYNAPT G2-Si and XS HDMS mass spectrometers (MS) equipped with a new DESI-XS source redesigned to achieve a comparable spatial resolution to MALDI (Waters, Milford, MA). HDMS is a novel mass spectrometry technology that relies on the mobility of ions based on the charge, shape, and size in the gas phase inside the MS instrument, which is suitable for small molecule drugs and their metabolites. We demonstrate that the SYNAPT XS, with redesigned Stepwave XS technology and superior MS resolution, especially for labile compounds like APAP, provides enhanced sensitivity for detection of APAP metabolites. Our report reveals that the enhanced performance (high spatial resolution, high mass resolving power, and high mass accuracy) improved the bio-localization of acetaminophen and its metabolites by reducing potential interference from endogenous compounds or matrices typically needed for MALDI-MSI analysis.

Materials and methods

Animals and experimental design

All animal procedures were approved by the Institutional Animal Care and Use Committee of the University of Kansas Medical Center. Experiments were performed following the National Research Council for the care and use of laboratory animal guidelines. This study used 8–12 week old male C57BL/6J mice purchased from Jackson Laboratories (Bar Harbor, Maine, USA). Upon arrival, animals were kept in a temperature-controlled room with a 14-hour light/10-hour dark cycle with ad libitum access to food and water. Food was removed 15 hours before in vivo studies, following which animals were injected intraperitoneally (i.p.) with the indicated doses of APAP (Sigma-Aldrich, St Louis, Missouri, USA) or saline vehicle⁵. Animals were then sacrificed by cervical dislocation under isoflurane anesthesia 30 minutes. At that point, blood was drawn from the caudal vena cava into heparinized syringes and plasma was obtained by centrifugation at 18,000 g for 3 min. Livers were divided into sections and snap frozen in liquid nitrogen.

Biochemical assays

To assess liver injury, plasma alanine aminotransferase (ALT) and aspartate transaminase activity (AST) were measured using kits from Pointe Scientific (Canton, MI) as per manufacturer’s instructions. Total tissue glutathione (GSH+GSSG) levels were measured using a modified Tietze assay as described⁴⁶.

Quantification of APAP-Cys protein adducts

APAP protein adducts were measured as described previously^{10, 47}. Briefly, free cysteine residues and low molecular weight metabolite conjugates that could interfere with the detection of APAP protein adducts were removed by filtering tissue homogenates through Bio-Spin 6 columns (Bio-Rad, Hercules, CA). Then, the filtrate was collected and mixed in a 1:1 ratio with an 8 U/ml solution of Protease from Streptomyces Griseus (Sigma-Aldrich, St. Louis, MO). The mixture was digested at 50° C for 15 hours and filtered through an ultrafree-MC Centrifugal filter of 0.22 μm pore size (Millipore Sigma, Burlington, MA). APAP-Cys derived from digested proteins were quantified by high-pressure liquid chromatography using a Coularray electrochemical detector (ESABiosciences, Chelmsford, MA) and compared to APAP-Cys standard from Toronto Research Company (Ontario, CA).

LC-MS/MS measurement of APAP and its metabolites

The level of APAP and its metabolites, APAP-glucuronide (APAP-Gluc), APAP-sulfate (APAP-Sulf), APAP-glutathione (APAP-GSH), free APAP-cysteine (APAP-CYS), and APAP-N-acetylcysteine (APAP-NAC) were analyzed using LC-MS/MS as previously described¹⁰. Briefly, a nine-point calibration curve in drug-free mouse tissue homogenate ranging from 0.25 to 75 μM was prepared. Then, 20 μl of the sample was supplemented with the internal standard 4-acetaminophen-d3 sulfate (APAP-d3) and analytes extracted with 100% methanol. All samples were vortexed, incubated on ice for 10 min, and centrifuged for 20 min at 20,000 g at 4 °C, followed by analysis of 5 μl of the supernatant by LC-MS/MS with a Waters Acquity Ultra-Performance Liquid Chromatography (UPLC) system equipped with a Xevo XS triple quadrupole mass spectrometer (Waters, Milford, MA). The lower limits of quantification (LLOQ) were determined to be 0.25 μM for APAP-GSH; 0.125 μM for APAP-Sulf and APAP-Gluc; 0.063 μM for free APAP-CYS; and 0.025 μM for APAP-NAC. Analyte levels below the limits of quantification were assumed to be zero. All standards were purchased from Toronto Research Chemicals (Toronto, Canada).

DESI-MSI mimetic model coupled with an in-line internal standard reference

A mimetic tissue model^48–50 was constructed to determine the limit of detection for APAP during DESI imaging on the quadrupole time-of-flight (QToF) mass spectrometers, SYNAPT HDMS G2-Si, and SYNAPT XS. Control snap-frozen liver tissue was homogenized and spiked with various concentrations of APAP, following which, DESI-MSI analysis was performed on SYNAPT HDMS G2-Si and SYNAPT HDMS XS mass spectrometers (Waters, Milford, MA) equipped with the Omni Spray 2-D version 2.0.1 (Prosolia, Indianapolis, IN), also modified to include a beta version of the high-performance sprayer in SYNAPT XS. General DESI acquisition parameters for the SYNAPT HDMS G2-Si MS were: MS acquisition mode: positive; MS acquisition rate: 0.5–1 s; TOF MS mode: resolution; sampling cone voltage: 25 V. MS source temperature: 150° C; Source sprayer voltage: 3 kV; Nebulizing nitrogen gas pressure: 0.5 MPa; DESI solvent: 98:2 methanol/water with 0.01% (v) formic acid; solvent flow rate: 3 μl/min; mass range: 50–1200 m/z; Pixel size: 200 μm. General DESI acquisition parameters for the SYNAPT HDMS XS MS were MS acquisition mode: positive; MS acquisition rate: 0.5–1 s; TOF MS mode: sensitivity; sampling cone voltage: 25 V. MS source temperature: 150° C; Step wave RF: 100; Ion guide RF: 150; Source sprayer voltage: 0.6 kV; Source sprayer gas pressure: 15 psi; DESI solvent: 98:2 methanol/water with 0.01% (v) formic acid; DESI solvent flow rate: 2 μl/min. Pixel size: 200 μm. To correct for any potential m/z drift during MSI experiments, leucine enkephalin was used as a lock mass after addition to the DESI sprayer solvent. Data acquisition was set up in the High-Definition Imaging (HDI) version 1.5 software and acquired using Masslynx software version 4.2 (Waters, Milford, MA). Following data processing, data visualization and regions of interest generation, and advanced analysis were all performed in HDI.

DESI-MSI of liver cryosections

Snap-frozen mouse liver tissue cryo-sectioned on glass slides were removed from the −80° C freezer and briefly dried under a desiccator immediately before DESI-MSI. Cryo-sections were subjected to DESI analysis on quadrupole time-of-flight (TOF) SYNAPT HDMS G2-Si and SYNAPT HDMS XS mass spectrometers (MS) (Waters, Milford, MA) equipped with the Omni Spray 2-D version 2.0.1 (Prosolia, Indianapolis, IN) modified to include a beta version of the DESI high-performance sprayer. All images were acquired with Omni Spray 2-D version 2.0.1 (Prosolia, Indianapolis, IN) combined with MassLynx version 4.2 (Waters, Milford, MA). DESI-MSI data was collected, processed, and analyzed using the High-Definition Imaging (HDI) version 1.5 software combined with MassLynx software version 4.1 (Waters, Milford, MA, USA). General acquisition parameters were: MS acquisition mode: positive; MS acquisition rate: 0.5–1 s; TOF MS mode: sensitivity; sampling cone voltage: 25 V. MS source temperature: 150° C; Step wave RF: 100; Ion guide RF: 150; Source sprayer voltage: 0.6 kV; Source sprayer gas pressure: 15 psi; DESI solvent: 98:2 methanol/water with 0.01% (v) formic acid; DESI solvent flow rate 2 μl/min. Pixel sizes differed per MSI analysis from 200 to 50 μm. Leucine enkephalin was added to the DESI sprayer solvent as a lock mass. Putative metabolite and lipid identification was made based on accurate mass searches in Lipidmaps⁵¹, Metlin⁵², and the Human Metabolome Database (HMDB)⁵³.

H&E and immunostaining of frozen tissue sections

Fresh snap-frozen liver tissues were cryo-sectioned at 12 μm thickness, following which sections were stained with hematoxylin and eosin (H&E) to illustrate morphological features on the liver section. Immunostaining was performed on liver cryosections after DESI MSI analysis. Tissues were fixed with 3% formaldehyde and then blocked with 5% normal goat serum. This was followed by overnight incubation with 1:200 diluted primary rabbit antibodies for cytochrome P450 2E1 (Abcam rabbit polyclonal antibody, cat. # ab28146) or cytochrome P450 2F2 (Cat. # sc-374540, Santa Cruz, Dallas, TX) at 4° C. The next day, sections were washed in PBS, followed by a 30 min application of SignalStain Boost Detection Reagent (Cell Signaling Technology, Rabbit no. 8114) or 1 h incubation with m-IgG Fc BP-HRP (sc-529765) followed by SignalStain DAB Chromogen detection (Cat. # 8059, Cell Signaling Technology, Danvers, MA) as per manufacturer’s instructions. After counterstaining with hematoxylin, tissue sections were imaged with a Zeiss Axioimager microscope (Carl Zeiss AG, Jena, Germany). Superimposition of images was done with ImageJ (http://rsb.info.nih.gov/ij/) and HDI.

Statistical analysis

All statistical analyses were performed by exporting data to SPSS Statistics 25 (IBM Co., Armonk, NY). Statistical analysis between two groups was performed with the Student’s two-tailed t-test while analysis between multiple groups was assessed by one-way analysis of variance (ANOVA), followed by Student-Newman-Keul’s test. In the case that data were not normally distributed, the Kruskal-Wallis test (non-parametric ANOVA) followed by Dunn’s Multiple Comparison Test was used. Differences with p < 0.05 were considered statistically significant.

RESULTS

Early hepatic glutathione depletion after APAP overdose without liver injury

Though acetaminophen is safe at therapeutic doses, an overdose can cause severe liver injury over time, and we initially examined circulating levels of ALT and AST as markers of liver injury after treatment of mice with a range of APAP doses. A concentration-dependent elevation in ALT and AST levels was evident by 30 minutes after the various doses, but since this is a very early time point, elevations were very mild and an increase above the upper range of normal for C57BL/6J mice⁵⁴ only occurred with the highest dose of 1200mg/kg (Figure 1A and B). So, was this lack of injury due to an absence of reactive metabolite generation after APAP at the lower doses? Depletion of hepatic glutathione due to scavenging of the reactive metabolite NAPQI is a central initiating feature of APAP-induced liver injury and so GSH levels were examined after administration of the various doses. As seen in Figure 1C, all doses of APAP resulted in significant depletion of liver GSH, confirming that formation of NAPQI and depletion of GSH occurred at all doses. Thus, it was the measurement at the very early time point that reflected the lack of appreciable liver injury. This is also illustrated in the liver histology (data not shown), where none of the samples showed evidence of necrosis at 30 minutes of APAP treatment, even at the highest dose of 1200mg/kg. Further confirmation of NAPQI generation and protein adduct formation is provided by the measurement of APAP-protein adducts, which demonstrates a distinct signal in livers from animals treated with 300mg/kg, which was substantially elevated after 1200mg/kg (Figure S1A). The dose-response curve also demonstrates a dose-dependent increase in protein adducts through all the concentrations tested (Figure S1B), confirming that APAP-mediated cell signaling to hepatotoxicity was initiated within 30 minutes of the overdose.

Fig. 1. Early hepatic response to the range of APAP overdose:

Fasted male C57BL/6J mice received various doses of APAP ranging from 300–1200 mg/kg or 10 ml/kg saline (control) and were sacrificed 30 minutes later. Plasma alanine transaminase (ALT) (A), aspartate aminotransferase (AST) (B) and tissue glutathione (GSH) (C) were then measured. Data represent means ± SEM of n = 6 animals per group. *P < 0.05 (compared with control).

APAP metabolism as assessed in whole liver homogenate

Since the focus of this study was the examination of APAP metabolism, this was then evaluated in liver homogenates from mice at the various doses indicated. As expected, the parent compound APAP showed a dose-dependent increase in the liver (Figure 2A). Glucuronidation and sulfation are the predominant metabolic pathways after a therapeutic dose of APAP and are also upregulated after an overdose. APAP-GLUC was prominent in the liver with levels slightly higher with the 1200 dose compared to others (Figure 2B). APAP-SULF levels were also appreciable, though lower than APAP-GLUC with the 1200mg/kg dose having much higher levels than the lower doses (Figure 2C). Looking at the oxidative metabolites, APAP-GSH though prominent, did not show appreciable differences between the various doses (Figure 2D). While APAP-CYS was much higher in doses of 600 and above compared to 300mg/kg (Figure 2E), APAP-NAC was the only oxidative metabolite that showed a distinct dose-response, being highest at the 1200mg/kg dose and consistently dropping at the lower doses (Figure 2F).

Fig. 2. Dose-dependent change in APAP and its metabolites in whole liver tissue 30 minutes after overdose:

Fasted male C57BL/6J mice received various doses of APAP from 300–1200 mg/kg or 10 ml/kg saline (control) and were sacrificed 30 minutes later. Hepatic concentrations of the parent compound APAP (A), the non-oxidative metabolites APAP-Glucuronide (B), and APAP-Sulfate (C), as well as the oxidative metabolites APAP-Glutathione (D), APAP-Cysteine (E), APAP-N-Acetylcysteine (F) were measured by LC/MS/MS. Data represent means ± SEM of n = 4 animals per group. *P < 0.05 (compared to 300 mg/kg).

Establishment of a DESI-MSI platform for mapping APAP metabolism

Now that we had established the characteristics of APAP metabolism after a range of APAP overdose in the total liver homogenate, further experiments were focused on determining their spatial localization. Towards this, initial experiments attempted to determine the efficacy of analyte extraction and the influence of ion suppression on the identification of metabolites by DESI. This was investigated using a tissue mimetic model spotted with various concentrations of APAP. To determine the most suitable MS instrument for our DESI-MSI analysis, we determined the limit of detection for APAP during DESI imaging at 200 μm pixel size on the SYNAPT G2 Si and SYNAPT HDMS XS mass spectrometers. While APAP signals from the tissue mimetic per se were below LOD (data not shown), spotted APAP was detectable on both instruments. For experiments on the SYNAPT G2-Si, supplementary figures S2A and C represent normalized intensities of a region of interest (ROI) of the APAP ion. Supplementary figure S2E shows the DESI images of the sodiated ion of APAP (green) overlaid with a common phospholipid (PC) lipid protonated ion (PC (36:5)). The plot, as well as the APAP ion intensity table and the ion images, show that LOD is between 0.1 to 0.01 mg/mL in the liver tissue matrix on the SYNAPT G2-Si. For experiments on the SYNAPT XS, supplementary figures S2B and D represent normalized intensities of regions of interest (ROI) of the APAP ions. Supplementary figure S2F represents DESI images of protonated, sodiated, and potassiated ions of APAP (green) overlaid with a common PC (36:5). The plot, APAP ion intensity table, and the DESI images show LOD between 0.01 to 0.001 mg/mL in the liver tissue matrix on the SYNAPT XS. Taken all together, a 10x lower APAP concentration was detected on liver tissue using the SYNAPT XS and new DESI sprayer, which corresponds to a 10x improvement in LOD. Thus, the SYNAPT XS MS was used for further DESI-MSI experiments.

Spatial visualization of APAP on the DESIXS-SYNAPT XS platform at 200 um pixel size

After establishing the DESI-MSI- SYNAPT XS platform, liver sections from animals treated with APAP (300mg/kg to 1200mg/kg) were analyzed and ion images were generated at a resolution of 200 μm for APAP species over the dose-response curve. The specificity of the signal is demonstrated by its complete absence in controls, which were not treated with APAP (Supplementary figure S3A–D). Protonated APAP was clearly detectable at the dose of 300mg/kg, with a much lower signal for the sodiated form and a very mild signal for the potassiated form (Supplementary figure S3). At higher doses of treatment, all forms of APAP were clearly detectable and the sum of the 3 ions (Supplementary figure S3D) indicates a concentration-dependent increase in the signal which replicates that seen on measurement of the parent compound by LC-MS.

Localization of APAP to distinct morphological features on liver sections

The data so far showed that APAP was detectable by DESI-MSI using the SYNAPT XS from liver sections of mice administered even a moderate (300mg/kg) overdose and showed a concentration-dependent increase at more severe overdoses. Yet, the primary aim of this study was to determine whether DESI-MSI could demonstrate the spatial differences in APAP metabolism which contribute to its characteristic pathophysiology. An APAP overdose induces centrilobular necrosis, with hepatocytes surrounding the central vein being the first cells to die, with the injury radiating outwards towards the portal triad. In order to determine spatial identity, we initially extracted ion spectra and averaged relative signal intensities from the regions of interest around the veins (Supplementary figure S4B) identified by H&E staining of the same tissue (Supplementary figure S4A) initially imaged at 200 μm pixel size. When we evaluated the intensities of APAP in the various regions of interest, we observed that the ion intensity of APAP was higher in areas of the vein as compared with the non-vein area (Supplementary figure S4C). Unfortunately, the H&E staining on cryosections did not allow clear demarcation of centrilobular areas from the portal area. Thus, we immuno-stained liver sections after DESI-MS analysis for Cytochrome P450 Cyp2E1, which is a marker of centrilobular cells, and Cyp2F2, a marker of periportal cells. As we have demonstrated earlier¹², this staining clearly demarcates areas on the section as pericentral (Cyp2E1 positive) and periportal (Cyp2F2 positive), which are mutually exclusive (Supplementary Figure S5 and magnified inset panel).

Visualization of spatial characteristics of GSH depletion and variance in signal intensity of APAP metabolites on liver sections

As described earlier, glutathione depletion due to the generation of the reactive APAP metabolite NAPQI is a critical feature of APAP pathophysiology and measurement in liver homogenates clearly indicates this (Figure 1C). Since enhanced spatial resolution would be critical to determine liver regions where glutathione levels change, we improved image resolution by performing DESI-MSI analysis at a higher resolution of 50 μm pixel size to obtain discrete spatial information in samples. The GSH signal is clearly visible throughout the tissue section in control animals when no APAP signal is evident (Supplementary Figure S6A). We were also able to reproduce GSH depletion seen in liver homogenates by DESI-MSI, as liver tissue from animals treated with a severe APAP overdose of 600mg/kg, which shows increased mortality at later time points with compromised liver regeneration⁵⁵ showed evident GSH depletion (Supplementary Figure S6A). Interestingly, the higher resolution of 50μm also revealed that the loss of GSH was predominant in the pericentral area (light green arrow) when compared to periportal regions (yellow arrow) at the 30 min time point after APAP overdose (Figure S6B). The next series of experiments examined the spatial distribution of APAP and its various metabolites such as APAP-GLUC, APAP-SULF, APAP-GSH, APAP-CYS, and the ultimate oxidative metabolite formed after APAP-GSH metabolism, APAP-NAC (Figure 3). The parent compound APAP (Figure 3A), as well as its non-oxidative and oxidative metabolites, were readily detected on liver sections with the APAP-GSH signal being much more elevated as compared to APAP-GLUC (Figure 3 D & B) corroborating the results from measurements in the whole tissue homogenate (Figure 2). The DESI-MSI ion images demonstrate that though the signal intensity for APAP-SULF (Figure 3C), APAP-CYS (Figure 3E), and APAP-NAC (Figure 3F) were lower than APAP-GSH, discrete signals with variable abundance and distinct regional specificity could be observed throughout the tissue.

Fig. 3. DESI-MS analysis of APAP and its non-oxidative and oxidative metabolites at 50 μm pixel resolution:

Fasted male C57BL/6J mice which received 600 mg/kg APAP were sacrificed 30 minutes later and liver sections used for DESI-MSI. DESI ion images of (A) APAP (m/z 152.0712) [M+H⁺], (B) APAP-GLUC (m/z 366.0591) [M+K⁺], (C) APAP-SULF (m/z 232.028) [M+H⁺], (D) APAP-GSH (m/z 457.1393) [M+H⁺], (E) APAP-CYS (m/z 2933.0362) [M+K⁺], (F) APAP-NAC (m/z 313.0858) [M+H⁺] at a resolution of 50 μm pixel size.

To further enhance the translational application of our findings, we compared these samples to those from mice that had been concurrently treated with 4-methylpyrazole (4MP), an FDA-approved antidote for ethylene glycol and methanol poisoning, which we have shown provides significant protection against APAP hepatotoxicity^{5, 9, 10} by preventing cytochrome P450 mediated reactive metabolite formation¹⁰ and inhibition of the MAP kinase JNK⁹, involved in the amplification of APAP-induced mitochondrial dysfunction^{56, 57}. The parent compound APAP, as well as APAP-NAC, were readily detected on liver sections of mice treated with 300mg/kg APAP, with significant elevation in parent APAP evident at the higher dose of 600mg/kg (Figure 4 A and C). Interestingly, animals treated with 4MP showed significant increase in parent APAP in both moderate and severe overdose, with concomitant disappearance of APAP-NAC (Figure 4 B and D) indicating robust inhibition of cytochrome P450 mediated generation of the oxidative metabolite.

Fig. 4. DESI-MSI visualization of the effect of 4MP on distribution of APAP and APAP-NAC following a moderate (300mg/kg) or severe (600mg/kg) APAP overdose at 50 μm pixel resolution:

Fasted male C57BL/6J mice received either 300 (A & B) or 600 (C & D) mg/kg APAP alone or along with 50 mg/kg 4MP (B & D) and were sacrificed 30 minutes later with liver sections used for DESI-MSI. Ion images of APAP (m/z 152.0712) [M+H⁺] + APAP (m/z 174.0531) [M+Na⁺], and APAP-NAC (m/z 313.0858) [M+H⁺] at 50 μm pixel size.

Correlating spatial distribution of APAP metabolism to pericentral and periportal areas

DESI MSI provides the unique ability to directly overlay the molecular information acquired from the mass spectrometry analysis on top of the same tissue section so that correlative comparisons of molecular and histologic information could be achieved. To determine whether the DESI ion signals for APAP and APAP-NAC could be localized to either the pericentral or periportal regions of the liver, we carried out DESI analysis at a 50 μm pixel size resolution and overlayed ion images on the corresponding sections stained for Cyp2F2 (positive staining in portal areas) in animals after a moderate 300mg/kg APAP overdose. Merging the images clearly denotes higher intensity signals for parent APAP mainly in the periportal areas (Figure 5A-soverlapping with stained areas), which would be expected since the liver blood supply flows from the hepatic artery and portal vein to the central hepatic vein where APAP is mainly metabolized by surrounding hepatocytes that highly express CYP2E1. The distribution of APAP-NAC was distinct from that of parent APAP (Figure 5B), but closer examination reveals interesting data (Figure 5C). The dynamic nature of ongoing APAP metabolism is clearly evident when comparing the APAP signal to that of its oxidative metabolite (APAP-NAC) in identical regions. Some areas show the overlap between the presence of APAP and APAP-NAC (yellow arrows), presumably due to ongoing metabolism, some others show an intense signal for APAP-NAC, but a faint signal for APAP (red arrows), indicative of just completed metabolism, while others show no APAP signal, but strong APAP-NAC signal (green arrows) suggesting complete APAP metabolism in that area (Figure 5C).

Fig. 5. Spatial localization of hepatic APAP and APAP-NAC by DESI-MSI analysis at 50 μm pixel resolution after moderate (300 mg/kg) APAP overdose:

Fasted male C57BL/6J mice received 300 mg/kg APAP followed by sacrifice 30 minutes later. A representative photomicrograph of the liver section immunostained for Cyp2F2 after DESI-MSI is shown on the left. DESI ion images for APAP (A) and APAP-NAC (B) were then superimposed onto the corresponding Cyp2F2 immunostained image to generate merged images which are enlarged in panels on the left and rotated for easier visualization below (C).

Moving on to animals after a severe APAP overdose (600mg/kg), it is immediately apparent that the APAP signal is much higher compared to the moderate 300mg/kg overdose (Figure 6A). Additionally, the higher dose of APAP administration also delays clearance of the parent compound from circulation, and an abundant signal is still visible within vessels (Figure 6A-enlarged panel, white arrow). However, the centrilobular areas do not have a significant signal, as in the animals given 300mg/kg, despite higher levels in circulation, which could be due to an increased rate of metabolism, especially to the oxidative metabolites as seen in Figure 2 E and F. This is reflected in the APAP-NAC ion signal (Figure 6B), which again localizes to some positions where intense APAP signal is also seen (Figure 6C-orange arrow), while at other locations APAP-NAC is much more intense than APAP (red arrow). Interestingly in some locations, the parent APAP signal is much more intense than the corresponding APAP-NAC (light blue arrow), again indicative of the dynamic nature of the events being examined. Further examination of the non-oxidative metabolites reveals that the APAP signal (blue) shows significant overlap with the APAP-GLUC (red) (Pink in Supplementary figure S7A-green arrow) and seems to be predominantly in the centrilobular area (Figure S7C). Interestingly, APAP-Sulf does not show significant overlap with the APAP signal (Figure S7B), but also seems localized to the centrilobular area of the liver (Figure S7D-yellow arrow). The oxidative metabolite APAP-GSH signal is very similar to APAP-GLUC, with significant overlap with the APAP signal (Figure S8A) and localization to the centrilobular area (Figure S8C-green arrow), while APAP-CYS again does not show significant overlap with parent APAP (Figure S8B) and seems concentrated in areas bordering pericentral and periportal areas (Figure S8D-yellow arrow) suggesting selective regional specificity of APAP metabolism.

Fig. 6. DESI-MSI analysis for spatial localization of APAP and APAP-NAC in liver sections at 50 μm pixel resolution after severe (600 mg/kg) APAP overdose:

Fasted male C57BL/6J mice received 600 mg/kg APAP followed by sacrifice 30 minutes later. A representative photomicrograph of the cryo-section used for DESI-MSI and then immunostained for CYP2F2 is shown on the left. DESI ion images for APAP (A) and APAP-NAC (B) were then superimposed on the same liver cryo-section immunostained for CYP2F2 to generate merged images enlarged in panels on the right and expanded with rotation below (C).

Effect of 4MP on the biodistribution of APAP and APAP-NAC after a severe 600mg/kg overdose

The data thus far clearly indicate that DESI-MS imaging provides sufficient resolution for spatial differentiation of APAP metabolism between the pericentral and periportal regions of the liver. But can it also illustrate clinically relevant alterations in this process with relevant spatial distinction? Thus, to further enhance the translational application of our findings, we examined whether changes in APAP metabolism induced by a clinically relevant intervention could be imaged by DESI-MS by repeating experiments in mice treated with 600mg/kg APAP, along with 4MP (50mg/kg). Inhibition of cytochrome P450 2E1 mediated oxidative metabolism by co-administration of 4MP with APAP resulted in a significant elevation in parent APAP compound throughout the liver (Figure 7A vs Figure 6A), indicating that at this higher dose, a significant percentage of APAP could be metabolized to NAPQI, despite upregulation of glucuronidation. This is confirmed by examination of the oxidative metabolite APAP-NAC, which is almost completely absent in animals treated with 4MP (Figure 4B & D), though faint signals are evident on closer examination (black arrow in enlarged panels on bottom) (Figure 7B vs Figure 6B).

Fig. 7. Effect of 4MP on biodistribution of APAP and APAP-NAC captured with DESI-MSI analysis of liver tissue sections at 50 μm pixel resolution after severe (600 mg/kg) APAP overdose:

Fasted male C57BL/6J mice received 600 mg/kg APAP followed by sacrifice 30 minutes later. A representative photomicrograph of the cryo-section used for DESI-MSI and then immunostained for CYP2F2 is shown on left. DESI ion images of APAP (A) and APAP-NAC (B) were then overlayed onto the same liver immunostained for CYP2F2 to generate merged images enlarged in panels on the right and expanded with rotation below (C).

It is important to note that stability of analytes on liver tissue sections is an important aspect to consider in DESI-MSI to obtain high-quality and reproducible results. We were cognizant of this aspect and every step of our analytical workflow; from tissue collection and sectioning to data collection and processing were optimized considering this. To achieve an unbiased evaluation of selected ions distribution within images, intensity values in selected ion images were normalized to the highest intensity of the internal standard, leucine enkephalin. This was performed for each metabolite separately as further demonstrated in Figure S9, where ion images from DESI-MSI analysis of two endogenous lipids LPA (32:0) and PC (38:6) reveal a non-uniform distribution of ions in tissues from control mice (Figure S9A). This pattern was replicated in samples from mice treated with both APAP and 4MP (Figure S9B) suggesting that distribution of analytes is inherently uneven within tissues irrespective of the treatment. Thus, our on-tissue normalization strategy with the leucine enkephalin proved to be important to account for the variation in analyte distribution across the tissue surface.

DISCUSSION

Understanding the biodistribution of drugs within tissue is an essential requirement of drug development as well as for the understanding of pathophysiology since most biological processes involve spatially distinct molecular associations of small molecules^{17, 27}. This knowledge is thus crucial for characterizing drug safety as well as understanding efficient target engagement in order to improve the efficacy of interventions for disease. MSI is now recognized as a powerful tool to determine the spatial distribution of small molecule drugs or their metabolites, without the requirement for labeling, in contrast to several other imaging methods. MSI also facilitates the assessment of drug safety and efficacy and provides insight into pathophysiological mechanisms by correlation of ion images with spatial cues. This is especially important in the case of tissue injury with distinct regional specificity such as centrilobular necrosis in drug-induced liver injury (DILI), which is a common clinical problem with poor prognosis^{1, 2}. One of the most well-studied examples is liver injury caused by an overdose of APAP, which is a widely used antipyretic/analgesic that helps to relieve pain and fever when taken at a therapeutic dose. APAP-induced liver injury is initiated by its metabolism by cytochrome P450, predominantly CYP2E1, to a reactive metabolite. This mainly occurs in the hepatocytes surrounding the central vein in the liver, where CYP2E1 is highly expressed¹². This zonated pattern of CYP2E1 expression correlates with the characteristic necrosis of hepatocytes surrounding the central vein seen after an APAP overdose. This dependence of hepatotoxicity on regional metabolism within the liver lobule illustrates the importance of understanding local variations in APAP metabolism within the liver since regional concentrations of APAP-derived metabolites may be differentially toxic to hepatocytes around the central vein or the portal triad.

At therapeutic doses, APAP is safe because of the liver’s ability to conjugate APAP to glucuronide and sulfate, but especially due to the ability of intracellular GSH within hepatocytes to scavenge the minimal amounts of CYP450-generated NAPQI. However, an overdose of APAP can generate an excessive amount of NAPQI that overcomes hepatic GSH stores^{58, 59}. This is precisely demonstrated by the almost complete depletion of hepatic GSH as measured in whole tissue homogenate (Figure 1C) and as revealed by the dramatically lower signal intensity of total GSH by DESI-MS at 30 min after APAP overdose in the murine model (Figure S6). Our initial assessment of APAP metabolism to determine total drug exposure along the dose-response curve by LC-MS/MS confirmed these early metabolic events and demonstrated that APAP-NAC was the only oxidative metabolite that showed a distinct dose-response. We also confirmed that no detectable liver injury was present at this very early time point despite the substantial decrease of hepatic GSH and dose-dependent increase of protein adducts. This illustrates the temporal course of cell signaling events related to APAP hepatotoxicity and highlights the fact that adducting cellular proteins alone, is not sufficient to cause hepatic damage since oxidative signaling through the mitochondria is a critical step for hepatocyte necrosis⁶⁰.

Despite the numerous clinical pharmacokinetic studies of APAP^61–64 and the emergence of MSI, only a few recent studies have investigated the liver zonation and whole-body distribution of APAP and its metabolites. These were carried out with MALDI-MSI, which is the most widely used MSI technique. However, the distribution or zonation of oxidative metabolites was not reported, with the exception of APAP-GSH, which was found to be restricted to the centrilobular area of mouse liver sections³⁹. This is despite imaging at a higher spatial resolution (20 μm) when compared to our analysis by DESI on the SYNAPT XS (50 μm), suggesting that the signals suffered from significant ion suppression from endogenous molecules or matrix adducts during the MALDI-MSI analysis. Following this early innovative study, sensitivity was improved for oxidative metabolites when using MALDI-MSI on the whole-body section of rats. However, this was at the expenditure of spatial resolution since that was only at 400 μm pixel size^{40, 41}. While this may suggest that the rat may be ideal for MSI analysis of drug metabolism after an APAP overdose, much higher APAP doses are needed to induce toxic effects, and this would have limited clinical relevance since it has been shown that the pathophysiology of APAP hepatotoxicity in the rat is distinct from that of humans and thus the rat is not a clinically relevant model for studying APAP hepatotoxicity⁴⁷. To circumvent these issues with MALDI-MSI for analysis of APAP metabolism, we now show that the label and matrix-free DESI-MSI technique⁴² allows us to spatially map changes in APAP metabolism in the liver of the clinicallyrelevant C57BL/6J murine model^{45, 47} after an APAP overdose. Our data demonstrate that there are distinct spatial differences in the biodistribution of APAP and its metabolites. The utility of advanced instrument capability and innovations in technology is illustrated by the enhanced results obtained from the SYNAPT XS, which has upgraded instrument capabilities as compared to the SYNAPT G2-Si originally used for this analysis. The StepWave XS ion guide in the SYNAPT XS used newer ion optic geometries to provide better ion beam focusing for increased sensitivity with less in-source fragmentation of labile compounds such as APAP during the transfer from the ion source to the quadrupole. Additionally, the extended TOF flight path in the XS also provided higher mass resolution under the same operating mode when compared to the G2-Si TOF. These improvements, coupled with the optimized focusing and sampling of the new DESI XS sprayer geometry, provided the necessary instrumentation method improvements that enabled tissue imaging of the effects of 4MP on APAP metabolism after an overdose. These capabilities allowed the successful detection of APAP and its metabolites, with significant elevation in the toxic oxidative metabolites after the severe APAP overdose at 600mg/kg. Discrete changes within the tissue illustrate higher intensity signals for parent APAP in the periportal areas, most likely because this is the area of first entry into the liver through the blood from the hepatic artery and portal vein, which then flows through the sinusoids towards the central hepatic vein for exit from the organ. Moreover, APAP metabolic reactions such as glucuronidation occur predominantly in cells surrounding the central vein¹³ and hence the disappearance of the parent compound in that area would be expected. In addition, the abundance of CYP2E1 in the central area would also result in the rapid conversion of parent APAP into NAPQI in this area. Thus, the data illustrates the distinct regional variation in parent APAP lifetime along the portal triad- central vein axis, with its rapid metabolism in the central vein microenvironment unlike the portal triad, allowing longer half-life in the portal region of the tissue. These data also illustrate the dynamic nature of ongoing APAP metabolism as would be expected in biological systems, since some areas showed intense signals for APAP-NAC while showing weak or no signal for APAP, whereas others had lower signals for APAP-NAC, but still showed signals for APAP.

The close replication of human pathophysiology after APAP overdose by the C57BL/6J mouse model has allowed us to demonstrate the safety and efficacy of 4MP against APAP-induced ALI in our earlier studies in mice^{5, 9, 10} and humans¹¹. We now show that DESI-MSI can replicate these clinically relevant findings and illustrate the significant increase in signal intensity from the parent APAP compound, along with the lack of detection of the oxidative metabolite APAP-NAC in animals treated with 4MP with APAP, which would be expected due to CYP2E1 inhibition by 4MP. These results, which to our knowledge, have never been reported with any other MSI techniques before, corroborate our earlier studies showing that 4MP plays an important role in protecting the liver against oxidative metabolite formation by blocking the formation of APAP-NAC¹¹. Thus, visualization of the spatial variance in signal intensity provides additional context on 4MP targets and its regional effects on APAP metabolism masked during the analysis of total liver homogenates.

In conclusion, for the first time to our knowledge, we report the successful ambient analysis of APAP metabolism using DESI-MSI at high mass and space resolution. The spatially distinct data which replicate biological indices also indicate that DESI on the SYNAPT XS^® is an innovative platform to perform analysis of drug metabolism at ambient ionization while minimizing matrix effect interference. The instrumentation allowed us to monitor the progression of APAP metabolism with increased sensitivity at high spatial resolution and high specificity due to the high mass accuracy for the parent drug and its metabolites. Though the identity of the parent APAP was confirmed by comparison to standard, it is true that the exclusive reliance on accurate mass might not be sufficient for identification of derivative APAP metabolites. However, confidence in our identification is enhanced by the replication of earlier studies evaluating the repurposed drug 4MP as an intervention for APAP hepatotoxicity which also highlights the biological and clinical relevance of this approach, with the additional benefit of spatial context. Supplementary experiments are ongoing to further confirm metabolite identification by additional MS/MS experiments and the use of isotopically labeled standards for each individual metabolite. Overall, our study provides an innovative label and matrix-free platform with high sensitivity that is advantageous to characterizing the spatial alterations in APAP metabolite abundance after 4MP treatment. This novel imaging platform could be used to investigate the influence of new therapeutic interventions on molecular targets and further drug development for the prevention of APAP hepatotoxicity.