Mass spectrometry imaging of phosphatidylcholine metabolism in lungs administered with therapeutic surfactants and isotopic tracers

Mass spectrometry imaging (MSI) visualizes molecular distributions throughout tissues but is blind to dynamic metabolic processes. Here, MSI with high mass resolution together with multiple stable isotope labeling provided spatial analyses of phosphatidylcholine (PC) metabolism in mouse lungs. Dysregulated surfactant metabolism is central to many respiratory diseases. Metabolism and turnover of therapeutic pulmonary surfactants were imaged from distributions of intact and metabolic products of an added tracer, universally 13C-labeled dipalmitoyl PC (U13C-DPPC). The parenchymal distributions of newly synthesized PC species were also imaged from incorporations of methyl-D9-choline. This dual labeling strategy demonstrated both lack of inhibition of endogenous PC synthesis by exogenous surfactant and location of acyl chain remodeling processes acting on the U13C-DPPC-labeled surfactant, leading to formation of polyunsaturated PC lipids. This ability to visualize discrete metabolic events will greatly enhance our understanding of lipid metabolism in diverse tissues and has potential application to both clinical and experimental studies.

Abstract Mass spectrometry imaging (MSI) visualizes molecular distributions throughout tissues but is blind to dynamic metabolic processes. Here, MSI with high mass resolution together with multiple stable isotope labeling provided spatial analyses of phosphatidylcholine (PC) metabolism in mouse lungs. Dysregulated surfactant metabolism is central to many respiratory diseases. Metabolism and turnover of therapeutic pulmonary surfactants were imaged from distributions of intact and metabolic products of an added tracer, universally 13 C-labeled dipalmitoyl PC (U 13 C-DPPC). The parenchymal distributions of newly synthesized PC species were also imaged from incorporations of methyl-D 9 -choline. This dual labeling strategy demonstrated both lack of inhibition of endogenous PC synthesis by exogenous surfactant and location of acyl chain remodeling processes acting on the U 13 C-DPPC-labeled surfactant, leading to formation of polyunsaturated PC lipids.
This ability to visualize discrete metabolic events will greatly enhance our understanding of lipid metabolism in diverse tissues and has potential application to both clinical and experimental studies.
Supplementary key words Pulmonary surfactant • lipids • isotope labeling • matrix-assisted laser desorption/ionization • remodeling Phospholipids represent the major surface-active components of lung surfactant. The phospholipid content of surfactant contains high levels of the disaturated species dipalmitoyl phosphatidylcholine (DPPC; PC16:0/16:0) that is maintained by a complex intraalveolar metabolism. Unsaturated PC lipids synthesized on the endoplasmic reticulum are subject to acyl remodeling catalyzed by sequential phospholipase A 2 and lysoPC acyltransferase activities (1,2), intracellular trafficking into lamellar body storage vesicles, and secretion into the alveolus to form the mature DPPCenriched surfactant. Surfactant enriched in DPPC is then secreted into the alveolar lining fluid by exocytosis of lamellar bodies, followed by rapid adsorption to the air:liquid interface. Surfactant is subsequently catabolized by alveolar epithelial type II (ATII) cell endocytosis (3), metabolism by alveolar macrophages (4) and loss up the bronchial tree (5). A proportion of the surfactant taken up by ATII cells is subsequently recycled into lamellar bodies for re-secretion into the alveolus (6,7).
Prior work by some of us used the incorporation of deuterated methyl-D 9 -choline into PC molecular species to probe surfactant metabolism in greater detail from biological extracts (8,9). This work characterized the mechanisms of acyl remodeling in surfactant PC synthesis by animal models (9), adult volunteers (10) and acute respiratory disease syndrome patients (11). Recently, an alternative acyl remodeling mechanism involved in surfactant metabolism was demonstrated, whereby a proportion of surfactant DPPC is selectively remodeled into PC species containing the polyunsaturated fatty acids arachidonate (20:4) and docosahexaenoate (22:6) (12). This study administered exogenous therapeutic surfactants to adult mice, with tracer amounts of the universally 13 C-labeled (U 13 C) isotopomer of DPPC (U 13 C-DPPC) added to the surfactant.
In addition to quantifying overall surfactant catabolism, this experimental approach enabled characterization of recycling mechanisms of surfactant DPPC by following the incorporations into intact PC species of U 13 C-labeled fragments derived from U 13 C-DPPC hydrolysis.
Lung surfactant components such as DPPC can also be analyzed by mass spectrometry imaging (MSI), which offers a powerful approach to study the distributions of lipids throughout biological tissues (13,14). MALDI-MSI has demonstrated enriched location of polyunsaturated PC species within the bronchial tree (15), a distribution potentially linked to the involvement of oxidation products of arachidonate in the pathology of inflammatory lung diseases (16,17). However, the location of synthesis within lung tissue of PUFA-containing PC species, both directly by the CDP:choline pathway and by acyl remodeling, has not been established. This information is lacking because conventional MSI is only capable of providing static lipid compositions within a given tissue region at a fixed moment in time. This static limitation can be overcome by coupling MSI with stable isotope labeling: isotope uptake into lipid metabolic processes can be detected in a time-resolved manner, reflected by measurements of mass-shifted lipid signals containing the isotopic tracer or the products of its metabolic conversion. Such approaches have been more widely applied using secondary ion mass spectrometry (SIMS), but because of the extensive fragmentation observed by SIMS whereby often only elemental or diatomic fragments are observed (e.g., Nano SIMS), it is difficult to study the metabolism of individual lipid species (18,19). Recently, several groups have coupled isotope labeling with MALDI (20) or with desorption electrospray ionization (21), both of which enable detection of intact lipid species. These studies utilized heavy water labeling of mice to study region-specific lipid synthesis in breast cancer and mouse brain tissue, respectively. The widespread and nonspecific nature of heavy water labeling were significant limitations of these studies. These limitations, together with the modest mass resolving power of the MSI instrumentation used, meant many lipid signals, representing combinations of different labeled and unlabeled species, and different metabolic processes remained unresolved.
Here, we used MALDI-MSI at high mass resolving power coupled to a dual labeling approach utilizing both methyl-D 9 choline chloride and U 13 C-DPPC administered to mice to visualize lung and surfactant lipid metabolism. We report the simultaneous characterization and visualization of surfactant PC synthesis and catabolism throughout the lung tissue.

MATERIALS AND METHODS
Preparation of labeled surfactants U 13 C-DPPC was mixed with the porcine-derived Poractant alfa and the synthetic CHF5633 surfactants as described previously (12). Isotopic enrichments of U 13 C-DPPC in Poractant alfa and in CHF5633 were, respectively, 7.55% and 2.87% of total DPPC.

Animal labeling and sample preparation procedure
Mouse lungs used in this study were provided from a previous study of exogenous surfactant turnover in mice (12). Tissues were stored at −80 • C, shipped on dry ice for MALDI-MSI, then stored at −80 • C until sectioning and analysis. Briefly, animal procedures were approved internally by the University of Southampton Animal Welfare and Ethical Review Body and externally by the Home Office, Animals in Science Regulation Unit. Wild-type (C57BL/6) male mice aged between 8 and 12 weeks were used for this study. Randomization was not relevant in this study. Similarly, blinding was not practicable as the two surfactant used had different phospholipid compositions. Each mouse was instilled intranasally with 50 μl (4 mg, equivalent to 200 mg/kg body weight) of either Poractant alfa or the synthetic surfactant CHF5633, both containing U 13 C-DPPC. At the same time, each mouse also received a 100 μl intraperitoneal injection of methyl-D 9 -choline chloride (10 mg/ml in water). After labeling, the mice were euthanized by carbon dioxide asphyxia at 6, 12, 18, or 24 h. Bronchoalveolar lavage was performed in situ with 4 × 0.9 ml PBS, and the recovered bronchoalveolar lavage fluid (BALF) aliquots were combined. BALF was centrifuged at 300 × g for 10 min at 4 • C to pellet cells, and the supernatants were then transferred to new vials and stored at −80 • C until extraction. Lung parenchyma was quickly dissected from the main bronchi, placed in cryotubes and snap-frozen in liquid nitrogen. Further details are provided in reference (12).

Alveolar macrophage labeling in vivo
For in vivo labeling studies, eight male wild-type C57BL6 mice were instilled with CHF5633 surfactant containing U 13 C-DPPC, then injected with methyl-D 9 -choline chloride 18 h later. After 3 h, mice were euthanized and BALF obtained, from which alveolar macrophages were isolated by centrifugation at 400g for 10 min BALF Macrophages were purified by adherence to plastic tissue-culture dishes and were subsequently washed three times with 0.9% saline to remove adherent surfactant.

Macrophage phospholipid analysis
Total lipids were extracted by scraping adherent alveolar macrophages into 800 μl of 0.9% saline after adding dimyristoyl PC (10 nmoles) as internal standard, followed by extraction with dichloromethane (2 ml), methanol (2 ml), and water (1 ml) (22). After mixing and centrifugation at 1,500g, 20 • C for 10 min, the dichloromethane-rich lower phase was recovered, dried under a stream of nitrogen gas, and stored at −20 • C until analysis by mass spectrometry. Mass spectrometry analysis was performed on a Waters XEVO TQ mass spectrometer (Waters UK, Elstree, UK) using electrospray ionization as described previously (9). Unlabeled PC and newly synthesized PC labeled with D 9 -choline were detected using precursor ion scans of the PC headgroup fragment ions at m/z +184 and m/z +193, respectively. Precursor ion scans of m/z +189 detected the PC species containing five labeled 13 C atoms in their phosphocholine head group.

Tissue preparation for MALDI-MSI
Frozen lung tissue was sectioned for MALDI-MSI analysis using a Leica CM 1860 UV cryomicrotome (Leica Microsystems, Wetzlar, Germany) at a temperature of −20 • C to produce 12 μm-thick tissue sections, which were thaw mounted onto clean indium tin oxide-coated glass slides (Delta Technologies, Loveland, CO). Tissue sections were stored at −80 • C until matrix application and analysis. 2,5dihydroxybenzoic acid was prepared as a 20 mg/mL solution in 2:1 CHCl 3 :MeOH (v/v) and applied to the sample using a SunCollect automatic pneumatic sprayer (Sunchrom GmbH, Friedrichsdorf, Germany). In total 15 layers of matrix were deposited onto the tissue. The first three layers were deposited using a flow rate of 10, 20, and 30 μL/min, respectively, and all subsequent layers at 40 μL/min.

Hematoxylin and eosin staining
Hematoxylin and eosin staining was performed on tissue sections after MSI analysis. The matrix was removed by immersion in 100% methanol for 30 s, after which the following protocol was used. The tissues were first washed with series of solutions (2 × 95% EtOH, 2 × 70% EtOH and deionized water for 2 min each), stained with hematoxylin for 3 min and subsequently washed with running tap water for 3 min. Tissues were then stained with eosin for 30 s and washed with running tap water for 3 min. After staining, the samples were placed into 100% ethanol for 1 min, then in xylene for 30 s. Finally, glass coverslips were placed onto the samples using Entellen mounting medium. High-resolution optical scans of the stained tissue sections were acquired using a Leica Aperio CS2 (Leica Biosystems Imaging, Vista, CA).

MALDI-MSI
MALDI-MSI was performed using a hybrid Orbitrap Elite mass spectrometer (Thermo Fisher Scientific GmbH, Bremen, Germany) coupled to a reduced pressure ESI/MALDI ion source (Spectroglyph LLC, Kennewick, WA). Further details on the experimental setup can be found in reference (23). All MSI data were collected using a nominal mass resolution setting of 240,000 @ m/z 400, an m/z range of 350-1,000 and a pixel size of 40 × 40 μm 2 . Tandem mass spectrometry (MS/ MS) was performed on MALDI ions generated directly from lung tissue sections to confirm the identification the selected signals. MS/MS spectra were acquired using resonant collision-induced dissociation, a ± 0.5 Da isolation window and Orbitrap detection of fragments.

Data analysis
MSI data analysis and visualization were performed using LipostarMSI (Molecular Horizon Srl, Perugia, Italy) (24). During data import, mass spectra were recalibrated using the [PC32:0+Na] + ion as a lock mass (theoretical m/z = 756.551374). Peak picking was performed using a relative intensity threshold of 0.05%, a minimum peak frequency of 0.5%, and an m/z alignment tolerance of 3 ppm. All MSI data visualizations were generated using total ion current normalization with hotspot removal (quantile thresholding, high quantile = 99%). An exception was the ratio images that were produced by normalization to the denominator ion signal ([PC32:0+Na] + or ([PC32:1+Na] + ).

Lipid nomenclature
Sum-composition PC molecular species are generally denoted by the number of total fatty acyl carbon atoms followed by the number of unsaturated double bonds. For example, PC36:4 corresponds to a PC lipid with 36 carbons and 4 double bonds distributed across the two acyl chains. In cases where the length and degree of unsaturation of each acyl chain are known, but the sn-positioning of individual fatty acyls is undefined, the underscore nomenclature is used, for example PC16:0_20:4. When the sn-positioning is defined, the "/" nomenclature is used. For example, PC16:0/20:4 corresponds to a PC lipid with a 16:0 acyl chain at the sn-1 position and a 20:4 acyl chain at the sn-2 position. Stereochemistry and double bond positions are undefined. Sum-composition lipid species can consist of multiple molecular species. For instance, PC36:4 could be either PC18:2_18:2 or PC16:0_20:4. The exception is for newly synthesized PC species containing 13 C 24 -palmitoyl-lysoPC derived from U 13 C-DPPC metabolism where, for instance, 13 C 24 -PC36:4 is unambiguously the palmitoylarachidonoyl PC16:0_20:4 species. Fragment ion nomenclature is based on that proposed by Pauling et al. in reference (25).

MSI of isotopically labeled lipids
We aimed to analyze lipid metabolism in mouse lung tissue after dosing intraperitoneally with D 9choline chloride and intranasally with an exogenous therapeutic surfactant containing U-13 C-DPPC. Both the complex mass spectra generated by MALDI analysis of biological tissue and the expected low abundance of labeled lipid species (∼1-2% relative to the corresponding unlabeled lipid species) make the unambiguous detection of both D 9 -choline-labeled PC and U 13 C-DPPC directly from lung tissue a challenge. Here, we have utilized the high resolving power of an Orbitrap mass spectrometer to enable unambiguous detection of isotopically labeled species in the presence of isobaric interferences arising from endogenous lipids. The average mass spectrum of a mouse lung tissue showed that protonated, sodiated, and potassiated ions of PC lipids dominate the spectrum with PC32:0 (predominately PC16:0/ 16:0), the major lipid component of lung surfactant, providing the highest signal intensities with the base peak at m/z 756.5513 assigned as the [PC32:0+Na] + ion (Fig. 1A). Importantly, we detected D 9 -labeled PC32:0, corresponding to PC32:0 synthesized over the 12 h between label injection and sacrifice, with the [D 9 -PC32:0+Na] + ion detected at m/z 765.6079 with an intensity of ∼1% compared with the unlabeled variant (Fig. 1B). The corresponding protonated and potassiated ions were also observed (supplemental Fig. S1). We detected [U 13 C-PC32:0+Na] + at m/z 796.6854 that arises exclusively from the therapeutic surfactant (Fig. 1C), as well as its corresponding protonated and potassiated ions (supplemental   pathway and D 9 -choline incorporation as well as exogenous surfactant using U 13 C-PC32:0 as a specific marker ion directly from lung tissue sections using MALDI-MSI.

IMAGING OF SURFACTANT PHOSPHATIDYLCHOLINE METABOLISM IN MOUSE LUNGS
Next, we sought to study the distribution of both D 9 -choline-labeled PC32:0 and intranasally dosed exogenous surfactant within mice lung tissue. We administered two different surfactants-synthetic CHF5633 and porcine lung-derived Poractant alfa. These preparations have different phospholipids, with CHF5633 being composed of a 50/50 mixture of PC16:0/16:0 and PG16:0/18:1 while Poractant alfa contains the full complement of surfactant lipids. U 13 C-PC16:0/16:0 was added to both surfactants to monitor exogenous surfactant metabolism before intranasal administration to mice that were also injected with methyl-D 9 -choline chloride to assess endogenous lung PC synthesis occurring via the Kennedy pathway (26).

CHF5633 synthetic surfactant
The distribution of several sodiated PC lipid ions throughout the mouse lung 12 h after administration of CHF5633 are shown in Fig. 2. The optical image of the hematoxylin and eosin stained of the same tissue after MSI is shown in Fig. 2A. The U 13 C-DPPC ion at m/z [M+Na] + 796.6855 was located in one parenchymal area but not within bronchioles, demonstrating one extreme example of the regional deposition (Fig. 2B). As expected, the distribution of unlabeled DPPC at m/z [M+Na] + 756.5514 (Fig. 2C) partially mirrored that of U 13 C-DPPC ( Fig. 2A) because of the excess amount of unlabeled DPPC in the CHF6533 surfactant. By contrast, the distribution of ion signals corresponding to the incorporation methyl-D 9 -choline into PC32:0 via the Kennedy pathway ( Fig. 2D; [M+Na] + , m/z 765.6079) exhibited a reduced intensity in regions where U 13 C-DPPC was highest, demonstrated by the ratio of D 9 -PC32:0+Na + /PC32:0+Na + (Fig. 2E); this observation could suggest inhibition of endogenous PC synthesis by the exogenous surfactant. However, both unlabeled PC32:1 (Fig. 2F, [M+Na] + , m/z 754.5359), which is an integral component of mouse lung surfactant but absent from CHF5633 and D 9 -PC32:1 (Fig. 2G, [M+Na] + , m/z 763.5923) were similarly present at lower intensity in regions enriched in CHF5633. Consequently, the ratio of [D 9 -PC32:1+Na] + /[PC32:1+Na] + was more uniform across the lung parenchyma ( Fig. 2H) with no decrease in areas of high U 13 C-DPPC signal. Similar distributions were apparent for other endogenous PC species with a parenchymal distribution (supplemental Fig. S3) and from tissue obtained from another mouse administered with both labels 6 h before tissue collection (supplemental Fig. S4). These results suggest the decreased [D 9 -PC32:0+Na] + /[PC32:0+Na] + ratios are a consequence of the regional, nonspecific accumulation of the DPPC-rich CHF5633 surfactant, rather than inhibition of PC synthesis. This is supported by extracted region-specific spectra that show regions of high U 13 C-DPPC to contain 2-3 fold higher signal for unlabeled DPPC than surrounding regions (supplemental Fig. S5). Although some lipid signal was observed on the glass slide adjacent to the tissue sections, which may arise because of smearing artefacts during tissue mounting, this did not alter the on-tissue distributions.
In agreement with previous reports (15), MSI demonstrated spatial segregation of unlabeled PC species, with the polyunsaturated PC species PC36:4 ( Fig. 3A, [M+Na] + , m/z 804.5514) and PC38:6 ( Fig. 3B, [M+Na] + , m/z 828.5515) restricted to bronchiolar regions (Fig. 3C, D). Similar results were obtained from another animal euthanized 6 h after label administration (supplemental Fig. S6). Little to no signal was observed for exogenous CHF5633-specific U 13 C-DPPC in the bronchiolar regions. Unfortunately, because of the low abundance of these polyunsaturated species compared with PC32:0 and PC32:1 that form major lipid components of surfactant, combined with the presence of an unresolved isotopologue from another lipid-related ion, the corresponding D 9 -containing PC36:4 and PC38:6 ions could not be unambiguously detected.
Incorporation of methyl-D 9 -choline also facilitated analysis of the distribution patterns of other newly synthesized PC species. Other PC species characteristic of lung tissue rather than surfactant, such as PC34:1 and PC34:2 (27) were distributed more equally between bronchiolar and parenchymal regions both for unlabeled and D 9 -choline-labeled species (supplemental Fig. S3). Similar distributions were observed for the same ions detected from lung collected from another mouse 6 h after label administration (supplemental Fig. S4). For all labeled ions discussed above, no signal was observed in a control lung tissue that was not administered with either D 9 -choline or U 13 C-DPPC (supplemental Fig. S7).

Poractant alfa porcine surfactant
Nasal administration of the porcine-derived Poractant alfa surfactant typically achieved a more widespread distribution in the lungs compared with CHF5633, illustrated by the distribution of the [U 13 C-DPPC+Na] + ion at 12 h postdosing (Fig. 4A). Analysis of another mouse 18 h after administration showed comparable results (supplemental Fig. S8). We observed a more uniform distribution of unlabeled [PC32:0+Na] + (Fig. 4B) and [D 9 -PC32:0+Na] + (Fig. 4C,  D), although the normalized signal intensity for [D 9 -PC32:0+Na] + was still slightly more intense in lung regions with a relatively low U 13 C-DPPC signal because of the presence of unlabeled PC32:0 in the Poractant alfa surfactant.

Alveolar metabolism of U 13 C-DPPC-labeled surfactant in vivo
Previous analysis of the in vivo metabolism of U 13 C-DPPC-labeled surfactants in mice by shotgun lipidomics analysis of extracts identified the labeled 13 C 5 -choline and 13 C 24 -lysoPC16:0 moieties, generated by the respective phospholipase D and phospholipase A 1 /A 2 enzymatic hydrolysis of U 13 C-DPPC, and their incorporation into PC molecular species (12). The molecular specificity for 13 C 5 -PC lipid synthesis closely mirrored that observed for D 9 -choline labeling. This indicates that 13 C 5 -choline generated by U 13 C-DPPC hydrolysis equilibrated with endogenous unlabeled choline and was then incorporated into PC by the CDP:choline (Kennedy) pathway (12). By contrast, reacylation of 13 C 24 -LPC16:0 in both lung tissue and bronchoalveolar lavage was much more restricted, generating primarily only two polyunsaturated PC species, 13 C 24 -PC16:0_20:4 and 13 C 24 -PC16:0_22:6 (12). Our analysis of the lavage cell pellet from this study (12) showed a higher enrichment of both these species than in either lung tissue or bronchoalveolar lavage (supplemental Fig. S9). As this cell pellet contained both alveolar macrophages and the more easily precipitated lung surfactant, we hypothesized that reacylation of 13 C 24 -LPC16:0 takes place in alveolar macrophages. To test this, mice received nasal administration of U 13 C-DPPC-labeled CHF5633, an intraperitoneal injection of methyl-D 9 -choline chloride (1 mg) 18 h later, were euthanized at 21 h after the first administration, and lung tissue was taken for analysis. Alveolar macrophages were purified by centrifugation at 400g for 10 min, followed by differential adherence and washing to remove surfactant. There was negligible incorporation of the D 9 -choline label into alveolar macrophage PC, indicating a very low rate of PC synthesis de novo (data not shown). This contrasts strikingly with the high rate of surfactant PC synthesis by alveolar type II epithelial cells, which was greatest at 3 h (12). In comparison, active re-acylation of 13 C 24 -LPC16:0 via the Lands cycle (28,29) by alveolar macrophages in vivo could be readily demonstrated 21 h after CHF5633 administration (12). Increased unlabeled PC32:0 in alveolar macrophages showed the uptake of exogenous surfactant by these cells (Fig. 5A). However, analysis of 13 C 24 -PC also showed considerable metabolism of this intracellular exogenous surfactant over the same time period. The relative abundance of U 13 C-DPPC decreased from the initial administered value of 100% at 0 h, whereas those of 13 C 24 -PC16:0_20:4 and 13 C 24 -PC16:0_22:6 re-acylation products increased (Fig. 5B). This finding was observed for all eight treated mice, with the mean relative abundances of U 13 C-DPPC, 13 C 24 -PC16:0_20:4, and 13 C 24 -PC16:0_22:6 being 38.7 ± 8.9%, 36.8 ± 9.7%, and 24.5 ± 6.4% (mean ± SD), respectively, relative to U 13 C-DPPC at t=0 (data not shown). There was no detectable synthesis of 13 C 24 -DPPC by alveolar macrophages in vivo, which is consistent with the absence of DPPC synthesis by acyl remodeling in this cell type (30). . All images were visualized using total-ioncurrent normalization and using hotspot removal (high quantile = 99%). H&E, hematoxylin and eosin; MSI, mass spectrometry imaging.

Imaging surfactant acyl remodeling in vivo
If re-acylation of 13 C 24 -LPC16:0 is restricted to alveolar macrophages in vivo, then 13 C 24 -PC16:0_20:4 should not co-localize with unlabeled PC36:4 in the bronchioles. Consequently, we monitored the distribution of 13 C 24 PC species in mouse lungs 12 and 18 h after administration of U 13 C-DPPC-labeled Poractant alfa, specifically chosen for this analysis rather than CHF5633 for two reasons. First, the more uniform distribution of the porcine surfactant (Fig. 4) facilitated a clearer analysis of the distribution of surfactant metabolites. Second, enrichment of U 13 C-DPPC was higher in administered Poractant alfa than in CHF5633 (12), thereby providing sufficient signal intensity for imaging of the low abundance 13 C 24 -labeled acyl chain remodeling products.
In our MALDI-MSI analysis, we detected the [ 13 C 24 -PC16:0_20:4+Na] + ion at m/z 828.6321 (Fig. 6A) with a relative intensity of ∼0.1% compared with the [DPPC+Na] + ion in the mean spectrum. This compound could be assigned with high confidence because of the low mass error of 0.2 ppm and absence of this signal in an undosed mouse lung, despite its low abundance (supplemental Fig. S7). Further confirmation for the assignment of this ion as [ 13 C 24 -PC16:0_20:4+Na] + was provided by the MS/MS spectrum (Fig. 6B)  Part-per-million (ppm) mass errors are indicated in parentheses. All images were visualized using total-ion-current normalization and using hotspot removal (high quantile = 99%). DPPC = PC16:0/16:0. MSI, mass spectrometry imaging; PC, phosphatidylcholine; U 13 C-DPPC, universally 13 C-labeled dipalmitoyl PC.
(red) that is synthesized by successive de-acylation/reacylation processes of U 13 C-DPPC through the Lands cycle. These processes led to replacement of a U 13 C-16:0 fatty acyl with an unlabeled 20:4 acyl chain, leading to the formation of 13 C 24 -PC16:0_20:4. Thus, the detection of both the lipid substrate and end products of this enzymatic-driven conversion provides the first direct observation of acyl remodeling events in tissue using MSI. Signal for [U 13 C-DPPC+Na] + was absent from the bronchioles and distributed heterogeneously throughout the parenchymal area. While this does not allow us to exclude the possibility of PC16:0/16:0 → PC16:0_20:4 metabolism within the bronchioles, the observation of [ 13 C 24 -PC16:0_20:4+Na] + provides direct evidence for PC16:0/16:0 → PC16:0_20:4 enzymatic synthesis within the lung parenchyma. The absence of 13 C 24 -PC16:0_20:4 and U 13 C-DPPC in the bronchial regions demonstrated the lack of surfactant uptake and PC metabolism by the bronchial epithelium. We found reasonable but not complete co-localization between   MS/MS data were obtained following accumulation of precursor ions over a larger sample area before fragmentation and detection. The results thereby present evidence for the specific acyl remodeling of surfactant DPPC into polyunsaturated PC species by alveolar macrophages in vivo.

DISCUSSION
For the first time we have combined isotopic labeling with MALDI-MSI at high mass resolving power to visualize: (i) uptake and distribution of therapeutic surfactants used to treat a variety of respiratory diseases; and (ii) the location of lipid synthesis and metabolic events throughout biological tissue. Injection of D 9 -choline into the mice enabled the detection and localization of newly synthesized PC lipids produce by the CDP:choline pathway by virtue of their expected m/ z shift of 9.0654 (Δ 9D-9H ). These metabolic events were found to be relatively homogenously distributed throughout the mouse lung tissue. The use of a U 13 C-DPPC label not only enabled the visualization of both CHF5633 and Poractant alfa porcine surfactant uptake independent of endogenous surfactant but also allowed detection of acyl chain remodeling events within the tissue. In particular, de-acylation/re-acylation of U 13 C-DPPC via the Lands cycle was observed, forming 13 C 24 -PC16:0_20:4 within the parenchyma. We note that while it is often assumed that this process is initiated by PLA 2 cleavage of the sn-2 fatty acyl, analogous PLA 1 cleavage (31,32), leading to the formation of 13 C 24 -PC20:4/16:0, is also possible. As a result, our approach not only provides lipid distributions, but the localization of lipid metabolic events within the tissue.
The metabolism of U 13 C-DPPC by alveolar macrophages in vivo, summarized in Fig. 5, provides strong circumstantial evidence for a major role for this cell type in the generation of 13 C 24 -PC16:0_20:4 by deacylation/re-acylation of DPPC. While the active reacylation of arachidonate into alveolar macrophage phospholipids has been reported previously (33,34), the results presented here are the first demonstration of a similar re-acylation that is specific for docosahexaenoate. If re-acylation of arachidonoate is indeed restricted to alveolar macrophages, this could explain the discrepancy of distribution of 13 C 24 -PC16:0_20:4 and U 13 C-DPPC shown in Fig. 6. This may possibly have been because of migration of alveolar macrophages within the lung parenchyma after initial uptake of the exogenous surfactant.
Surfactant catabolism and turnover have previously been studied both in vivo (35) and in isolated ATII cells (36) and alveolar macrophages (37), which are thought to be responsible for the turnover of surfactant phospholipid in healthy lung. However, there has been no demonstration of their relative contributions, as opposed to, for instance secretory phospholipase A 2 (38), to the decreased bronchoalveolar surfactant concentration in severe respiratory diseases such as acute respiratory distress syndrome. In contrast, the extensive accumulation of surfactant phospholipid and proteins in alveolar proteinosis is clearly because of impairment of granulocyte macrophage colony stimulating factor-dependent alveolar macrophage metabolism (39). MSI offers considerable potential to probe these mechanisms in increased detail, particularly as the sensitivity continues to improve, enabling the detection of even lower abundance labeled lipids and their metabolites (e.g., LPC species and D 3 /D 6labeleld PC species that may form via metabolism of D 9 -PC species). This detection sensitivity will be especially important if the relative roles of different individual cell types can be visualized, a possibility that is becoming feasible as current research aims to reduce the spatial resolution of MSI to the single-cell level (40,41).

Data availability
Mass spectrometry imaging datasets in the open source imzML format are available for download from the MetaboLights repository (42)