Introduction

Cardiolipin (CL) is a class of complex phospholipid exclusively located in the mitochondria of non-photosynthetic eukaryotic cells and is an integral part of the mitochondrial electron transport chain. The tight association of CL with cytochrome C (cyt C) is critical to the stability of respiratory chain supercomplexes [1, 2] and the genesis of the proton electrochemical potential across the inner mitochondrial membrane that drives the synthesis of adenosine triphosphate (ATP). Peroxidation of CL and its subsequent dissociation with bound cyt C together are both critical steps which lead to the eventual release of cyt C into the cytoplasm and trigger a cellular cascade resulting in apoptosis [2, 3]. Due to the high efficiency of mitochondria in chemical energy conversion, the density of this organelle in organ tissue generally reflects the energy demand for the underlying tasks. However, in organs with a seemingly uniform parenchyma, regions of functional difference may only be distinguished by the regionalized metabolic profiles. For example, the homogeneous liver parenchyma is divided into functional zones based on the predominant metabolic features in situ [4, 5]. This functional division may reflect differences in energy demands that control the underlying metabolic reactions. Such differences could be associated with local variations in the efficiency of ATP conversion by mitochondria. The fatty acyl composition of CLs on the mitochondrial membrane was thought to affect the ATP conversion efficiency [6]. Therefore, characterizing the in situ distribution of CL species likely demarcates the functional zones in those organs.

Mass spectrometry is one of the utilities used in the detection and characterization of CLs due to its high sensitivity and unparalleled accuracy in molecular identification. Previous studies have reported the analyses of CLs and their peroxidized products using electrospray ionization mass spectrometry [7–13]. CLs have also been detected and characterized using matrix-assisted laser desorption/ionization mass spectrometry (MALDI-MS) technique [14–19] coupled with thin-layer chromatography or other offline separation techniques. However, to date, only limited studies have reported direct detection of CLs from tissue sections [16, 20]. The low relative abundance of CLs compared with other phospholipids and the stringency in matrix selection for in situ ionization [16] are contributing factors to the fractional coverage of CL species by MALDI-MS in these direct detection studies.

Mass spectrometry imaging (MSI) is a technique that examines the two-dimensional distribution of molecular ions in tissue sections. This technique is well-suited to probe the distribution of wide variety of exogenous compounds and their metabolites in organs and whole-body sections [21–24]. In addition, MSI is used in proteomic [25–28] and lipidomic studies [29–32] to screen and identify biological markers in organ tissues that correspond to healthy and the diseased states. The combination of MSI and the conventional histological techniques extends the traditional morphological studies with advances in molecular sciences. Such a combination will provide new insights in biomedical research.

Given recent advances in sample processing for MALDI imaging mass spectrometry (MALDI-MSI) of phospholipid distribution in situ, we directly visualized CL distribution in rat tissue sections from organs with high mitochondria content. The improved tissue preparation and processing techniques enhanced the ionization efficiency of minor CL species and provided the matrix stability needed for MALDI-MSI of CL distribution in organ sections under a high vacuum environment. The distribution patterns of CLs in the organ sections corresponded to their respective histological features and further correlated with regional functions in the organs.

Material and methods

Chemicals

The MALDI matrix 2,5-dihydroxybenzoic acid (DHB) was purchased from Alfa Aesar (Lancashire, U.K.). ACS-grade ammonium acetate and poly l-lysine stock solution (0.1 %) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Cardiolipin standard from bovine heart was purchased from Avanti Polar Lipid Inc. (Cat. No. 840012P, Alabaster, AL, USA). HPLC-grade methanol and chloroform were purchased from Macron Fine Chemicals (Center Valley, PA, USA).

Animal use, tissue collection, and section processing

The use of laboratory animals followed the National Institutes of Health guideline on the humane use of laboratory animals. All animal protocols were approved by the Institutional Animal Care and Use Committee of National Sun Yat-Sen University.

Male Sprague–Dawley rats were group-housed (three rats per cage) in the local colony room under 12-h dark–light cycle and constant room temperature (25 °C). Standard laboratory rat chow (MFG Experimental Animal Feed Laboratory Animal Diet, Oriental Yeast Co. Ltd., Tokyo, Japan) and drinking water were provided to the animals ad libitum. Rats between 280 and 320 g were euthanized with a lethal overdose of isofluorane for fresh organ collection. The rats were rapidly decapitated after respiration ceased and muscle tone lost. Then, the heart, quadriceps femoris, and liver were rapidly dissected and excised. The organs were immersed in pre-chilled (−80 °C) isopentane for 10 to 15 s, retrieved, then rapidly wrapped in aluminum foil, and stored in a −80 °C freezer until cryosectioning. The isopentane container was stored at −80 °C until immediately before use. Previous study has demonstrated that this snap-freezing protocol did not cause any detectable loss of tissue lipids [33]. The frozen rat organs were trimmed, mounted on the specimen disk, and cut into 14-μm thick cryosections with a cryostat (Leica CM 3050, Nossloch, Germany). Cryosections were mounted on the indium-tin oxide (ITO)-coated slides (ITO slides; 25 mm × 75 mm; 80 Ω/cm). ITO-slides were pre-coated with 0.01 % poly l-lysine to enhance adhesion to tissue sections for further processing. A drip-wash step using 500–700 μL of 150 mM cold ammonium acetate solution (4 °C) to remove the interfering endogenous salts [33, 34] was applied to collected tissue sections. A brief evaluation of the poly l-lysine coating and desalting wash on MALDI ionization of phospholipids was carried out in positive and negative ion modes using rat cortex to ensure that no suppression of ionization to the in situ phospholipids had occurred. Immediately after the desalting wash, tissue sections were vacuum-dried for 20 min and sublimed with 0.1 g DHB at 0.26–0.28 Torr under 230 °C for 9 min [33]. For the tandem mass spectrometry study of the CL standard, purchased lipid pellets were dissolved in the mixture of methanol and chloroform (1:2; v/v) at 1 mg/mL, spotted on the ITO slides, and air-dried before matrix sublimation.

After MALDI-MSI experiments, the imaged tissue sections were stained with hematoxylin and eosin (H&E) and mounted with coverslips. The stained tissue sections were scanned with Aperio ScanScope CS digital slide scanner (Vista, CA, USA), and the images were saved into JPEG files for histological reference and image co-registration studies.

Mass spectrometer and data processing

An Autoflex III MALDI TOF/TOF mass spectrometer with a 355 nm Nd-YAG Smartbeam laser (Bruker Daltonics, Breman, Germany) operating at 200 Hz was used for profiling, tandem mass spectrometry (MS/MS) confirmation, and MALDI-MSI of CLs in tissue sections. The laser spot size was set at “medium” for the acquisition of CL molecular ions which generated a spot of 60–70 μm in diameter on the tissue section. The CL molecular ions were acquired using negative ion mode with reflectron. The voltages applied to Ion Source 1 and 2 were 19 and 16.85 kv, respectively. A 50-ns delay was used for the pulsed ion extraction. The laser fluence was set at slightly above the ionization threshold of CLs in tissue for all studies. The profiling spectrum of CL was summed from 300 consecutive laser shots. The MS/MS confirmation of CL was performed using LIFT mode, which is essentially the enhanced post-source decay mode, of the instrument where the precursor and product ions were collected from 1000 and 2000 consecutive laser shots, respectively. The mass selection window was opened from 12C m/z-0.5 Da to 12C m/z +1.4 Da of the selected CL species. The resolution of every peak in the MS/MS spectra was at least 1000 except the lowest m/z peak which exhibited a resolution of 800. During MS/MS confirmation, no collision cell was used. For MALDI-MSI studies, the tissue sections were raster-scanned at a lateral resolution of 100 μm for both X- and Y-axes. The profiling and MS/MS spectra were processed with FlexAnalysis (version 3.0; Bruker Daltonics) and exported by mMass (version 5.4.1; http://www.mmass.org) [35, 36]. The LIPID MAPS mass spectrometry database (http://www.lipidmaps.org) and the result of Tyurina et al. [11] were used for the identification of CL species. Naming of CL fragments followed that in previous studies [10, 16]. The ion density maps of CLs were extracted using FlexImaging (version 2.0, Bruker Daltonics) and saved into JPEG files. To illustrate the in situ distribution of liver CLs, the extracted CL ion density maps were co-registered with the scanned H&E stain image of the liver section. Manipulations of the exported images were limited to rotation and cropping of the JPEG files using GNU imaging manipulation program (GIMP; version 2.8.2; http://www.gimp.org).

Results

Direct profiling of myocardial cardiolipin

In general, a good MALDI-MS profiling of a particular molecular ion in situ constitutes the basis for successful MALDI-MSI studies of such molecule. However, when we directly profiled the rat heart section that was sublimed with DHB without any additional processing, the resulting spectrum resembled instrument noise and showed no detectable CL molecular ions. To remedy such lack of detectable CL molecular ion signals, the immediate adjacent rat heart section was drip-washed with 150 mM ammonium acetate solution and vacuum-dried before matrix sublimation. Figure 1 demonstrates the CL profiling result from the rat heart section after the desalting drip-wash. A single detectable molecular species at m/z 1447.97 without the associated sodiated or potassiated species was revealed by the desalting drip-wash. This result indicates the effectiveness of such tissue processing step in strengthening the ionization of in situ CL and simplifying the mass spectrometric presentation of this molecule. The inset of Fig. 1 shows the isotopic pattern of this CL ion. It appears that a desalting drip-wash is necessary for successful in situ CL profiling. Due to its effectiveness in enhancing the ionization of in situ CL, all the subsequent profiling and imaging studies would include the desalting drip-wash step in sample processing.

Fig. 1
figure 1

Negative ion mode MALDI-MS profiling spectra of cardiolipin (CL) in rat heart section. The spectrum was acquired from tissue section treated with desalting drip-wash using 500 μL of 150 mM ammonium acetate solution before matrix sublimation. Inset: Magnification of the m/z 1440–1460 range, showing the isotopic pattern of the in situ CL from rat heart

Full size image

MALDI-MS/MS confirmation of myocardial cardiolipin

To confirm the molecular identity of the m/z 1447.97 ion in Fig. 1b, a MALDI-MS/MS experiment was carried out to reveal the structural characteristics and the fatty acyl constituents of this molecular ion. Upper panel of Fig. 2 shows the tandem mass spectrum of the m/z 1447.9 ion from the rat heart section while the lower panel of Fig. 2 illustrates the tandem mass spectrum of the same m/z 1447.9 molecular ion from the commercial CL standard. Both MS/MS spectra show that only the m/z 695.5 ion species was generated as the [a/b]− fragment ion which was essentially the phosphatidic acid moiety of the CL precursor [10, 16]. This suggests that the phosphatidic acid (PA) moiety in the CL precursor consists of a single PA 36:4 species. Both the [a+56/b+56]− fragment ion at m/z 751.5 and the [a+136/b+136]− fragment ion at m/z 831.5 support the structural features of the m/z 1447.9 precursor. The fragment ion representing the loss of a fatty acyl moiety from the [a/b]− ion was also detected at m/z 415.2 only. Parallel formation of the m/z 433.3 fragment ion by the loss of a fatty acyl as ketene from the [a/b]− ion was also noted. Finally, only one fatty acyl moiety was observed at m/z 279.4 that corresponded to the (18:2) fatty acyl fragment. Based on the structural features noted above, and the matching of fragments from the commercial CL standard, the m/z 1447.9 CL in rat heart section was identified as the (18:2)4 CL.

Fig. 2
figure 2

Negative ion mode tandem mass spectrometry (MS/MS) of CL. a MS/MS spectrum of CL from rat heart section. b MS/MS spectrum of CL from the commercial standard. The fragment ion pattern of CL in rat heart section appears identical to the fragment ions from the commercial standard

Full size image

MALDI-MSI of CL in the myocardial and skeletal muscle sections of the rat

Figure 3a shows the MALDI-MSI result of the m/z 1447.9 molecular ion from the rat heart section. Figure 3b shows the image of m/z 1449.9, the second isotopic peak of the m/z 1447.9 CL from the same section. The walls of the ventricles and the left and right ventricular chambers (LV and RV respectively, Fig. 3b) are clearly delineated by the CL image when referenced to the H&E image of the same tissue section (Fig. 3c). The pseudo-color scheme of Fig. 3a and b indicates that the CL distribution in the ventricular myocardium is mostly homogeneous. The relative abundance of m/z 1447.9 and m/z 1449.9 CL ions largely agrees with the isotopic ratio of CL in the averaged mass spectrum (Fig. 3d). Furthermore, based on the relative abundance of the m/z 1447.9 and the m/z 1449.9 isotopic peaks of the detected CL molecular ion [16], it appears that only one species of CL was detected in the rat myocardia in Fig. 3d.

Fig. 3
figure 3

Negative ion mode MALDI-MSI of CL in rat heart section. a The ion image of m/z 1447.9, showing a largely homogeneous distribution of this ion across the section. b The ion image of m/z 1449.9. c H&E stain image of the rat heart section prepared after MALDI-MSI study. Bar = 1 mm. d The averaged spectrum from the MALDI-MSI experiment that generated (a and b), showing the isotopic pattern of CL in rat heart. No other CL clusters were observed

Full size image

In addition to myocardia, we also explored the CL distribution in a cross-section of quadriceps femoris, the skeletal muscle of anterior thigh. Figure 4a, b shows the molecular images of m/z 1447.9 and m/z 1449.9, respectively, in the cross-section of this muscle. In contrast to the myocardial section, the images of these two CL ions did not completely overlap with the H&E stain image of the tissue section (Fig. 4c). The pseudo-color scheme further indicates that the relative abundance of these two CL species does not differ as significantly as that in cardiac muscle. The averaged mass spectrum from the MALDI-MSI experiment (Fig. 4d) also supports this result and further suggests the presence of low abundant (18:2)3(18:1) CL in skeletal muscle with the (18:2)4 CL [16]. In spite of this difference, the averaged mass spectra in Figs. 3d and 4d suggest that the speciation of CL is relatively straightforward and uncomplicated in striated muscle.

Fig. 4
figure 4

Negative ion mode MALDI-MSI of CL in the cross-section of rat quadriceps femoris. a The ion image of m/z 1447.9, showing the distribution of CL in the cross-section. b The ion image of m/z 1449.9. c H&E stain image of the rat quadriceps femoris section prepared after MALDI-MSI study. Bar = 1 mm. d The averaged spectrum from the imaging experiment that generated (a and b). The elevated relative abundance of m/z 1449.9 suggests the presence of low abundant (18:2)3(18:1) CL species [16]

Full size image

MALDI-MSI of CL in rat liver

Unlike CLs in striated muscles, CL speciation and localization in rat liver appear more complicated. The averaged mass spectrum of liver CL imaging study in Fig. 5a indicates that liver CLs are detected in three mass ranges. The imaging results also indicate that CLs in each mass range share a distinctive distribution pattern. The first CL-containing mass range falls between m/z 1420 and 1430 and includes the m/z 1421.9 ((16:1)(18:2)3), 1423.9 ((16:0)(18:2)3), and 1425.9 ((16:1)(18:2)(18:1)2) CL species [11]. The in situ distribution pattern of these three CLs is represented by the image of m/z 1423.9 CL where the high-intensity areas surround the void regions and form elliptical circles in liver section (Fig. 5b). The second mass range contains detectable CL species between m/z 1445 and 1455, and includes the m/z 1447.9 ((18:2)4), 1449.9 ((18:1)(18:2)3), and 1451.9 ((18:1)2(18:2)2) CL species. The unique distribution pattern of these three CLs is represented by the image of m/z 1447.9 (Fig. 5c). The in situ distribution pattern of these three CLs exhibits a cord-like, continuous, yet somewhat irregular pattern. The third mass range with detectable CLs falls between m/z 1470 and 1480 and contains the m/z 1473.9 ((18:1)(18:2)2(20:4)), 1475.9 ((18:1)2(18:2)(20:4)), and 1477.9 ((18:1)3(20:4)) CL species [11]. These three CLs, like CLs in the previous two mass ranges, share a third distribution pattern represented by the image of m/z 1473.9 CL (Fig. 5d). The in situ distribution pattern of CLs in this mass range appears similar to that of CLs in the second mass range. However, the distribution boundary extends slightly beyond the territory of CLs in the second mass range. The small yet obvious dot-like low-intensity areas in the middle of most of the highest intensity patches in Fig. 5c are more differentiated in Fig 5d. This structure resembles a small-diameter hepatic vessel such as an arteriole in the hepatic portal tract. Figure 5e shows the H&E image of the liver section processed after the MSI study. For reference purpose, the images of m/z 1421.9, 1425.9, 1449.9, 1451.9, 1475.9, and 1477.9 CL are shown in panel a, b, c, d, e, and f, respectively, in Fig. S1 (Electronic supplementary material).

Fig. 5
figure 5

Negative ion mode MALDI-MSI of CLs in rat liver section. a The averaged spectrum from the MALDI-MSI of rat liver section, showing the three mass ranges containing the detected CL species. The left and right insets show the ×10 intensity magnification of CL peaks at the first and the third mass ranges, respectively. b The ion image of m/z 1423.9 CL that represents the distribution of CLs in the first mass range between m/z 1420 and 1430. c The ion image of m/z 1447.9 CL, a representation of CL distribution in the second mass range between m/z 1445 and 1455. d The ion image of m/z 1473.9 CL that exemplifies the CL distribution pattern in the third mass range between m/z 1470 and 1480. e H&E stain image of the liver section prepared after MALDI-MSI study. Bar = 1 mm

Full size image

In order to visualize the relationship of the above three distribution patterns of liver CLs, and to reveal the potential associations of these CLs to the functional zones of the hepatic acini, we co-registered the CL images in Fig. 5b–d using their respective pseudo-color scheme. Figure 6a shows the co-registration of CLs in the first (Fig. 5b) and the third (Fig. 5d) CL-containing mass range. The pseudo-color scheme shows that CLs in these two mass ranges are largely distributed in a complementary manner, since each of the pseudo-color remains mostly separate and the individual distribution patterns remain easily distinguishable in the co-registered image. In contrast, Fig. 6b demonstrates the result of co-localizing CLs in the second (Fig. 5c) and the third (Fig. 5d) CL-containing mass range. The highly overlapped pseudo-color scheme indicates that CLs in these two mass ranges are distributed mostly in the same location. Nevertheless, the reddish margin surrounding the high-intensity light blue or whitish areas confirms the previous observation that the distribution boundary of CLs in the third mass range would extend beyond the territory of CLs in the second mass range. To correlate the CL distribution with histological structure, Fig. 6c illustrates the co-registration of Fig. 6a with the H&E image in Fig. 5e. The superimposed image reveals small yet clearly visible areas encircled by the green CL signal. The morphological feature of these areas denotes the venule space in liver parenchyma resulting from sectioning across the centrilobular venule [4]. Figure 6d shows the co-registration of Fig. 6b and the histological image in Fig. 5e. The venule space discerned in Fig. 6c appears to locate in regions of lowest intensity in Fig. 6b, indicating that CLs in the second and third mass ranges are away from the venule space and complementary to the distribution of CLs in the first mass range. Together, it appears that CLs in the first mass range are seen in the perivenous zone of the hepatic acini. On the other hand, CLs in the second mass range are distributed in the periportal zone, whereas CLs in the third mass range are distributed in the periportal and the transitional zone of the hepatic acini [4]. For reference purpose, panels a and b in Fig. S2 (Electronic supplementary material) show the co-registration of Fig. 5b–d and of Fig. 5b–e, respectively.

Fig. 6
figure 6

Co-registration of liver CL images and the H&E stain image. a Co-registration of CL images in Fig. 5b and d, showing the distribution relationship of CLs in the first and the third mass ranges. b Co-registration of CL images in Fig. 5c and d, showing the distribution relationship of CLs in the second and the third mass ranges. c Co-registration of a and the H&E stain image in Fig. 5d. d Co-registration of b and the H&E stain image in Fig. 5d

Full size image

Discussion

In this study, we enhanced the ionization and detection of low abundant phospholipids such as CLs in situ by the combination of direct tissue desalting [34] and matrix sublimation [37–39]. As was mentioned previously, MALDI-MS profiling of CLs in situ can be achieved by using 2,6-dihydroxyacetaphenone (DHA) as the matrix [16]. However, DHA is easily sublimed under a high vacuum environment within 30 to 45 min, which renders it unsuitable for the MSI studies of CLs using the instrument equipped with a high-vacuum ion source such as the MALDI-TOF/TOF mass spectrometer used in this study. Importantly, we determined that the desalting wash was necessary (Fig. 1). We also tested other widely utilized MALDI matrices in phospholipids analysis such as 9-aminoacridine (9-AA), α-cyano-4-hydroxycinnamic acid (CHCA), and 1,5-diaminonaphthalene (DAN) for in situ CL imaging. We found that sublimation of 9-AA onto tissue section was quite difficult to accomplish even with elevated heating temperature and heating duration, perhaps due to its high melting point. Although it was possible to sublime CHCA and DAN onto tissue sections, the former would generate the tissue CL ion signals with significantly lower mass resolution and signal-to-noise ratio, while the latter would generate no CL ion signals. In our hands, the combination of tissue desalting wash and sublimation of DHB delivered the optimal MALDI-MSI results for CLs in situ.

In addition to in situ desalting and matrix sublimation, we also discovered that adding a thin layer of poly l-lysine as the adhesive on the conductive surface of the ITO slide provides additional benefits. Not only did this coating cause no visible interference to the ionization of phospholipids in tissue section, it also provides additional adhesive properties to the ITO slides to securely hold the tissue sections. Such advantages may be applicable to the in situ MALDI studies of peptides and proteins and may be particularly important in mass spectrometry studies of fragile tissue sections.

In the previous study, Eibish et al. has pointed out that phosphatidylcholine (PC) dimers could be easily misinterpreted as CL during MALDI-MS detection of phospholipids from hepatocyte [19]. Although the tissue sections for MALDI-MSI study contains PCs as well as CLs, the formation, however, of PC dimers was unlikely in our setting. First, the homo-dimers of the commonly encountered high abundant PC species in rats such as PC 32:0 and PC 34:1, if formed, should have been detected at m/z 1462.2 and m/z 1518.2, respectively, under negative ion mode. Similarly, the hetero-dimer of these two PCs would have resulted in a m/z 1492.2 peak. However, in the averaged spectra shown in Figs. 3d, 4d, and 5a, no detectable peaks were found between the m/z 1460 and m/z 1470 range, or in mass range above m/z 1490. Second, during the validation of poly l-lysine coating, we found that the negative ion mode MALDI-MS profiling of rat brain sections that contained high level of PCs did not produce any detectable monomer molecular ion of the most abundant PCs, let along the formation of PC dimers. Third, had the PC dimers been mistakenly identified as the reported CL species, the subsequent MS/MS confirmation would have generated rarely occurred short-chain fatty acyl moieties containing an odd number of carbons which were rarely seen in normal rats. However, all our MS/MS results indicated that the fatty acyl moieties of the presumed CL precursors contained even number of carbons and were at least 16 carbons in length. The observed [a/b]− fragments also accurately reflected the phosphatidic acid moieties in the CLs. Hence, it was unlikely that the PC dimers were mistakenly identified as the reported CL species.

Even though the MS/MS study was able to reveal the fatty acyl residues of the CL ions, it was difficult to discern the stereo-specific positions of the revealed fatty acyl moieties on the CL molecule using this MS/MS technique from a tissue section containing essentially no alkali metal ions after desalting. Perhaps the alternative technique such as ion mobility mass spectrometry [40] that couples in-line separation and mass detection will be a more suitable tool to separate the regio-isomers of the CLs.

From the averaged mass spectrum of liver cardiolipins (Fig. 5a), it was apparent that the relative abundance of CLs in the three CL-containing mass ranges differed quite significantly. Such difference was most likely due to the actual discrepancy in the relative abundance of CL species in the liver section. However, in a previous MALDI-MS quantification of diacylglycerols (DAG), the impacts of the fatty acyl length and the double bond contents of the acyl moieties were also briefly explored [41]. Since each CL molecule contains four fatty acyl moieties, and some of them bear different extent of saturation, the same factors that influenced the ionization efficiency of DAG may also influence the ionization of the in situ CLs and further impact their mass spectrometric presentation.

In this study, we have confirmed the simple speciation of (18:2)4 CL in rat heart muscle as reported previously [13, 16]. Also, homogeneous distribution of this (18:2)4 CL in rat myocardial section echoed the observation in a previous study that reported a homogeneous distribution of mitochondria in rat heart muscle [42]. However, in skeletal muscle, we observed a slightly different speciation results from the previous study [16]. Such a minor discrepancy could be attributed to the enhancement in ionization of in situ CL adapted in this study which permitted the revelation of such minute differences. Alternatively, the cumulative difference in the composition of dietary fatty acids between the current and previous study could also influence the fatty acyl composition of CLs [6, 43] and contributed to this small difference. Furthermore, such a minor difference in the CL speciation was also seen in the second mass range of liver CL. Specifically, the relative abundance of m/z 1449.9 seen in this study was higher than that of m/z 1447.9, whereas in the previous study the relative abundance of these two species was identical [16]. In spite of the cause leading to the observed discrepancy in isotopic pattern of CLs, the results reported here support the combination of matrix sublimation and tissue desalting methodology as a means to enhance the ionization efficiency of low abundant CLs in tissue.

It was interesting to note that area containing CL ion signals in the cross section of skeletal muscle was not in full agreement with the H&E tissue stain image. Skeletal muscle fibers are divided into red, intermediate, and white muscle fibers based on the myoglobin content and mitochondrial number [44]. However, the ratio of these fibers in muscle will determine the predominant physiological characteristics of the entire muscle. A muscle cross section showing no detectable CL suggests its heavy reliance on glycolysis as the primary means of ATP production, a feature of fast-twitching yet fatigue-sensitive white muscle fibers, rather than through oxidative phosphorylation typical of red muscle fibers [4, 44]. Nevertheless, it may be far-fetched to distinguish the slow-twitching red muscle fibers from the fast-twitching yet fatigue-resistant intermediate muscle fibers merely by the presence of CL, since both heavily rely on mitochondria as the primary source of ATP.

Functional zones within the liver parenchyma based on metabolic heterogeneity of carbohydrate [45], amino acid, and ammonia [46], and even fatty acids [47], has been well-documented. The liver parenchyma could be roughly divided into the periportal, transitional, and perivenous zones based on the distance to the supplying branch of hepatic portal arteriole and the consequential difference in oxygen tension along the hepatic sinusoids [5, 48]. The periportal zone predominantly contains enzymes for oxidative reactions and carries out functions such as energy metabolism, beta-oxidation of fatty acids, amino acid catabolism, ureagenesis from amino acid, gluconeogenesis, and oxidation protection [49], whereas the perivenous zone, due to its close proximity to the terminal hepatic venule, contains high amounts of esterases, carries little or no oxidative capacity, and performs functions like conjugation, transformation, and detoxification. The transitional zone, on the other hand, features a mixed metabolic feature of the periportal and perivenous zones [4]. The distribution of local metabolic products such as phospholipids and the catabolic enzymes could correspond to the functional zonation in the liver. Indeed, previous studies have reported the zonation of liver parenchyma based on phospholipids and triglycerides in the regenerated murine liver [50] and changes in such zonation in human livers after steatosis and nonalcoholic steatohepatitis [51]. Based on our current results, regions containing CLs in the second mass range, including the (18:2)4 CL, bears histological and metabolic resemblance to the periportal zone. The region containing CLs in the first mass range would represent the perivenous zone in liver parenchyma, since the fatty acyl composition of such CLs was less ideal for the mitochondrial oxidative functions [6]. Regions containing CL in the third mass range, especially the narrow area beyond the distribution boundary of CLs in the second mass range, likely reflects the transitional zone that features mixed metabolic capacities [4]. We expect that such zonation in liver parenchyma based on the distribution pattern of CLs will be altered by acute and chronic metabolic abnormalities and oxidative stress.

In conclusion, the combination of the in situ desalting drip-wash and the sublimation of DHB successfully enhances the ionization efficiency and permits the observation and imaging of low abundant CLs in situ by MALDI-MSI. CL imaging in heart, skeletal muscle, and liver sections permits the exploration of metabolically critical machineries and further extends the boundary of metabolomic and lipidomic research.