Formal Lithium Fixation Improves Direct Analysis of Lipids in Tissue by

Article pubs.acs.org/ac

Formal Lithium Fixation Improves Direct Analysis of Lipids in Tissue by Mass Spectrometry Rian L. Griffiths,† Joscelyn Sarsby,†,‡ Emily J. Guggenheim,‡ Alan M. Race,†,‡ Rory T. Steven,†,‡ Janine Fear,§ Patricia F. Lalor,§ and Josephine Bunch*,†,‡ †

School of Chemistry, University of Birmingham, Edgbaston, Birmingham B15 2TT, United Kingdom PSIBS Centre, University of Birmingham, Edgbaston, Birmingham B15 2TT, United Kingdom § Centre for Liver Research and NIHR BRU, School of Immunity and Infection, University of Birmingham, Edgbaston, Birmingham B15 2TT, United Kingdom ‡

S Supporting Information *

ABSTRACT: Mass spectrometry imaging is a powerful method for imaging and in situ characterization of lipids in thin tissue sections. Structural elucidation of lipids is often achieved via collision induced dissociation, and lithium−lipid adducts have been widely reported as providing the most structurally informative fragment ions. We present a method for the incorporation of lithium salts into tissue imaging experiments via fixation of samples in formal lithium solutions. The method is suitable for preparation of single tissue sections, or as an immersion fixation method for whole tissue blocks or organs prior to sectioning. We compare lithium adduct detection and MALDI-MSI of murine brain from analysis of tissues prepared in different ways. Tissues prepared in formal solutions containing lithium or sodium salts before coating in matrix via air-spray deposition are compared with fresh samples coated in lithium-doped matrix preparations by either dry-coating or air-spray deposition. Sample preparation via fixation in formal lithium is shown to yield the highest quality images of lithium adducts, resulting in acquisition of more informative product ion spectra in MALDI MS/MS profiling and imaging experiments. Finally, the compatibility of formal lithium solutions with standard histological staining protocols (hemotoxylin and eosin, Van Giessen and Oil Red O) is demonstrated in a study of human liver tissue.

F

acid to enhance nuclear detail in light microscopy. Alternatives based on zinc have also proved successful for improved tissue contrast for light microscopy7 and increased staining density of cell nuclei.8 Contrast of tissue or cell features can be amplified for electron microscopy by staining with heavy metals which efficiently scatter electrons.9,10 Suitable stains should preferentially bind a particular structural constituent or species.10 Thus, heavy metals such as osmium (OsO4),11,12 lead,13 and uranium14 have all effectively been included in fixation and staining preparations for electron microscopy. Many are described in targeted approaches for a specific analyte. Osmium tetroxide stains for unsaturated lipids,15 ruthenium for polymers,16 lead hydroxide for RNA containing cell components,17 and barium hydroxide for the golgi regions of hepatic cells.17 Formal solutions can also be used as primary fixatives prior to other treatment to improve a particular analysis.18 Clearly careful consideration of the analyte of interest is important when choosing a staining or fixation method, and it

ixation of tissue samples is important to prevent proteolytic degradation,1 stabilize fine structure, and obstruct bacterial growth. Formaldehyde is a noncoagulant fixative which fixes tissue via the formation of cross-linking methylene bridges between neutral amino (or similar) groups in proteins.2,3 The chemical process of formaldehyde fixation is slow.4 Complete fixation is limited by the rate of chemical reaction, and so tissue must be immersed long enough for cross-linking to occur.4,5 Formaldehyde is generally used as a 4% solution in water. This solution of formaldehyde is known commercially as formalin. Formalin is widely reported to be compatible with histological staining and is extensively used. With a view to tissue storage, formalin is often used as an isotonic solution such as formal saline (sodium chloride in formalin). The isotonic nature of such formal solutions limits diffusion of tissue components; however, formal saline is mildly acidic. For this reason some laboratories prefer to use sodium phosphate salts to neutrally buffer the solution. Both preparations involve sodium salts to maintain physiological conditions.6 A number of other salts have been included in formalin (formal solutions) with a view to improving tissue studies by analytical techniques such as microscopy. Mercuric chloride has been included with formaldehyde in combination with acetic © 2013 American Chemical Society

Received: March 11, 2013 Accepted: June 19, 2013 Published: June 19, 2013 7146

dx.doi.org/10.1021/ac400737z | Anal. Chem. 2013, 85, 7146−7153

Analytical Chemistry

Article

is not unusual to use numerous fixation techniques for multiple analyses of a single tissue sample.4 Mass spectrometry imaging offers analysts the opportunity to simultaneously acquire information regarding the sample composition and the spatial distribution of analytes, with no requirement for radiolabeling and with minimal sample preparation. Belazi et al. utilized osmium tetroxide staining for targeted analysis of lipids by time-of-flight secondary ion mass spectrometry (ToF-SIMS);19 hence, it is possible to direct mass spectrometry analysis using stains and fixatives. Formaldehyde does not act as a fixative for lipids and is therefore a suitable medium to consider for their analysis.20,21 MALDI-MS analyses of tissue fixed in formal saline or phosphate buffered formalin have been shown to lead to the detection of a high abundance of sodium adducts of lipids.22,23 Identification of species detected at a given m/z can be determined by tandem mass spectrometry (MS/MS) in which a selected ion is fragmented and then product (fragment) ions are detected. Complete structural identification is achieved by dissociation at structurally informative sites. Complete assignment of phospholipid structures, for example, requires product ions or neutral losses indicative of the headgroup and both fatty acid side chains. Phospholipid headgroups, such as phosphocholine (PC), are indicated by either the neutral loss of 183 from the intact (parent) lipid species or the detection of the charged phosphocholine headgroup at m/z 184. Similarly, neutral loss of a fatty acid or detection of the acyl fragment (RCO+) would be indicative of the nature of the side chain composition. Lipids form protonated [M + H]+ and cationic adducts such as [M + Na]+ and [M + K]+. Dissociation studies of different adducts have shown that product ions indicative of these fatty acid residues are most easily detected during dissociation of lithium−lipid adducts.24−32 Furthermore, the sn-1 and sn-2 side chains can be assigned; the fragment ion indicating neutral losss of the sn-1 fatty acid will be detected in greater abundance.31,33 This becomes particularly important when identifying potential disease biomarkers in complex tissue samples by mass spectrometry.34−38 Lithium salts are not naturally abundant in tissue, and so, the task of lithium ion introduction arises. To date this has been accomplished by many groups simply via pipetting a solution of a matrix/lithium salt mixture onto tissue for in situ analysis.26,39 However, analyte delocalization is a concern with this approach; hence, characterization of lipids from a single specific anatomical region is unrealistic. This can be overcome by including a lithium salt in the matrix during sample preparation. Cerruti et al. introduced lithium during the matrix application step via an automated deposition spraying technique;24 however, problems with nozzle blocking were experienced. We present formal lithium as a fixative solution in a targeted approach for the analysis of lipid species by mass spectrometry techniques. Formation of lithium−lipid adducts in situ by this route allows the analyst to acquire compositional information that can be mapped in ion images. Thus, we can show the spatial distribution of a particular lipid within tissue, as well as enhance their identification by collision induced dissociation. Using this sample preparation technique we were also able to acquire MS/MS images of lipids in which structurally characteristic product ions of a particular lipid can be mapped, showing the spatial location of a single identifiable analyte. The compatibility of formal lithium fixative with a number of common histological staining techniques was established using

sections of human liver from normal and diseased specimens, thus highlighting the potential of this approach for biomarker identification.

■

EXPERIMENTAL SECTION Materials. All salts (NaCl, LiCl, and LiNO3), α-cyano-4hydroxycinnamic acid matrix (CHCA), and formic acid (FA) were purchased from Sigma Aldrich (Gillingham, U.K.). Formaldehyde was purchased from Adams Healthcare (Swindon, U.K.). Methanol (HPLC grade) was purchased from Fisher Scientific (Leicestershire, U.K.). Water was purified by an ELGA Option 3 system (Marlow, U.K.). Hematoxylin and eosin (H&E), Van Gieson, Oil Red O and Celestine Blue stains, Scott’s tap water substitute, acid alcohol, and alcohol were purchased from Leica Biosystems (U.K.). DPX mountant (a mix of distyrene, plasticizer, and xylene), acetone, and Immu-Mount were all purchased from Thermo Scientific (Shandon, U.K.). Stainless steel MALDI target plates were purchased from AB Sciex (Framingham, MA), conductive ITO coated glass slides from Bruker Daltonik (Bremen, Germany), and plain glass and polylysine coated glass slides from Leica (U.K.). Human Tissue Samples. Explanted liver tissue from patients being transplanted for chronic end stage liver disease (nonalcoholic steatohepatitis, NASH) were obtained from the Queen Elizabeth Hospital, Birmingham, U.K. Fresh tissue was cut into approximately 1 cm3 pieces and immediately frozen by immersion in liquid nitrogen. Samples were then stored at −80 °C until cryosectioning was performed. All tissue was collected with informed written consent under local ethics committee approval. Fixative Solutions. Formal saline was prepared according to standard protocols (0.154 M NaCl in formalin) where formalin is a solution of 10% (by volume) formaldehyde solution (40% in methanol) in water (90% volume) resulting in a solution which contains 4% formaldehyde. Formal lithium solutions were prepared similarly (0.154 M LiCl or LiNO3 salt in formalin). Tissue Block Fixation. A snap frozen whole rat brain was sectioned into four. Each was treated differently; the first remained frozen and was stored at −80 °C, and the rest were immersion fixed in either formal saline or formal lithium (LiCl or LiNO3) at room temperature. Immersion fixation of tissue blocks was achieved by immersing tissue blocks in formal saline or formal lithium (LiCl or LiNO3 salt) in a volume 20× their mass for 48 h. Once fixed, tissue blocks were stored at −80 °C until sectioning at −18 °C [Leica CM 1850 Cryostat (Milton Keynes, U.K.)] at 12 μm before thaw mounting onto a stainless steel MALDI imaging target or plain glass slides for LESA−ESIMS or at 6 μm before thaw mounted onto polylysine coated glass slides for staining. Freezing of fixed tissue leads to greater ease when preparing tissue blocks for sectioning. Single Section Fixation. Snap frozen mouse brain tissue was sectioned differently for mass spectrometry (12 μm) and staining (6 μm) [Leica CM 1850 Cryostat (Milton Keynes, U.K.)]. Human liver tissue blocks were sectioned 6 μm thick [Bright OTF (Cambridgeshire, U.K.)]. Tissue sections were thaw mounted onto ITO coated glass slides for analysis on a ToF−ToF instrument, plain glass slides for LESA−ESI analysis, or polylysine coated glass for staining. Slides were then submerged in a fixative solution for 30 min. Control (fresh) sections were not treated with a fixative. 7147



Article

Staining. Serial tissue sections to those used for mass spectrometry were treated by fixation in a formal solution or not fixed (control fresh tissue) and stained. Groups of serially treated tissue sections were stained with H&E, Van Gieson, or Oil Red O according to standard protocols. A Ziess Axioskop 40 microscope, 10× objective, was used to obtain microscopic images using Ziess AxioVision software. Matrix Application. Matrix solutions of CHCA (20 mg/ mL, 80% MeOH) with or without the inclusion of 5−20 mM LiCl were deposited using an artist airbrush purchased from Draper (Hampshire, U.K.) propelled by dry N2. Two consecutive spray passes were followed by 10 s drying time, until 10 mL was deposited, from a distance of 20 cm from the plate. For dry-coating,40 CHCA matrix (2 g) was ground in a pestle and mortar with or without 5 mM equivalent LiCl or LiNO3 for 10 min. Matrix samples were deposited onto the MALDI target plate through a 20 μm mesh sieve. Mass Spectrometry. MALDI-MSI and MS/MS imaging studies of murine samples were carried out on a QSTAR XL QqTOF mass spectrometer (AB Sciex). An Elforlight (Daventry, U.K.) Nd:YVO4 DPSS laser (355 nm) was triggered by a Thurlby Thandar Instruments (Huntingdon, Cambridgeshire) TGP110 10 MHz pulse generator, coupled to the MALDI source via a 100 μm core diameter fiber optic patchcord (Edmund Optics, NA = 0.22) and operated at 5 kHz and approximately 8 μJ. Analysis was performed in positive ion mode with a pixel size of 100 by 100 μm2, m/z range 400−900, focusing potential (FP) of 85, and declustering potential (DP2) of 35. For MS/MS imaging CID was performed with N2 gas at collision energy of 40. All images were collected in raster mode at a speed of 1 mm s−1 and acquired in ≈25 min, except matrixlithium chloride samples prepared by airspray. For these the raster speed was decreased to 0.2 mm s‑1 leading to an acquisition time of 100 min. MALDI-MSI of human liver tissue was acquired on an ultrafleXtreme ToF−ToF mass spectrometer (Bruker Daltronics) with a smartbeam-II Nd:YAG (355 nm) laser operated at 1 kHz, and laser energy was optimized between 35% and 55% of the available power. Analysis was performed in reflectron ToF and positive ion mode with a pixel size 100 by 100 μm2, m/z region 60−1000. Images were acquired in spot imaging mode summing 500 laser shots per pixel. Each full tissue section image was acquired in ≈5 h. Electrospray analysis was performed on an LTQ Orbitrap Velos from Thermo Fisher Scientific (Leicestershire, U.K.), and tissue sections were surface sampled by LESA (Advian Ithaca, NY) with solvents composed of 70:30:0.1 MeOH:H2O:FA. Further details are provided in the Supporting Information. Data Conversion and Analysis. MALDI data acquired using the QSTAR XL was analyzed using Analyst QS 1.1 and MATLAB. The data were converted from the AB Sciex .wiff proprietary file format to mzML using AB MS Data Converter (AB Sciex version 1.3) and then converted to imzML using imzMLConverter41 and processed in MATLAB [version 7.8.0 (2009a), Math Works Inc.]. For all images, displayed peak information is the summed ion intensity from within a 0.1 Da window centered on the peak of interest. MALDI data acquired using the ultrafleXtreme was analyzed using MATLAB. Data were converted from the Bruker proprietary format to mzML using a custom MATLAB script and then converted to imzML using imzMLConverter. LESA data were converted to mzML using msconvert as part of ProteoWizard.42 Spectra acquired from a single injection were summed using a custom MATLAB

script and output to mzML. The summed spectra were then converted to imzML using imzMLConverter41 and processed in MATLAB.

■

RESULTS AND DISCUSSION Conventional Approaches to Additive Incorporation for MALDI-MS Imaging. Conventional methods for lithium incorporation into tissue samples for MALDI are based on inclusion in the matrix. Lithium chloride was added to CHCA matrix solutions prior to airspray deposition or ground with CHCA matrix before dry-coating deposition. A number of problems were encountered by these approaches. Airspray deposition of matrix−lithium additive solutions led to clogging of the sprayer nozzle at additive concentrations which did not lead to the detection of abundant lithium-lipid adducts (see Figure S1). Similar issues were described by Cerruti et al. even though they deposited matrix−additive solutions via a heated nozzle.24 They also comment upon optimization of additive concentration, and the formation of “elongated clusters” and compact crystal structures when including some lithium salts. These are undesirable as they can cause analyte delocalization or require higher laser fluences for ablation. Spraying of matrix−additive solutions also led to the formation of crystals which could not be as readily ablated by the Nd:YVO4 diode pumped solid state (DPSS) laser employed in this study, when compared to solutions of matrix only. This particular laser delivers a relatively low energy per pulse at a higher repetition rate compared to traditional nitrogen lasers.43,44 In order to analyze these samples the interrogation time (which dictates the number of shots delivered in a particular location) needed to be increased in order to ablate the sample crystals, leading to much lengthier acquisition times. Inclusion of lithium chloride in CHCA matrix for dry-coating preparations also proved troublesome. Grinding the matrix and lithium salt together led to the formation of agglomerates, probably owing to the hygroscopic nature of lithium chloride, which was unsuitable for this deposition technique. A suitable sample was formed when lithium nitrate was ground with CHCA matrix; however, deposition of this led to the detection of relatively high abundances of protonated adducts alongside lithium and other cationic adducts compared to analysis of the same tissue using the airspray method (see Figure S1). Relatively high abundances of protonated alongside cationic lipid adducts have been reported previously using solvent free methods.45 Even though solvent free methods have proven useful for analytes which predominantly form protonated adducts,46 lipid analysis was only further complicated. MALDI-MS Imaging Analysis of Formal Preparations. Preparation of formal saline according to the standard protocol (0.154 M in formalin) substituting sodium chloride with a lithium salt did not alter the final pH of the solution. Furthermore, no lipids were detected by MALDI-MS analysis of the fixative solutions post fixation of tissue, indicating that there was no lipid migration into the supernatant solution. Immersing tissue blocks in a formal lithium fixative solution for a 48 h period was sufficient to allow full penetration through tissue and led to similar indicative discoloration of the tissue as that using formal saline. Sectioning of tissue fixed in formal lithium did not appear to differ from the standard formal saline preparation, and thaw mounting of the tissue sections onto either stainless steel MALDI target plates or plain or coated glass slides was unaffected by treatment in this fixative. 7148



Article

Preparing formal lithium fixed tissue samples for imaging analysis by the widely employed deposition method of airspray, with CHCA matrix solutions, led to similar matrix crystal homogeneity as when spraying fresh frozen and formal saline fixed tissue sections. Introducing lithium during tissue fixation removes the need for salt inclusion in MALDI matrix solutions during the deposition step; the crystal homogeneity is unchanged by formal preparations, and hence, many of the problems described above and by Cerutti et al.24 are eliminated by immersion fixation. Furthermore, these images were acquired with 500 laser shots per pixel, which is similar to the conditions of Cerruti’s experiments (300); however, continual raster methods, with a stage speed of 1 mm s−1, enabled relatively high speed imaging of lithium adducts. Hence, our methodology lends itself to high throughput analysis. Lipid species detected in formal saline fixed tissue were similar to those detected in fresh frozen tissue. Predominantly sodium−lipid adducts were detected when analyzing fixed tissue prepared and/or buffered with a sodium salt, as reported by Carter et al.22 The three most abundant lipid peaks in the m/z region 700−900 in fresh frozen tissue analysis are m/z 772, 798, and 826, [M + K]+ of PC 32:0, 34:1, and 36:1, respectively. The most abundant peaks in the lipid region shift to m/z 756, 782, and 810 in formal saline fixed tissue, indicating a change in predominant adduct formation to [M + Na]+ lipid adducts, see Figure 1. The lipid species detected in formal lithium fixed tissue were similar to those detected in fresh frozen and formal saline samples, with a mass shift of Δ − 32 compared to fresh frozen ([M + K]+ to [M + Li]+). When analyzing formal lithium fixed tissue samples the most abundant peaks detected in the region m/z 700−900 were 740, 766, and 794, [M + Li]+ of PC 32:0, 34:1, and 36:1, respectively (see Figure 1), similar species to those reported by Cerutti et al. by introduction of lithium during matrix deposition.24 It would appear that fixing tissue in an isotonic lithium salt fixative solution leads to a shift in predominant adduct formation similar to that previously reported in isotonic or buffered sodium salt fixative solutions. Mean spectra of species detected in tissue during imaging analysis are shown in Figure 1. It is clear from this figure just how powerful this preparation is for forming the predominant lipid adducts detected. Further adduct confirmation can be achieved via MS/MS analysis as the headgroup product ion changes corresponding to the lipid adduct dissociated.27 The spectral quality of formal lithium fixed samples is similar to that of fresh frozen and formal saline fixed tissue. There did not appear to be a significant change in the signal-to-noise by preparing the tissue using this method, and the ion counts of detected lipid species were of a similar order of magnitude. In short, introducing lithium by formal lithium fixation does not appear to compromise spectral quality in any way. Nitrate salts have recently been reported to increase ion counts of a desired lipid adduct in MALDI.30 However, substitution of lithium chloride for lithium nitrate did not lead to a significant increase in ion counts of detected species. Results indicate that this method is tolerant of different salt types and differences in anatomical localization of specific lipid species can be identified equally well with both fixatives, see Figure S2. This has been described for a number of other formal preparations, for example, with formal zinc solutions prepared with zinc salicylate7 or both acetate and chloride salts.8

Figure 1. Mean spectra of species detected between m/z 700 and 900 on tissue during acquisition of MALDI-MS images. (A) Fresh frozen tissue. (B) Formal saline fixed (NaCl). (C) Formal lithium fixed (LiCl). Four lipid species are highlighted in the spectra showing the lipid-adducts detected by each tissue preparation: diamonds show sphingomyelin (SM) 36:1, circles indicate phosphatidylcholine (PC) 32:0, stars denote phosphatidylcholine (PC) 34:1, and triangles represent phosphatidylcholine (PC) 36:1. Potassium adducts are colored green, sodium adducts are purple, and lithium adducts are red and outlined. The spectral quality and species detected by each sample preparation do not change, only the dominant lipid adduct.

In Situ Structural Characterization of Lipids and MALDI-MS/MS Image Acquisition. A particular advantage of driving ionization toward lithium adduct formation is the improved structural characterization afforded by the dissociation of these adducts compared to sodium and potassium−lipid adducts.26,28,47,48 Lipid profiling of different lipid adducts in situ shows that dissociation of either the potassium, sodium, or lithium lipid adduct allows headgroup assignment [neutral loss of 59 (choline) and neutral loss of 183 (phosphocholine headgroup)]. However, more product ions are detected in high relative abundance which inform the fatty acid side chain identities when lithium lipid adducts are dissociated in formal lithium fixed tissue compared to potassium adducts detected in fresh frozen tissue or sodium adducts in formal saline fixed tissue (see Figures S3−5). The ability to dissociate highly abundant lithium lipid ions in situ enables structural assignment via detection of the following types of product ions: neutral loss of the fatty acid, neutral loss of the fatty acid lithium salt, and neutral loss of the fatty acid alongside the neutral loss of choline. The detection of three product ions in the region m/z 340−540 indicates that the two fatty acid side chains are the same (Figure S3), six product ions indicate that the two fatty acid side chains are different (Figure S4), and 12 product ions indicate that there are two isobaric lipids with two sets of different fatty acid side chains (Figure 7149



Article

Figure 2. (A) MALDI-MS ion images of lithium adducts of three different PC lipids detected in formal lithium fixed rat brain tissue and an overlay of all three ion images, yellow m/z 812 (PC 38:6), blue m/z 794 (PC 36:1), and purple m/z 740 (PC 32:0). Ion image areas of no signal intensity are shown in black. All images are normalized to TIC. (B) MALDI-MS/MS of PC 32:0 [M + Li]+. Single ion images of m/z 551 (neutral loss of 189) and 425 (neutral loss of palmitic acid and choline) and the mean spectrum of ions detected on tissue are shown. A schematic illustrating dissociation sites is shown above the spectrum (annotation of NL is neutral loss). Ion image areas of high signal intensity are shown in pink. The potential to spatially map distributions of a single lipid using MS/MS imaging with greater confidence than may be possible with MS imaging is shown.

S5). Ions corresponding to loss of the sn-1 fatty acid are detected in greater abundance than those of the sn-2 loss;26,28 hence, structural assignment is possible. In situ profiling of formal lithium fixed tissue led to the detection of peaks revealing side chain identities in relatively high abundance, enabling lipid characterization. Exemplary spatial distributions of three different PC lipid species detected in formal lithium fixed tissue by MALDI-MS imaging, and an overlay of their distributions in the cerebellum, are shown in Figure 2A. Further to this, MS/MS imaging of lithium fixed tissue sections enables generation of ion images showing the spatial distribution of structurally characteristic product ions. The mean spectrum acquired from tissue regions during MS/ MS imaging of one of these lipids, m/z 740 (PC 32:0 [M + Li]+), is shown in Figure 2B. The mean product ion spectrum from the imaging experiment clearly shows structurally characteristic peaks. Neutral loss of 59 and 183 reveals that this species is a phosphocholine lipid. Neutral loss of 189 indicates loss of a lithium adduct of the phosphocholine headgroup. Of greatest interest are the peaks which indicate the identity of the fatty acid residues. The peak at m/z 239 is indicative of the acyl fragment (RCO+) of palmitic acid (16:0). Product ions detected at m/z 484, 478, and 425 further support this assignment as they can be attributed to the neutral loss of palmitic acid (256), neutral loss of the lithium salt of palmitic acid (262), and neutral loss of palmitic acid alongside choline (315), respectively, as previously described.26 Therefore, it is possible to assign this species as PC 16:0/16:0. Mapping the spatial distribution of the peak at m/z 551 clearly shows that this lipid is only detected in gray matter. Mapping of the neutral loss of palmitic acid and choline at m/z 425 further confirms that this lipid is only present in gray matter in the cerebellum. These spatial distributions directly correlate and concur with the distributions in the original data set. Therefore, it is possible to map ion images of structurally characteristic product ions by preparing tissue in a formal lithium fixative. Moreover, this data was acquired by the same rapid acquisition parameters described for MS imaging. The mean spectrum acquired from tissue regions during MS/ MS imaging of m/z 766 (PC 34:1 [M + Li]+) shows a greater number of fatty acid side chain indicative product ions than that of PC 32:0. Six product ions in the region m/z 340−540

indicate that this phosphocholine lipid has two different fatty acid side chains, see Figure S7. Mapping the spatial distribution of the most abundant product ion indicative of each fatty acid side chain, m/z 451 [neutral loss of palmitic acid alongside choline (315)] and 425 [neutral loss of 18:1 fatty acid alongside choline (315)], shows that this lipid is homogeneously distributed in the cerebellum. The sensitivity of the MS/MS imaging method is such that the spatial distribution of this lipid species can be mapped by a single product ion relating to either the sn-1 or sn-2 fatty acid side chain. MS/MS imaging is widely documented for other analytes,49−52 and recent studies have considered sodium adducts of fatty acids and triacylglycerols in fingerprints,53 phospholipids in murine brain,54 and also mapping the spatial distribution of isobaric lipids with different headgroups.55 The ability to perform MS/MS imaging from lithium lipid adducts shows real opportunity for in situ tissue analysis of a range of lipid species. Liquid Extraction Surface Analysis−Electrospray Ionization (LESA−ESI). LESA coupled to electrospray ionization (ESI)56 and a high resolution mass spectrometer allowed further identification of lipids in formal lithium fixed specimens. Lipid species were preferentially detected as potassium adducts in fresh tissue sections with less abundant peaks relating to protonated and sodium adducts. In spectra collected from formal saline fixed tissues, sodium adducts were the most abundant adducts detected. Protonated and potassium adducts were detected in much lower relative abundance. Lithium adducts were detected in relatively high abundance in lithium fixed tissues; less abundant peaks relating to protonated, potassium, or sodium adducts were found to be significantly less abundant in comparison to fresh tissue sections. Numerous lipids were detected in all tissue preparations, readily identified by accurate mass (see Figures S8−11). Dissociation of lithium adducts of a number of different lipid classes have previously been reported to improve characterization by electrospray ionization. The assignment of the fatty acid residues of triacylglycerol lipids is reportedly enriched by lithium adduction.57 Furthermore, double bond assignment of unsaturated fatty acids has also been enhanced by dissociation of lithium−lipid adducts.58 Analysis of formal lithium fixed tissue by electrospray also led to the detection of lithium−lipid adducts. Collision induced dissociation of select peaks showed that lithium adducts provide superior structural information 7150



Article

Currently, diagnosis depends on the routine fixation of tissue with neutral buffered formalin and subsequent histopathological analysis using light microscopy. However, with the additional molecular information available from techniques such as MALDI-MSI, the integration of MS techniques is desirable, and standard protocols suitable for use with histopathology and MS imaging could be beneficial for histology driven studies.59−62 The effect of formal metal fixation on common staining techniques (H&E, Van Gieson, and Oil Red O) was considered. Fresh tissue, formal saline, and formal lithium prepared with either lithium chloride or nitrate salts were assessed. NASH liver tissue is characterized by chronic inflammation steatosis and extensive fibrosis, which causes major disruptions to the architecture of the tissue. Our staining confirms that our NASH specimen is inflamed, contains major fibrotic septa (stained pink in the Van Giesen stain), and abundant steatosis (highlighted by red neutral lipid accumulations in the Oil Red O samples). Microscopic observation confirmed that both microvesicular and macrovesicular steatosis was evident and distribution was not uniform with some regenerative nodules being more intensely steatotic than others. Visualization of this architecture was possible irrespective of tissue treatment (see Figure 3 and Figures S12−13), and all features were identifiable in all samples. Importantly, our examination of the histological sections appears to show that all formal preparations facilitated the uptake of each dye and maintained the microarchitecture of the tissue samples. All tissue features remain evident and distinguishable when tissue sections are treated with a formal fixative and subsequently stained with common histological reagents. Formal lithium fixed tissue sections appeared to closely resemble those from fresh frozen samples (Figure 3). Therefore, it would seem that treating the sample with formal lithium does not limit histology and would be suitable for histology driven lipid imaging studies owing to the improved characterization afforded by lithium adducts. MALDI imaging studies are typically conducted alongside standard histological evaluations of tissue sections. This is becoming increasingly important in clinical studies in which spectra and single ion images can be spatially correlated to different tissue regions in stained sections providing unrivalled insight into the molecular composition and cellular organization of samples. This usually involves analysis of a first section by MALDI, and a serial section is then used for staining. Less commonly, the same tissue section may be surveyed by both approaches. Tissue staining has been performed both preand post-MALDI imaging. Although common dyes such as hemotoxylin and eosin have not lent themselves to this strategy, other routinely used dyes such as cresyl violet and methylene blue have proven more compatible.62,63 Nonetheless, it is widely accepted that structural stains such as hemotoxylin and eosin provide the most useful detail of cellular structure and tissue architecture which is required for histopathological surveys and grading of clinical samples.

over other lipid adducts upon fragmentation during electrospray ionization. It follows that analysis of formal lithium by imaging techniques such as desorption electrospray ionization (DESI) should enable the acquisition of similar data to that presented here by MALDI and LESA. Human Liver Case Study: Compatibility of Formal Metal Fixatives with Common Staining Methods. In order to assess the compatibility of our fixation strategy with standard histological staining techniques, we performed a study using human tissue. Specimens from patients with nonalcoholic steatohepatitis (NASH) were analyzed by MALDI-MSI and LESA−ESI-MS. Serial sections were also stained using traditional chromogenic stains to evaluate any effect of tissue fixation on histological information. MALDI ion images of PC 34:1 show lipid distributions that correlate well with the morphology of the tissue section. This lipid is localized within the hepatocyte nodules in the NASH specimen; intensity varies between individual features, and there is a tendency for increased signal closest to the fibrotic areas. Mapping of the [M + K]+ adduct in fresh frozen tissue and the [M + Li]+ adduct in formal lithium fixed tissue shows similar distributions in fresh and fixed tissue samples, see Figure 3. Due to the heterogeneous nature and complex composition of biological samples it is extremely important to identify the composition and morphology of different tissue types.

Figure 3. MALDI-MSI (−) ion images of lipid PC 34:1 detected in NASH liver sections. A summed spectrum of peaks detected in the region m/z 750−850 is presented in the top panel, and ion images are shown below. There is a clear change in the predominant adduct detected in fresh frozen ([M + K]+) and formal lithium (LiCl) ([M + Li]+) fixed tissue samples. Gross and microscopic images of serial sections of tissue stained with H&E, Van Gieson, and Oil Red O are shown below each tissue image. Scale bars show 100 μm; larger pictures can be viewed in Figures S7 and S8. The opportunity to correlate MS images with histological staining techniques from formal lithium fixed samples, similar to fresh tissue samples, is shown.

■

CONCLUSIONS Our approach of formal lithium fixation is a simple and reproducible method for incorporating lithium salts into tissue which we show to enhance in situ structural characterization in MALDI and ESI experiments. Importantly, the distributions of selected lipids were not found to differ between fresh and fixed 7151



Article

(12) Wigglesworth, V. B. Proc. R. Soc. London, Ser. B 1957, 147, 185− 199. (13) Reynolds, E. S. J. Cell Biol. 1963, 17, 208−212. (14) Stempak, J. G.; Ward, R. T. J. Cell Biol. 1964, 22, 697−701. (15) Adams, C. W. M. J. Histochem. Cytochem. 1960, 8, 262−267. (16) Trent, J. S.; Scheinbeim, J. I.; Couchman, P. R. Macromolecules 1983, 16, 589−598. (17) Watson, M. L. J. Biophys. Biochem. Cytol. 1958, 4, 727−730. (18) Holt, S. J.; Hicks, R. M. J. Biophys. Biochem. Cytol. 1961, 11, 31− 45. (19) Belazi, D.; Solé-Domènech, S.; Johansson, B.; Schalling, M.; Sjövall, P. Histochem. Cell Biol. 2009, 132, 105−115. (20) Deierkauf, F. A.; Heslinga, F. J. M. J. Histochem. Cytochem. 1962, 10, 79−82. (21) Boskey, A. L.; Cohen, M. L.; Bullough, P. G. Calcif. Tissue Int. 1982, 34, 328−331. (22) Carter, C.; McLeod, C.; Bunch, J. J. Am. Soc. Mass Spectrom. 2011, 22, 1991−1998. (23) Palmer, A. D.; Griffiths, R.; Styles, I.; Claridge, E.; Calcagni, A.; Bunch, J. J. Mass Spectrom. 2012, 47, 237−241. (24) Cerruti, C. D.; Touboul, D.; Guérineau, V.; Petit, V. W.; Laprévote, O.; Brunelle, A. Anal. Bioanal. Chem. 2011, 1−13. (25) Ho, Y.-P.; Huang, P.-C.; Deng, K.-H. Rapid Commun. Mass Spectrom. 2003, 17, 114−121. (26) Jackson, S. N.; Wang, H.-Y. J.; Woods, A. S. J. Am. Soc. Mass Spectrom. 2005, 16, 2052−2056. (27) Stubiger, G.; Pittenauer, E.; Allmaier, G. Anal. Chem. 2008, 80, 1664−1678. (28) Stübiger, G.; Belgacem, O. Anal. Chem. 2007, 79, 3206−3213. (29) Jackson, S. N.; Wang, H.-Y. J.; Woods, A. S. Anal. Chem. 2005, 77, 4523−4527. (30) Griffiths, R. L.; Bunch, J. Rapid Commun. Mass Spectrom. 2012, 26, 1557−1566. (31) Hsu, F.-F.; Bohrer, A.; Turk, J. J. Am. Soc. Mass Spectrom. 1998, 9, 516−526. (32) Turk, F. H. a. J. J. Am. Soc. Mass Spectrom. 2003, 14, 352−363. (33) Hsu, F.-F.; Turk, J. J. Am. Soc. Mass Spectrom. 2003, 14, 352− 363. (34) De Oliveira, L.; Câmara, N. O. S.; Bonetti, T.; Turco, E. G. L.; Bertolla, R. P.; Moron, A. F.; Sass, N.; Da Silva, I. D. C. G. Clin. Biochem. 2012, 45, 852−855. (35) Feldstein, A. E.; Lopez, R.; Tamimi, T. A.-R.; Yerian, L.; Chung, Y.-M.; Berk, M.; Zhang, R.; McIntyre, T. M.; Hazen, S. L. J. Lipid Res. 2010, 51, 3046−3054. (36) Ashraf, M. Z.; Kar, N. S.; Podrez, E. A. Int. J. Biochem. Cell Biol. 2009, 41, 1241−1244. (37) Stauber, J.; Lemaire, R.; Franck, J.; Bonnel, D.; Croix, D.; Day, R.; Wisztorski, M.; Fournier, I.; Salzet, M. J. Proteome Res. 2008, 7, 969−978. (38) Touboul, D.; Roy, S.; Germain, D. P.; Chaminade, P.; Brunelle, A.; Laprévote, O. Int. J. Mass Spectrom. 2007, 260, 158−165. (39) Hart, P.; Francese, S.; Claude, E.; Woodroofe, M.; Clench, M. Anal. Bioanal. Chem. 2011, 401, 115−125. (40) Puolitaival, S. M.; Burnum, K. E.; Cornett, D. S.; Caprioli, R. M. J. Am. Soc. Mass Spectrom. 2008, 19, 882−886. (41) Race, A. M.; Styles, I. B.; Bunch, J. J. Proteomics 2012, 75, 5111− 5112. (42) Chambers, M. C.; Maclean, B.; Burke, R.; Amodei, D.; Ruderman, D. L.; Neumann, S.; Gatto, L.; Fischer, B.; Pratt, B.; Egertson, J.; Hoff, K.; Kessner, D.; Tasman, N.; Shulman, N.; Frewen, B.; Baker, T. A.; Brusniak, M.-Y.; Paulse, C.; Creasy, D.; Flashner, L.; Kani, K.; Moulding, C.; Seymour, S. L.; Nuwaysir, L. M.; Lefebvre, B.; Kuhlmann, F.; Roark, J.; Rainer, P.; Detlev, S.; Hemenway, T.; Huhmer, A.; Langridge, J.; Connolly, B.; Chadick, T.; Holly, K.; Eckels, J.; Deutsch, E. W.; Moritz, R. L.; Katz, J. E.; Agus, D. B.; MacCoss, M.; Tabb, D. L.; Mallick, P. Nat. Biotechnol. 2012, 30, 918− 920. (43) Spraggins, J.; Caprioli, R. J. Am. Soc. Mass Spectrom. 2011, 22, 1022−1031.

tissues proving this as a viable sample preparation strategy for MALDI imaging. We have demonstrated the compatibility of formal lithium fixation (of single sections and whole tissue blocks) with three common stains for examination of liver tissue. We report that formal lithium fixation not only enhances the in situ information that may be obtained in MALDI-MS and MS/ MS imaging studies, but also the uptake and usefulness of hemotoxylin and eosin, as well as Van Gieson and Oil Red O, are not compromised. With increasing pressure to identify potential biomarkers in clinical investigations, we hope that this methodology will provide lipid analysts with a single, histologycompatible protocol for multimodal analysis of tissue sections: Formal lithium is a useful tissue preservative and an effective route to introducing lithium for CID in MALDI and ESI studies and proves to be compatible with common histological stains.

■

ASSOCIATED CONTENT

S Supporting Information *

Figure S1 showing the incorporation of lithium additives via airspray and dry coating deposition methods. Figure S2 showing tile of images showing changing adduct formation in fresh frozen and formal fixative types. Tables S3−6 listing lipids identified by accurate mass in LESA−ESI experiments. Tables S7 and S8 showing staining of NASH liver tissue. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was funded through studentships to A.M.R., R.T.S., J.S., and E.J.G. via the EPSRC through the PSIBS Doctoral Training Center, Grant EP/F50053X/1, and through a studentship from The School of Chemistry, University of Birmingham, to R.L.G.

■

REFERENCES

(1) Werner, M.; Chott, A.; Fabiano, A.; Battifora, H. Am. J. Surg. Pathol. 2000, 24, 1016. (2) Puchtler, H.; Meloan, S. N. Histochemistry 1985, 82, 201−204. (3) Kiernan, J. A. Microsc. Today 2000, 1, 8−12. (4) Nowacek, J. M.; Kiernan, J. A. Fixation and Tissue Processing. In Special Stains and H&E; Kiernan, J. A., Kumar, G. L., Eds., Dako North America: Carpinteria, CA, 2010; pp 141−152. (5) Buesa, R. J. Ann. Diagn. Pathol. 2008, 12, 387−396. (6) Glauret, A. M. 2.2 Vehicles for Fixatives. In Practical Methods in Electron Microscopy: Fixation, Dehydration and Embedding of Biological Specimens; Glauret, A. M., Ed.; North-Holland Publishing Company: Amsterdam, 1978; pp 7−21. (7) Mugnaini, E.; Dahl, A. L. J. Histochem. Cytochem. 1983, 31, 1435− 8. (8) Wester, K.; Asplund, A.; Bäckvall, H.; Micke, P.; Derveniece, A.; Hartmane, I.; Malmström, P. U.; Pontén, F. Lab. Invest. 2003, 83, 889−899. (9) Watson, M. L. J. Biophys. Biochem. Cytol. 1958, 4, 475−478. (10) Hall, C.; Jakus, M.; Schmitt, F. J. Appl. Phys. 1945, 16, 459−465. (11) Palay, S. L.; McGee-Russell, S.; Gordon, S., Jr.; Grillo, M. A. J. Cell Biol. 1962, 12, 385−410. 7152



Article

(44) Trim, P. J.; Djidja, M. C.; Atkinson, S. J.; Oakes, K.; Cole, L. M.; Anderson, D. M. G.; Hart, P. J.; Francese, S.; Clench, M. R. Anal. Bioanal. Chem. 2010, 1−11. (45) Hankin, J. A.; Barkley, R. M.; Murphy, R. C. J. Am. Soc. Mass Spectrom. 2007, 18, 1646−1652. (46) Goodwin, R. J. A.; Scullion, P.; MacIntyre, L.; Watson, D. G.; Pitt, A. R. Anal. Chem. 2010, 82, 3868−3873. (47) Jackson, S. N.; Woods, A. S. J. Chromatogr., B 2009, 877, 2822− 2829. (48) Jackson, S. N.; Wang, H.-Y. J.; Woods, A. S. J. Am. Soc. Mass Spectrom. 2007, 18, 17−26. (49) Perdian, D. C.; Lee, Y. J. Anal. Chem. 2010, 82, 9393−9400. (50) Prideaux, B.; Dartois, V.; Staab, D.; Weiner, D. M.; Goh, A.; Via, L. E.; Barry, C. E.; Stoeckli, M. Anal. Chem. 2011, 83, 2112−2118. (51) Khatib-Shahidi, S.; Andersson, M.; Herman, J. L.; Gillespie, T. A.; Caprioli, R. M. Anal. Chem. 2006, 78, 6448−6456. (52) Reyzer, M. L.; Hsieh, Y.; Ng, K.; Korfmacher, W. A.; Caprioli, R. M. J. Mass Spectrom. 2003, 38, 1081−1092. (53) Yagnik, G. B.; Korte, A. R.; Lee, Y. J. J. Mass Spectrom. 2013, 48, 100−104. (54) Steven, R.; Bunch, J. Anal. Bioanal. Chem. 2013, 1−10. (55) Garrett, T. J.; Prieto-Conaway, M. C.; Kovtoun, V.; Bui, H.; Izgarian, N.; Stafford, G.; Yost, R. A. Int. J. Mass Spectrom. 2007, 260, 166−176. (56) Edwards, R. L.; Creese, A. J.; Baumert, M.; Griffiths, P.; Bunch, J.; Cooper, H. J. Anal. Chem. 2011, 83, 2265−2270. (57) Hsu, F.-F.; Turk, J. J. Am. Soc. Mass Spectrom. 1999, 10, 587− 599. (58) Hsu, F.-F.; Turk, J. J. Am. Soc. Mass Spectrom. 1999, 10, 600− 612. (59) Jeon, Y. E.; Lee, S. C.; Paik, S. S.; Lee, K. G.; Jin, S. Y.; Kim, H. R.; Yoo, C. W.; Park, H. M.; Han, S. Y.; Choi, D. H.; Kim, H. K. Pathol. Int. 2011, 61, 449−455. (60) Agar, N.; Yang, H.; Carroll, R.; Black, P.; Agar, J. Anal. Chem. 2007, 79, 7416−7423. (61) Cornett, D. S.; Mobley, J. A.; Dias, E. C.; Andersson, M.; Arteaga, C. L.; Sanders, M. E.; Caprioli, R. M. Mol. Cell. Proteomics 2006, 5, 1975−1983. (62) Chaurand, P.; Schwartz, S. A.; Billheimer, D.; Xu, B. J.; Crecelius, A.; Caprioli, R. M. Anal. Chem. 2004, 76, 1145−1155. (63) Walch, A.; Rauser, S.; Deininger, S.-O.; Hofler, H. J. Histochem. Cell Biol. 2008, 421−434.

7153


Formal Lithium Fixation Improves Direct Analysis of Lipids in Tissue by

Recommend Documents