Nondestructive, Histologically Compatible Tissue Imaging by Desorption Electrospray Ionization Mass Spectrometry

Maintaining your integrity: Morphologically friendly DESI-MS imaging of tissue can be performed prior to histochemical and immuno-histochemical analysis on same tissue section by using new solvent systems. This novel DESI-MS imaging capability allows chemical information to be obtained while preserving tissue integrity. Imaging mass spectrometry (MS) is a powerful technique that has been gaining increased attention and broad use in many fields of science. Many biological applications are being pursued that make use of the comprehensive information it provides on the distribution of multiple endogenous and exogenous molecules within animal and plant tissues.1 Amongst the imaging MS techniques,2 those based on ambient ionization3—desorption electrospray ionization mass spectrometry (DESI-MS),4 laser ablation electrospray ionization (LAESI),5 and atmospheric pressure femtosecond laser imaging mass spectrometry (AP fs-LDI IMS)6 amongst others7—allow analysis at atmospheric pressure without significant sample preparation. In the past few years, much effort has gone into advancing ambient imaging mass spectrometry, especially in cancer diagnostics.8 The prospect of improving the accuracy of histopathological cancer evaluation by adding chemical information to morphological microscopic analysis now represents an attainable advance. The capabilities of DESI-MS imaging as a diagnostic tool are demonstrated in many studies in which the lipid profiles directly detected from different types of tissues, such as kidney,9 bladder,10 prostate,11 and brain12 cancers, enabled diagnosis of multiple types of human cancers. However, technical challenges remain, and validation studies are still needed to successfully merge microscopic and MS information into a routine histopathology workflow. Correlation between histology and DESI-MS has thus far been performed by comparing the ion images obtained to the diagnosis from pathological evaluation of a serial section, typically a hematoxylin and eosin (H&E)-stained section.9 Even though this strategy is sufficient for optical image evaluation under routine microscopic pathology, unambiguous correlation demands the use of the same tissue section for morphological and MS imaging evaluation.13 The primary reason for the incompatibility of DESI-MS imaging with histochemical examination is that the most commonly used solvent systems, such as methanol/water and acetonitrile (ACN)/water (1:1, v/v), yield extensive chemical information but completely destroy the native tissue morphology in the process, thereby precluding any subsequent analysis on the same tissue section. In an effort to overcome this limitation, we have developed a methodology that allows DESI-MS imaging to be performed while preserving tissue integrity. New solvent systems, which we refer to as "morphologically friendly", allow a spot-by-spot extraction of lipid species from tissue during DESI-MS analysis while the morphology of the tissue remains undisturbed. This is demonstrated here by imaging lipids from multiple types of human cancerous and normal tissue to achieve a molecular diagnosis and then examining the same tissue section by H&E staining or immunohistochemistry to allow an independent histochemical diagnosis. DESI-MS imaging is based on microscopic-scale solvent extraction, and many studies have shown that the chemical and physical properties of the solvent system chosen affect the molecular information obtained.14 However, the effect of the solvent system on the morphology of the tissue being analyzed by DESI-MS is an important property that has not yet been carefully explored. In this study, many solvents such as ACN, H2O, MeOH, ethanol, tetrahydrofuran, N,N-dimethylformamide (DMF), chloroform, and acetone as well as their mixtures, were explored in the analysis of 15-μm-thick serial coronary mouse-brain tissue sections (Figure S1 in the Supporting Information). The majority of the ions observed in the mass spectra obtained from the solvent systems tested correspond to lipid species commonly observed in brain tissue when using standard MeOH/H2O (1:1),15 such as deprotonated free fatty acids, phosphatidylserines (PS), phosphatidylinositols (PI), and sulfatides (ST), with variations in the relative abundances of the lipid species and in the total ion signal. Remarkably, the use of DMF as the spray solvent did not cause visible damage to the tissue section.16 The effect of DMF on the tissue was further explored by combining this solvent with other solvents in binary (1:1, v/v) and tertiary mixtures (Figure S2). None of the DMF solvent combinations tested caused damage to the tissue sections analyzed; however, they did provide extensive chemical information. For example, combining DMF with ACN, EtOH, THF, or CHCl3 yielded high ion signals and chemical information similar to that seen when using standard MeOH/H2O, while combinations with either H2O or MeOH greatly enhanced the signal of low m/z compounds, such as small metabolites, fatty acids (FA) and FA dimers.16 To confirm the preservation of tissue integrity, H&E staining was performed directly on the tissue sections after DESI-MS imaging, without any intermediate sample preparation steps. Microscopic examination of the H&E tissue sections revealed no damage to or change in the cellular morphology of the sample after DESI-MS analysis with DMF/EtOH and DMF/H2O solvent systems, while the tissue analyzed with MeOH/H2O was found to be significantly damaged, as observed visually (Figure S3). The physical and chemical effects of the DMF/EtOH solvent system were further investigated by performing ten consecutive DESI-MS image acquisition experiments on the same region of a 15-μm-thick mouse-brain tissue section (Supporting Information). After these DESI-MS imaging analyses, the section was H&E stained, and found to still show no morphologic damage. A plot of the ion count of a representative ion observed in the gray-matter region, m/z 834.3 (PS 18:0/22:6), as a function of the DESI-MS analysis number revealed an approximately exponential decay in ion count; this is consistent with the spot-by-spot microextraction mechanism proposed for DESI-MS (Figure S4).17 The extraction process that occurs in DESI-MS can be compared to the fixative procedures commonly used in histology, in which lipids are removed whereas the nucleoproteins and the intracellular and extracellular proteins stained in the histochemical treatment are preserved. All combinations of DMF with other solvents tested in this particular study on mouse-brain tissue sections (2, 3, 5, 10 and 15 μm thicknesses) by DESI-MS imaging were found to be morphologically friendly (Supporting Information). Some other pure solvents, such as ACN, DMF, THF, and EtOH, as well as a few other solvent combinations that did not contain DMF, such as ACN/EtOH (1:1), MeOH/CHCl3 (1:1) and ACN/CHCl3 (1:1), also appeared to be morphologically friendly, whereas many more, such as pure H2O, THF/H2O (1:1), EtOH/H2O (1:1), MeOH/H2O (1:1) and ACN/H2O (1:1), were not, and the tissue after DESI-MS imaging was damaged physically. The morphological effect that the DESI-MS spray has on tissue is related to the physical and chemical properties of the solvent system.17 The relative solubility of lipids and insolubility of the cellular and extracellular protein components of the tissue section in the DESI spray solvent helps to conserve tissue morphology, while providing high-quality lipid mass spectra. The physical properties of the solvent, such as surface tension and its effects on the dynamics of the DESI spray, are the proximate cause of the damage to the tissue. Solubilization of cellular and extracellular constituents of the cells allows the tissue to become more susceptible to the mechanical action of the DESI spray droplets, and this appears to be the case for conventional solvents such as MeOH/H2O.18 In the experiments shown, our choice of DMF combinations such as DMF/EtOH and DMF/ACN arises from the fact that high ion signal intensity with low background was achieved for the compounds of interest, and that a very stable and very well defined spray spot was obtained, which is ideal for imaging applications. Our focus in this communication is to demonstrate the ability of DESI-MS imaging to be histologically compatible for biomedical applications. The underlying reasons for this compatibility must be related to the physical and chemical effects of these solvents on tissue, but the nature of these effects remains to be elucidated. Chemical information and image quality are the most important factors in DESI-MS imaging applications. Figure 1 A shows an average mass spectrum of a mouse-brain tissue section obtained with DMF/EtOH as the solvent system. The spray geometry, gas pressure, and solvent flow rate used in the imaging experiments were optimized so that a spot size of 180 μm was obtained (Supporting Information).15 An optical image of the histologically compatible DESI-MS imaging experiment is shown in Figure 1 B. The two distinctive negative-ion-mode MS patterns associated with the lipid compositions of the gray and the white matter of the mouse are seen in the ion images shown in Figure 1 C–F.15 Lipids were tentatively assigned based on tandem MS experiments, although isomers are known to occur. Figure 1 G shows an optical image of the same tissue section, which was H&E stained after DESI-MS imaging. The high-quality 2D DESI-MS ion images can be directly compared and overlaid with the H&E stained tissue section for better correlation between the spatial distribution of the lipid species and brain substructures. A) Average DESI-MS mass spectrum of a mouse-brain coronal section including regions of white and gray matter obtained by using DMF/EtOH as the solvent system. B) Picture of the DESI-MS imaging experiment. DESI-MS ion images show the distribution of C) m/z 834.3, PS(18:0/22:6); D) m/z 888.6, ST(24:1); E) m/z 885.6, PI(18:0/20:4), and F) m/z 303.3, FA(20:4). G) Optical image of the same tissue section after first being imaged by DESI-MS and then H&E stained. The ability to perform DESI-MS imaging and histochemical analysis of the same tissue section is important for accurate correlation between molecular signatures and disease state. Because the information from the two methods is orthogonal, it is expected to greatly improve discrimination and diagnosis of the disease state. This is especially true in the analysis of cancerous tissue sections, which are very often highly heterogeneous with regions containing various tumor cell concentrations,19 infiltrative normal tissue,10 and precancerous lesions,11 within other features. The integration of DESI-MS imaging into a traditional histopathology workflow requires that any MS analysis before histology not interfere with the morphology of the tissue section. To investigate this capability, human bladder, kidney, and prostate cancer tissues along with adjacent normal tissues were analyzed by DESI-MS imaging in the negative-ion mode by using one of the histology-compatible solvents and then H&E stained or subjected to immuno-histochemical (IHC) analysis with p63 antibody.20 Figure 2 shows a series of negative-ion-mode DESI-MS ion images of human bladder transitional cell carcinoma and adjacent normal tissue samples. As previously shown by DESI-MS imaging,10 the ions that most significantly discriminate between cancerous and normal bladder tissue are the free fatty acids (FA) and their dimers (Figure S5), which consistently appear at increased intensities in the cancerous tissue (Figure 2 D and E). Detailed pathological examination of the same tissue sections stained with H&E after DESI-MS imaging (Figure 2 F) compared to a control section confirmed that no damage to the cells' morphology or to the overall tissue integrity was caused as a result of DESI-MS imaging analysis (Figure S6). The nondestructive nature of the DMF-based solvent system enables ion images to be overlaid with an H&E stain of the same tissue section for unambiguous diagnosis and correlation. For example, a small region of tissue within the cancerous section detected by DESI-MS as negative for bladder cancer based on the distribution of FA dimer (m/z 537.2) was confirmed as normal tissue by pathological evaluation of the overlaid DESI-MS ion image and H&E stain of the same tissue section (Figure 2 G). DESI-MS imaging of human-bladder cancerous and adjacent normal tissue sections by using morphologically friendly DMF/EtOH as the solvent system. Ion images show the distribution of A) m/z 788.4, PS(18:0/18:1); B) m/z 885.6, PI (18:0/20:4); C) m/z 835.6, PI(16:0/18:1); D) m/z 281.6, FA (18:1) and E) m/z 537.2 (FA dimer). F) After the DESI-MS imaging experiment, the tissue sections were subjected to H&E staining, evaluated by an expert pathologist, and diagnosed. G) Overlay of the ion image of m/z 537.2 and H&E stain of the same tissue section. Compatibility between tissue DESI-MS imaging and IHC was also investigated. Figure 3 shows magnified bright-field optical images of control tissues and prostate cancer tissue sections that were first imaged with DMF/EtOH or conventional ACN/H2O and then H&E stained or subjected to p63 IHC. Compared to the respective control tissues, no spatial delocalization of the cellular proteic content stained during the IHC or H&E protocols was observed as a consequence of DESI-MS imaging analysis when using DMF/EtOH, as confirmed by detailed pathological evaluation. The destructive nature of conventional solvents system is clearly observed and prevents any further morphological analysis. The same histological observations were true for analysis of other samples of human bladder, prostate, and kidney cancers (Figures S7–S9, respectively) and their adjacent normal tissue. Previous molecular information that allowed a diagnosis to be obtained for these types of cancer was consistent with information obtained using the new solvent systems.9, 11 Magnified bright-field optical images of prostate cancer tissue sections subjected to H&E stain or p63 IHC. Optical images of the entire tissue sections are shown to the left of each magnification. When compared to controls (A), no effect on tissue morphology was observed in tissue sections first imaged by DESI-MS using DMF/EtOH (B). C) Tissue destruction does occur when using conventional solvent systems, such as ACN/H2O. Scale bars=100 μm. The results reported here introduce a novel capability of DESI-MS tissue imaging—histological compatibility. Performing DESI-MS imaging while maintaining tissue integrity and cell morphology allows ambient MS analysis of tissue sections to be combined with traditional histopathology to provide more accurate disease diagnoses. This means that DESI-MS imaging may now be inserted at any point into a workflow profile dealing with tissue analysis. We anticipate that these advances will allow DESI-MS to be more broadly applied as a molecular diagnostic tool. DESI-MS imaging could be included as the initial step in a clinical tissue-analysis workflow. Furthermore, the morphologically compatible solvent system allows DESI-MS imaging to be combined with other analytical techniques for chemical analysis of the same tissue section. These include immunofluorescence and other spectroscopic and imaging experiments. We are currently investigating higher-spatial-resolution applications of DESI-MS imaging by using the developed morphologically friendly solvent systems. The methodology being introduced here now allows DESI-MS to be considered as an essentially nondestructive technique in terms of tissue morphology preservation, introducing new possibilities to expand the applications of DESI-MS imaging further in the biomedical field, to incorporate this technique into routine histopathological analysis for better accuracy of cancer diagnosis and grading, and ultimately to the strategy of intraoperative DESI-MS analysis. Abbreviations: ACN: acetonitrile, AP fs-LDI IMS: atmospheric-pressure femtosecond laser imaging mass spectrometry, DESI-MS: desorption electrospray ionization mass spectrometry, DMF: N,N-dimethylformamide, FA: fatty acids, H&E: hematoxylin and eosin, LAESI: laser ablation electrospray ionization, MS: mass spectrometry, PI: phosphatidylinositols, PS: phosphatidylserines, ST: sulfatides This work was supported by the U.S. National Institutes of Health (grant 1R21EB009459-01) and by the Indiana Clinical and Translational Sciences Institute (in part from the NIH, National Center for Research Resources, Clinical and Translational Sciences Award through grant RR025761). We gratefully acknowledge Dr. Timothy Masterson at the IU School of Medicine for assistance in obtaining tissue samples and Prof. Marcos Eberlin for valuable discussions. Detailed facts of importance to specialist readers are published as "Supporting Information". Such documents are peer-reviewed, but not copy-edited or typeset. They are made available as submitted by the authors. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.