Direct Tissue Analysis by MALDI MS: Application to Kidney Biology

Abstract

Direct tissue analysis using MALDI MS provides in situ molecular analysis of a wide variety of biological molecules including xenobiotics. This technology allows measurement of these species in their native biological environment without the use of target specific reagents such as antibodies. It can be used to profile discrete cellular regions and obtain region-specific images, providing information on the relative abundance and spatial distribution of proteins, peptides, lipids, and drugs. In this chapter, we report the sample preparation, MS data acquisition and analysis, and protein identification methodologies utilized in our laboratory for profiling/imaging MS and how this has been applied to kidney disease and toxicity.

Introduction

MALDI MS, introduced in the late 1980s [1,2], has become an enabling analytical technology to study proteins in biological systems because of its molecular specificity and high-throughput capabilities [3,4]. One of the more recent developments to this technology is its application to direct tissue analysis for both molecular profiling and imaging [5]. The technology allows for the analysis of molecules in tissues without the need for target specific reagents such as antibodies or the need for tissue homogenization, thereby maintaining the integrity of the tissue sample and allowing assessment of molecular spatial distribution.

The principles of MALDI MS have been described [2,6–9]. In brief, the analyte of interest is mixed with an energy absorbing compound (matrix) on a MALDI target plate. As the solvent evaporates, the analyte co-crystallizes with the matrix. Inside the mass spectrometer, molecules are desorbed from the sample by irradiation with a UV laser. In this process, analytes become protonated and primarily give rise to [M+H]⁺ ions that are subsequently measured according to their mass-to-charge ratio (m/z). Typically, the time-of-flight analyzers are employed to measure m/z values of ions [8,9] (Fig. 1).

Fig 1.

MALDI-TOF Schematic. A sample is irradiated with a brief laser pulse and molecules are desorbed from the surface. Energy is absorbed by the matrix and transferred to the analyte, which becomes protonated and is then accelerated down a field free drift tube. Ions collide with a detector at the end of this tube and their time-of-flight is measured and through calibration converted to a mass-to-charge ratio (m/z).

MALDI MS can be used to ‘profile’ molecules in a sample, that is, analyze a small number of discrete spots scattered throughout the sample, or “image” a sample, that is, systematically raster across a tissue section to produce an ordered array of spots. In the latter, the laser performs a raster over the tissue surface in a pre-defined two-dimensional array or grid, generating a full mass spectrum at each grid coordinate. Mass spectra acquired in this manner display ions in the m/z range of 500 to over 100,000, corresponding to many hundreds of different molecules present in the tissue. The coordinates of the irradiated spots are used to generate two-dimensional ion density maps, or images, that represent individual m/z values with their corresponding intensities (Fig. 2). Likewise, drug analysis in tissue can be accomplished by monitoring the protonated drug or its metabolites and their corresponding fragments at each discrete coordinate of such an array.

Fig 2.

Images generated by imaging MALDI MS. A) Optical hematoxylin and eosin stained (top) and overlaid ion images (bottom) of a normal rat kidney section. Shown are three ions represented in color to indicate localization (pink: m/z 8451 in the cortex, green: m/z 8560 in the outer medulla, blue: m/z 4965 localized primarily in the inner medulla). B) Optical hematoxylin and eosin stained (top) and ion images (bottom) of a human clear cell renal cell carcinoma tissue section with tumor and adjacent non-tumor tissue. Shown are two overlaid ions (red: m/z 11090 localized to tumor, green: m/z 4563 in non-tumor cortex).

Direct tissue analysis by MALDI has been used to detect drugs and their metabolites [10–13] as well as intact proteins and peptides [14,15] directly from tissue. Profiling/imaging MS has been used to study molecular aspects of cancer, providing tumor-specific markers as well as diagnostic and prognostic-specific markers [16,17]. In this chapter, we describe the current profiling/imaging methodologies utilized in our laboratory and their application to glomerulosclerosis, drug-induced renal toxicity and renal cancer.

Sample Prep for Profiling and Imaging Tissues by MS

Direct MS analysis is usually performed on tissues that have been excised and immediately frozen in liquid nitrogen in order to maintain tissue morphology and minimize molecular degradation. Maintaining the integrity of the tissue throughout this process maximizes the information obtained in the analyses by ensuring that the original three-dimensional structure is not compromised and the molecular species monitored have not been altered as a result of the sample procurement and preparation process.

Frozen tissues are sectioned in a cryostat, if necessary, using a small amount of optimum cutting temperature (OCT) media on the cutting block for support. Preferably, tissue should not be embedded in the polymer media as surface contamination with polymer caused by the cutting process tends to reduce ionization efficiency. A cryostat temperature of between −15 °C to −25 °C is generally optimal for sectioning kidney tissue. Sections are generally cut at 12 μm thickness but may range between 10–20 μm depending on the application [4,10]. After cutting, sections are thaw-mounted onto a MALDI target plate and washed and/or allowed to dry in a vacuum desiccator.

It is advantageous to wash tissue sections that are rich in salts and other contaminants as well as hemoglobin prior to matrix deposition because these species can contribute to a high spectral baseline and generalized signal suppression. Because of the kidney’s role in filtration of blood, washing kidney sections removes excess hemoglobin and salts and enhances signal quality. The conventional tissue washing method involves two separate 70% ethanol submersions for 20–30s followed by 10–15s in 95% 200-proof ethanol (or 100% reagent grade ethanol). Ethanol is a known tissue fixative in histology [18] and delocalization of proteins appears not to be significant; however it is noted that proteins soluble in these aqueous solutions may be removed during the washing step. Previous studies have indicated minimal protein loss during this process, but the washing procedure should be confirmed for each tissue type [19,20].

Selection of the matrix/solvent combination for direct tissue analysis depends on the molecular weight, hydrophobicity, and salt content of the analyte. Typically, 3,5-dimethoxy-4-hydroxycinnamic acid (sinapinic acid, SA) is favorable for proteins (MW > 2 kDa), and α-cyano-4-hydroxycinnamic acid (CHCA) is optimal for peptides (500–2000 Da). Small molecule analysis, including drugs and lipids, is usually performed with 2,5-dihydroxybenzoic acid (DHB). The optimal matrix concentration range is 10–30 mg/ml SA for protein analysis and 10–20 mg/ml CHCA for peptide analysis. The usual matrix solvent is 50% acetonitrile (ACN) with 0.1% trifluoroacetic acid (TFA). More nonpolar solvents such as methanol or isopropanol can be used for more hydrophobic matrices. The optimal matrix/solvent combination may vary between tissue types and analytes and must be assessed for each study.

Various methods of matrix deposition on thin tissue sections have been successfully employed. For tissue profiling, matrix can be manually deposited either with a pipette or pulled glass capillary. Syringe pumps provide an alternative, more automated approach with higher reproducibility and smaller matrix spot diameters. Robotic deposition offers superior reproducibility, smaller matrix spot diameters and a higher throughput platform, and also allows for the ability to perform histology-directed analyses [21] and imaging. Two types of robotic devices, an acoustic spotter [19] and a chemical inkjet printer [22], have been successfully utilized for tissue profiling and imaging. General matrix deposition guidelines for both robotic types have been described [19,22], but for each tissue type, method optimization is recommended to determine the number of matrix drops and passes required to obtain high quality spectra. In some cases matrix seeding may be desired to aid in the crystallization process [19]. The current minimum spot-to-spot spacing achievable with commercial robotic spotters is in the range of 100–200 μm. For applications that require a resolution higher than 100 μm, spray coating may be used. A spray coating device may be as simple as a glass thin layer chromatography (TLC) spray nebulizer that generates and sprays fine droplets of matrix under slight positive pressure, or automated spray devices. The matrix solution must wet the tissue surface to ensure co-crystal formation between matrix and analyte. For manual spraying methods, a one minute delay between passes provides sufficient drying time. Multiple passes are necessary to coat the entire tissue, but over-coating can suppress analyte signal. The matrix coverage can be monitored as necessary under a microscope.

Histology Directed MS Analysis

Region specificity is important for analysis of heterogeneous tissue types such as kidney. One MALDI compatible method to accomplish this is to stain tissues using cresyl violet stain or other suitable stains (H&E stains give poor MS results), allowing histology and MS analysis to be performed on the same tissue section with minimal interference in signal quality [23]. This method usually requires that tissue sections be applied to a conductive glass slide to obtain histo-pathological detail. A new method, histology-directed MS analysis, utilizes digital imaging to merge traditional histopathology, or hematoxylin and eosin tissue staining, with tissue profiling (Fig. 3). Details of this methodology have been reported [21]. In brief, a pathologist can systematically select areas on a stained tissue section with the usual cellular specificity. By co-registration, the spot coordinates are transferred to a robotic matrix spotter and then to the mass spectrometer. This allows the profiles and other data processing algorithms such as statistical analysis, to be directly mapped onto the histological image and correlation between molecular profiles and histopathology.

Fig 3.

Histology-directed MS sample preparation for tissue analysis. A pathologist systematically selects discrete cellular areas (spots) of interest on a stained tissue section. The coordinates of the spots are transferred to the robotic spotter for matrix deposition on a co-registered serial section with subsequent laser irradiation of these spots (outlined in black for visualization) to give mass spectra.

MS Analysis

The mass spectral acquisition methods used in direct tissue analysis are performed in an automated fashion. When profiling discrete matrix spots, mass spectra are obtained by performing a raster over the area of interest on the tissue. The number of spectra obtained and averaged from a single spot on tissue depends on how many shots can be acquired at one location before signal is depleted and how big the matrix spot is. Most experiments sum between 250 and 500 spectra on a single spot about 150 um in diameter. Mass spectral acquisition from spray coated images does not involve averaging shots at different locations within a pixel. Instead, each ablated location represents one pixel and the number of laser shots within that pixel depends on how many shots can be obtained before the signal significantly decreases.

Spectral pre-processing is carried out on the acquired data to reduce inter- and intra-experimental variance. Such variations may be produced from the ionization process, background noise, and calibration offsets that result from biological or sample preparation differences. Mass spectra are processed by removing background noise, normalizing intensities, and performing a final mass calibration. The latter is primarily used for profiling experiments in which multiple sample groups are being compared, enhancing the efficacy of statistical algorithms to determine biological patterns and changes in these patterns. Imaging applications do not typically require the recalibration step, but background subtraction and normalization significantly enhance the image quality. Illustrations of these processes have been presented [24].

Data Analysis

The large, complex datasets produced by these experiments require robust bioinformatics tools to assess spectral patterns and decipher molecular species that differentiate one group of samples from another. MS data from each group are compared, to determine spectral features (peaks) that are significantly different between the two subject groups. The experimental design and biostatistical methods utilized are critical to the success of such analyses. Many of the algorithms used to decipher the vast amount of proteomic information were initially employed for genomic microarray experiments and have been adapted for proteomic dataset analyses [25–28].

Protein Identification

Identification of statistically relevant mass spectral peaks is necessary to gain insight into biological processes. There are two approaches to protein identification. The top-down approach involves ionization and gas phase fragmentation of the protein of interest inside the mass spectrometer [29], whereas the bottom-up approach utilizes MS to identify peptides obtained from protease digestion of that protein, often in a mixture of other proteolytic fragments [30]. The resulting mass spectra are searched against theoretical protein/peptide databases for corresponding sequence patterns. Searches are performed using conventional algorithms such as MASCOT and Sequest, [31,32]. When possible, manual validation of reported identifications is recommended to reduce false positives.

Commonly, the bottom-up strategy is used and this involves homogenization of the tissue followed by reversed-phase liquid chromatography (RP-LC) separation, where the eluate is continuously collected into fractions. Fractions are vacuumed to dryness, dissolved in 40% acetonitrile with 0.1% TFA, spotted onto a MALDI plate and analyzed for the fractions containing the peptides of interest. Based on the complexity of the fraction, one can enzymatically digest the entire fraction or further separate the mixture by one dimensional gel electrophoresis followed by excision and enzymatic digestion of the band of interest. Peptides are then subjected to further separation and fragmentation by RP-LC-tandem MS analysis and subsequent database searches.

On tissue digestion and protein identification

Recent work has shown that digestion and identification of proteins may be coupled with direct tissue analysis [22]. The technique involves automatically depositing a spotted array of enzymatic solution onto the tissue at room temperature. After hydrolysis, MALDI matrix (25 mg/ml DHB in 50% methanol, 0.5% TFA) is deposited onto the array for subsequent MALDI MS analysis. Figure 4 illustrates an image obtained from on-tissue tryptic digestion of a clear cell renal cell carcinoma section, where a unique distribution of peptides is found in the tumor. This procedure provides the ability to simultaneously assess the distribution of proteins and their corresponding peptides by imaging MS (IMS) to obtain a more complete molecular signature of a tissue section. Additionally, the in situ identification of proteins requires less time than conventional protein identification strategies and the ability to compare a given protein image to the image of its subsequent peptides increases identification confidence. This process is best used for validation, i.e., where the presence of one or more specific proteins is needed to be established. For unknown proteins, the bottom-up or top-down procedure is recommended.

Fig 4.

On-tissue tryptic digestion. Kidney tissue containing clear cell renal cell carcinoma and adjacent non-tumor was sectioned at 12 μm thickness and transferred to a MALDI target plate. The section was washed with ethanol and spotted with an ordered array of 120 picoliter droplets of trypsin solution placed at 250 μm lateral resolution using a robotic spotter. Subsequent to hydrolysis, matrix (CHCA) was directly deposited on these spots and a peptide image obtained. A) Optical image of spotted trypsin and matrix array. B) MALDI ion image of tryptic peptide m/z 1988 localized to the tumor. Figure generated by M. Reid Groseclose.

Application to Kidney

Glomerulosclerosis

Focal segmental glomerulosclerosis (FSGS), or scarring of the glomeruli, is found in scattered regions of the kidney and impairs kidney function. Its etiology and biological causes are unknown, making treatment complicated and diagnosis difficult without a biopsy [33]. The focal segmental characteristic of FSGS provokes the question of whether the nonsclerotic glomeruli in sclerotic tissue are already programmed to become sclerotic or whether these nonsclerotic glomeruli have less prosclerotic activation and therefore are more receptive to therapy. Using direct tissue analysis by MALDI MS, Xu et al [34] used a glomerulosclerotic rat model to determine if proteomic profiles from sclerotic glomeruli could be differentiated from nonsclerotic glomeruli, and secondly, if nonsclerotic glomeruli have a prosclerotic phenotype at the protein level.

To achieve the selective isolation and analysis of glomeruli from tissue, laser capture microdissection (LCM) was utilized. This technique allows for the dissection of specific cells or cell populations from a heterogeneous tissue section. In the FSGS study, LCM was combined with MALDI MS for tissue protein profiling (Fig. 5). The LCM procedure requires that a thin tissue section (~8 um thickness) is mounted on a glass microscope slide and dehydrated in serial ethanol dilutions followed by xylene rinses. The LCM procedure utilizes a narrow laser beam (~7–30 um diameter) to irradiate a heat-sensitive transparent polymer film resulting in adhesion of the film to the section. The cells bound to the polymer are separated from the section when the polymer is removed. This polymer membrane is mounted onto a MALDI target plate using conductive double-sided tape. The cells are spotted with very small volumes of matrix (sinapinic acid) using a thin pulled capillary or a robotic matrix deposition device.

Fig 5.

Schematic of LCM preparation for MALDI MS analysis. A thin tissue section is placed on a glass slide where it is irradiated with a laser beam through a thermoplastic polymer film cap. When the cap is removed, cells that were irradiated remain attached to the cap. The cap is affixed to a MALDI plate and matrix is applied for subsequent MS analysis.

The analysis of LCM-captured cells combined with MALDI MS profiling and statistical analysis resulted in the proteomic differentiation and classification of normal, nonsclerotic, and sclerotic glomeruli. Statistical analysis revealed that nonsclerotic glomeruli in the presence of progressive renal scarring also have an altered proteomic profile that is more similar to sclerotic than normal glomeruli. For example, thymosin β4 was identified as one of several key differentiators between nonsclerotic and sclerotic glomeruli. Its expression was also elevated in these samples as compared to normal glomeruli, suggesting its role as an early activator and indicator of prosclerotic mechanisms.

Drug Nephrotoxicity

The kidney is particularly susceptible to toxic damage because of its large surface area and its role in blood filtration. Current methods of assessing nephrotoxicity, such as monitoring serum creatinine levels and changes in cellular morphology, often fail to detect changes until the onset of toxic nephropathy. Characterization of toxicity markers with profiling and imaging MALDI MS may provide a robust and high-throughput method to detect toxicity in early lead compounds.

A pilot study was performed to examine the proteomic changes induced by gentamicin (Sigma), an antibiotic known to cause proximal tubule damage in kidneys. Experimental details for this study have been described [35]. Briefly, male Wistar rats were treated for seven consecutive days with 100 mg/kg/day gentamicin or vehicle and sacrificed 24 hours after the last treatment. Kidney sections were either manually spotted for profiling or automatically spotted with an acoustic robotic spotter for imaging. Mass spectra from each region contained hundreds of peaks in the m/z range of 3500–25,000. Some signals were confined to anatomical regions, such as the cortex, medulla and papilla, giving each region its own molecular signature. Notably, a peak at m/z 12959 was found to be statistically significant in differentiating control and treated animals. IMS results indicated that this feature was localized to the cortex, the site of damage in these animals (Fig. 6). This protein was extracted directly from the cortex of three tissue sections and the extracts were pooled and separated by RP-LC. The LC fraction containing the peak of interest was subjected to RP-LC-tandem MS analysis and identified as transthyretin, a marker of nutritional status [36–38] and a ligand of the megalin receptor [39] which binds gentamicin [40,41]. The identification was confirmed by western blot and immunohistochemistry.

Fig 6.

Imaging results from gentamicin study. Rats were dosed with gentamicin once daily for seven days, sacrificed and their kidneys excised. Differential protein expression was compared between control and dosed kidneys using IMS. This figure shows the image analysis of m/z 12959 differentially expressed in the cortex of dosed tissue with corresponding average peaks in the spectrum. Figure adapted with permission [35].

Renal Tumor Margins

Another study currently ongoing is the investigation of the molecular distributions in clear cell renal cell carcinoma (ccRCC) with respect to molecular tumor margins. One of the major concerns in clinical oncology is ensuring complete tumor removal to minimize local recurrence and ensure long-term patient survival [42–44]. Depending on the tumor type, recurrence can occur a few months to a few years after removal of the primary tumor, suggesting there are underlying molecular processes in the remaining normal tissue that go undetected using current histopathological techniques. The discovery of new molecular markers for these processes has the potential to provide additional, complimentary approaches to identify abnormal tissue environments.

It is important to be able to measure the molecular characteristics of the tumor and surrounding tumor margin of ccRCC. The radical nephrectomy procedure is the standard treatment for ccRCC but it is excessive for small tumors or patients in whom the tumor is confined to a solitary functioning kidney, has arisen bilaterally or when the remaining kidney could be under future threat from renal deficiencies. The partial nephrectomy procedure maximizes the amount of kidney spared but presents a challenge for clinicians due to concerns over tumor involvement of the surgical margin [45–47].

Proteomic technologies such as profiling/imaging MALDI MS may allow for the elucidation of tumor margin characteristics to further understand tumor spread. Early results have identified molecular signatures that show atypical tissue significantly beyond the histological margin. MALDI-TOF MS was performed on 34 ccRCC biopsies containing tumor and adjacent non-tumor tissue. Matrix arrays were deposited on the tissue, covering the tumor and non-tumor tissue (Fig. 7A). Automated MS acquisition was carried out on each of the matrix spots. After processing mass spectra, statistical analysis was performed on the regions as follows: tumor versus normal and margin normal versus normal (Fig. 7B). Many of the differences between tissue on the normal side of the margin and tissue distant normal were also observed between tumor and nearby histologically normal tissue. Protein identification revealed that one of the families contributing to these abnormal characteristics was mitochondrial electron transport proteins. These proteins were underexpressed in the tumor as compared to normal and underexpressed in the near normal compared to normal (Fig. 6C). Based on the results as well as other studies [48–56], it is probable that increased glycolysis at the expense of mitochondrial oxidative phosphorylation plays a significant role in ccRCC tumor spread into the normal tissue.

Fig 7.

Tumor margin analysis by MALDI MS. A.) Optical image of section with matrix spots with regions of interest highlighted. B.) Optical image of the corresponding hematoxylin and eosin stained section. C.) The top figure shows the amplitude of six m/z values plotted as a function of distance (of corresponding matrix spot analyzed) from the histological tumor margin in one representative tissue sample. Each line represents a different m/z value, each identified as a member of the mitochondrial electron transport chain. These are smoothed trendlines of the original scatterplot data. The bottom figure is a plot of the first derivative of the trend line. Both plots illustrate that the features represented are underexpressed in the tumor and in the margin normal tissue. These features remain underexpressed until ~4000 μm past the histological tumor border.

Conclusions

Direct tissue analysis by MALDI MS is an important technology for assessing the localization of molecular species and for revealing the underlying molecular signatures indicative of disease. The molecular species identified in these experiments can provide insight into mechanisms and etiology of disease. Applications to kidney diseases have just begun, but the potential of this technology to help unravel the complex molecular processes involved are extraordinary. With increasing advances in the field, the utility of this technology will continue to evolve and will play a fundamental role in understanding kidney biology.