Imaging Mass Spectrometry: Principles and Potentials

Abstract

Direct tissue profiling and imaging mass spectrometry (MS) allow for detailed mapping of the complex protein pattern across a tissue sample. Utilization of these tools provides spatial information across a tissue section for target protein expression and can be used to correlate changes in expression levels with specific disease states or drug response. Protein patterns can be directly correlated to known histological regions within the tissue, allowing for the direct monitoring of proteins specific for morphological regions within a tissue sample. Profiling and imaging MS have been used to characterize multiple tissues, including human gliomas and lung cancers, as well as tumor response to specific therapeutics, suggesting the use of proteomic information in assessing disease progression as well as predicting patient response to specific treatments. This article discusses both the technology and methods involved in analyzing proteins directly from tissue samples as well as several MS applications, including profiling human tumors, characterizing protein differences between tumor grades, and monitoring protein changes due to drug therapy.

Introduction

In recent years, mass spectrometry has become an indispensable tool for proteomic studies (Aebersold and Goodlett, 2001; Godovac-Zimmermann and Brown, 2001; Lahm and Langen, 2000; McDonald and Yates, 2000; Pandey and Mann, 2000; Roepstorff, 1997; Russell and Edmondson, 1997). Desorption and ionization techniques such as matrix-assisted laser desorption ionization mass spectrometry (MALDI MS) (Hillenkamp et al., 1991; Karas and Hillenkamp, 1988) and electrospray ionization mass spectrometry (ESI MS) (Fenn et al., 1989) have literally revolutionized our ability to analyze proteins. These improvements offer levels of sensitivity and mass accuracy never before achieved for the detection, identification and structural characterization of proteins. It is now possible to routinely measure molecular weights above 200 kDa as well as obtain low parts per million mass measurement accuracies for the determination of peptides and proteins. Protein identification has been greatly facilitated because of the rapid expansion of protein and gene databases. Modern mass spectrometers can now rapidly map and fragment peptides that result from protease digestion in order to obtain sequence information and identify proteins.

MALDI MS is an ideal tool to investigate complex protein mixtures. It utilizes a matrix, a small acidic aromatic molecule that absorbs energy at the wavelength of the irradiating laser. The analyte molecule is mixed with the matrix in a ratio of typically 1/5000, deposited on a target plate and allowed to dry. During the drying process, matrix-analyte co-crystals form. These crystals are then submitted to very short laser pulses (typically UV laser light), resulting in the desorption and ionization of the analyte molecule. Mostly intact protonated molecular ions are formed ([M+H]⁺, where M is the molecular weight of the analyte molecule). The mass-to-charge (m/z) of the ion is typically measured in a time-of-flight mass analyzer (Cotter, 1999) (Figure 1).

There are several technologies available to analyze proteins in a tissue specimen. To date, the most commonly used technology is the separation and visualization of proteins by 2-dimensional (2-D) gel electrophoresis and subsequent identification by mass spectrometry and database searching (Lahm and Langen, 2000; Godovac-Zimmermann and Brown, 2001). One of the drawbacks of the 2-D gel technology is that sample preparation removes the direct relationship between morphological tissue regions and a specific protein. One solution to this problem is to purify cells from thin tissue sections by laser capture microdissection prior to protein extraction. Although such an approach has been successful (Curran et al., 2000), the extraction of a sufficient quantity of material is very labor intensive and requires a large amount of microdissected cells.

One of the recent applications of MALDI MS is its use to profile and image proteins directly from thin tissue sections (Todd et al., 2001; Chaurand and Caprioli, 2002; Chaurand et al., 2002, 2004b). MALDI imaging mass spectrometry (IMS) is a new technology that allows for simultaneous mapping of hundreds of peptides and proteins present in thin tissue sections with a lateral resolution of about 30–50 μm. Matrix is first uniformly deposited over the surface of the section, utilizing procedures optimized to minimize protein migration. Proteins are then desorbed from discrete spots or pixels upon irradiation of the sample in an ordered array or raster of the surface. Each pixel thus is keyed to a full mass spectrum consisting of signals from protonated species of molecules desorbed from that tissue region. A plot of the intensity of any one signal produces a map of the relative amount of that compound over the entire imaged surface. This technology provides an extremely powerful discovery tool for the investigation of biological processes because the identities of proteins observed do not need to be known in advance. IMS under various forms has already been successfully used to further characterize the expression of proteins and other organic biological compounds in numerous normal and diseased tissues. For example, protein organization in mouse colon (Chaurand et al., 1999), brain (Stoeckli et al., 2001; Todd et al., 2001; Chaurand et al., 2002, 2004b) and epididymis (Chaurand et al., 2003) as well as phospholipid organization in mammalian lens tissue (Rujoi et al., 2004) has been studied. Variations in protein expression have been investigated in the cases of Parkinson’s (Pierson et al., 2004) and Alzheimer’s (Stoeckli et al., 2002) diseases. Several forms of cancers have also been investigated including gliomas (Stoeckli et al., 2001; Chaurand et al., 2004b; Schwartz et al., 2004), breast cancer (Palmer-Toy et al., 2000; Xu et al., 2002), prostate cancer (Masumori et al., 2001), colon cancer (Chaurand et al., 2001a) and lung cancer (Bhattacharya et al., 2003; Yanagisawa et al., 2003). In this later study (Yanagisawa et al., 2003), protein patterns have been shown to be predictive of diagnosis and prognosis. Methodologies aimed at detecting and mapping pharmaceutical compounds by direct MALDI MS analysis of sections from dosed tissues have also been described (Troendle et al., 1999; Reyzer et al., 2003).

We review here the basic methodologies for tissue sample handling and preparation for analysis by MALDI MS using various mouse and human cancer tissue samples and discuss current and future potentials of the methodology for basic research and in clinical applications.

Generating Protein Profiles from Thin Tissue Sections

A schematic diagram of the general experimental procedure is given in Figure 2 (Chaurand et al., 2004b). After dissection, the tissue sample should be loosely wrapped in aluminum foil, immediately snap-frozen in liquid nitrogen and stored at −80°C until analyzed. Thin frozen sections are typically cut at −15°C using a cryostat (the optimum cutting temperature may be tissue dependent) and thaw-mounted onto a conductive target plate (Schwartz et al., 2003). The tissue sample is maintained with the desired orientation on the cutting block using a medium such as OCT (Optimum Cutting Temperature) polymer. However, it is important to note that OCT is only used to mount the tissue on the cryostat block and should not come into contact with the surface of the tissue to be sectioned. The surface of the tissue should be left available for sectioning free from the polymer. To align tissue features and molecular images, 2 sections are typically used, one is stained (typically with hematoxylin and eosin) for optical evaluation and the second is investigated by IMS. Although this approach is robust, misalignment between the 2 sections is of concern especially for very small tissue samples such as needle biopsies. One of the latest tissue analysis protocols developed utilizes optically transparent glass slides as target plates (which have a thin conductive coating on the surface) together with MALDI MS friendly tissue staining protocols (Chaurand et al., 2004a). This makes possible the microscopic evaluation of a tissue section by a pathologist followed by the molecular imaging of the same section by IMS.

In general, there are two basic modes of data acquisition: profiling and imaging (Figure 2). In the profiling experiment, one is interested in comparing protein patterns from a discrete number of spots or areas. Although there is no spot limitation, typically this is done for 5–20 regions across a given tissue section. Matrix is applied, as discrete droplets (spots) to the regions of interest, and the sample is placed into the MS source. We have found sinapinic acid (saturated in 50/50/0.1—acetonitrile/H₂O/trifluoroacetic acid) to be the matrix of choice for the analysis of proteins (Schwartz et al., 2003). Typically, 200 to 500 nL droplets of matrix are deposited using an automatic pipette covering an area of 2–4 mm². Smaller drop sizes (5–100 nL) can be applied with a fine capillary attached to a Hamilton type syringe containing the matrix. In this case, matrix deposition is generally performed under low to medium magnification to precisely deposit the matrix droplets at the desired tissue coordinates. The laser beam irradiates each sample spot and ion signals from 200–1000 consecutive shots are averaged across the droplet surface generating a mass spectrum. Upon analysis, the resulting mass spectra typically yield from 300 to in some cases over 1000 signals, of various intensities over three to four orders of magnitude, in a m/z range from 2000 up to, in some cases, over m/z 200,000 (Chaurand and Caprioli, 2002). However, because of the inefficiency of MALDI time-of-flight mass spectrometers to resolve (Bahr et al., 1997) and detect higher molecular weight compounds (Brunelle et al., 1997; Westmacott et al., 2000), most signals detected are below m/z 30,000. One may also presume that the most intense signals come essentially from soluble and abundant protein species.

In the imaging experiment (Figure 2), the goal is to display a detailed molecular image of an entire tissue section or a specific subregion. In this case, matrix needs to be homogeneously deposited across the section without generating any major lateral protein migration. In an attempt to “fix” the proteins in a section, the sections mounted on the target plate may be soaked in an ethanol fixing solution containing matrix (sinapinic acid, 20 mg/mL in 90/10/0.1—ethanol/H₂O/trifluoroacetic acid) for 5–10 minutes and allowed to dry at room temperature. The fixing procedure also tends to seed matrix on the sections. Sections are then coated with matrix solution (sinapinic acid, 20 mg/mL in 50/50/0.1—acetonitrile/H₂O/trifluoroacetic acid) using a pneumatic handheld Venturi glass sprayer. At this stage, special care needs to be taken not to “overwet” the section. On average, 10 spray cycles are necessary to achieve a thin, relatively homogeneous, matrix coating that covers the entire tissue section. The coating procedure may be monitored under a microscope to assess crystal size and density (Schwartz et al., 2003). A second coating approach currently developed in our laboratory consists of matrix deposition using an automated spotter (Rapidspotter TM, Labcyte Inc., Sunnyvale, CA), generating a Cartesian microdroplet array over the surface of the section (Aerni et al., 2003, 2004). Although this tissue-coating approach does not currently permit imaging resolution below 200 μm, it has the advantage of limiting protein migration within the surface covered by each microdroplet.

In an imaging experiment, thousands of spots can define the array with each spot associated with a full mass spectrum. Maximum resolution depends primarily on the dimensions of the ionizing laser beam. With commercially available instruments, beam dimensions of about 25–100 μm in diameter are attainable with minimum efforts (Caprioli et al., 1997). Specialized instrument control software is used to set a data acquisition grid that defines a discrete Cartesian pattern across the sample surface (Stoeckli et al., 1999, 2002). This pattern has a fixed center to center distance between spots, typically ranging from 25 to 250 μm depending on the imaging resolution required. The mass spectrometric data is then acquired utilizing this grid pattern with a predetermined number of laser shots per grid coordinate. The signal intensity for a selected m/z value at every acquisition coordinate is then integrated and a 2-dimensional ion density map, or image, is reconstructed. An image can be generated for each of the mass signals detected throughout the section. From a single acquisition, several hundred images, each at a specific molecular weight, can be reconstructed. Data acquisition and processing (image reconstruction) is done with specialized software (Stoeckli et al., 2002).

All of the samples presented here were cut at −15°C in 10–12 μm sections using a Leica CM 3050 S cryostat (Leica Microsystems AG, Welzlar, Germany). Sections were analyzed using sinapinic acid as matrix. Mass spectrometric data were acquired in the linear mode under optimized delayed extraction conditions using a Voyager DE-STR MALDI time-of-flight mass spectrometer (Applied Biosystems, Framingham, MA).

Profiling and Imaging Mammalian Tissue Sections

Examples of protein profiles obtained by MALDI MS from various human cancer tissue biopsies are presented in Figure 3. Tissues were cut in 12 μm sections and processed as described above. Overlaid in Figure 3 are three typical profiles obtained from a grade IV glioma (blue trace), a colorectal adenocarcinoma (red trace) and a breast adenocarcinoma (green trace). Signals were detected throughout the studied m/z range (from 3,000 to 70,000) with the large majority of these below m/z 30,000. A close inspection of these profiles revealed that, although some signals were common to 2 of the cancer forms, very few signals are common to all 3 forms of cancer. Pattern recognition studies could therefore determine protein signals specific for each tissue studied, illustrating the high specificity of the profiles for a given tissue sample.

Preliminary investigations of human tumor biopsies by MALDI MS demonstrate that proteomic patterns can be used to distinguish cancerous tissue from normal tissue as well as subclassify cancers by histological grade (Yanagisawa et al., 2003; Schwartz et al., 2004). In these studies, prospectively collected, snap-frozen normal and tumor specimens from multiple patients were examined using MALDI MS. Peptide and protein expressions were compared and the patterns assessed through hierarchical cluster analysis. In the study from Schwartz et al., the mass spectral patterns could reliably distinguish gliomas from nontumor brain tissue as well as subclassify grade IV gliomas from grades II and III. Figure 4 presents the simultaneous analysis by imaging mass spectrometry of two 12 μm sections obtained from a grade II (low-grade) and a grade IV (high-grade) resected human glioma biopsy (Figure 4b, 4a, respectively). The sections were coated with matrix using the automated spotter, and the images were acquired with a lateral resolution of 250 μm. Figure 4c presents survey protein profiles obtained from the low-grade (red trace) and high-grade glioma (blue trace) samples. In the m/z range displayed, numerous signals were found to be expressed with different intensities in both the low-grade and the high-grade biopsies (highlighted by stars). These differences have been used to subclassify low grade from high grade gliomas (Schwartz et al., 2004). Figure 4d–h presents 5 ion density maps, or images, acquired for different molecular weight signals which displayed strong intensity differences between the low-grade and the high-grade tumor. The mass signals found to statistically discriminate between normal and tumor and the different grades of cancer are currently being identified. Biomarker identification is performed by well-established methods that consist of extraction of the proteins from the tissue followed by protein separation (RP-HPLC, anion exchange, and size exclusion). After screening by MALDI MS, the HPLC fractions containing the targeted molecular weight markers are digested with trypsin and the resulting peptides mapped and sequenced by mass spectrometry. The proteins are identified by interrogating gene or protein databases with the experimentally recovered peptide mass maps and sequences (Chaurand et al., 2001b, 2003; Stoeckli et al., 2001).

Based on signal (or protein) expression measured at precise coordinates across the sections using high resolution IMS, changes in cellular morphology may be distinguished and in some cases, the cellular origins for the different protein signals can be determined. This later point is demonstrated for the signal at m/z 10836 in the previous analysis, identified as S100β protein (Swiss-Prot accession Number P04271), which was found increased by about a factor of 4 in the high-grade sample (Figure 5a). Figure 5b–c presents the corresponding mass spectrometric ion density maps obtained when integrating the signal for the S100β protein at m/z 10836. The images clearly show stronger S100β protein expression in the high-grade tumor with respect to the low-grade tumor. Based on signal expression across the high grade glioma section (Figure 5b), stronger intensity signals for S100β were correlated to regions with higher concentrations of astrocytes. In parallel, S100β protein expression levels between low-grade and high-grade gliomas were also investigated by immunohistochemistry. Figure 5d–e displays magnified photomicrographs of high-grade and low-grade glioma tissue sections (cut from the same biopsies investigated by IMS) after immunoreaction (epifluorescence microscopy). The photomicrographs show a number of astrocytes in the high-grade tumor with pronounced S100β immunoreactivity in the cytosol (arrowhead) and weak immunopositivity in the oligodendrocytes in the low-grade tumor. Similar IMS and immunostaining patterns have been seen when comparing grade IV astrocytomas with grade II astrocytomas.

Imaging mass spectrometry can also be used to measure the precise localization of drugs and metabolites in tissue sections (Reyzer et al., 2003; Troendle et al., 1999). One of the difficulties of imaging lower molecular weight compounds by MALDI mass spectrometry comes from significant interference and overlap with signals from endogenous compounds in the tissue or originating from various matrix cluster ions typically present in the m/z range up to about 2,000. In order to increase selectivity as well as sensitivity, collisionally activated dissociation (CAD) is employed (Reyzer et al., 2003). The general strategy involves mass selection of the intact drug ion, fragmentation of the ion by CAD, and detection of a unique fragment ion. Imaging of drugs on tissue is then accomplished in an analogous fashion to imaging of proteins on tissue. In this case, the intensity of the drug specific fragment ion is plotted as a function of pixel coordinate. Similar software is required to control the x/y movement of the sample stage, to identify the area to be imaged, and to define the imaging resolution.

Imaging drugs by mass spectrometry offers significant advantages over traditional imaging techniques (such as autoradiography and fluorescence spectroscopy). Because intact drugs may be desorbed directly from tissue surfaces, no additional syntheses are required, thus allowing this type of localization analysis to occur earlier in the drug discovery process with less cost. In addition, any confounding pharmacological effects due to a bulky label are eliminated, as are the environmental issues associated with radioactivity. Finally, the molecular specificity of mass spectrometry allows the intact drug to be distinguished from its metabolites that differ in mass. The resulting images generated by mass spectrometry are thus specific for the intact drug, while the independent localization of metabolites may also be determined.

In parallel to drug localization, variations of the proteome induced by the drugs may be investigated as a function of dosage or time to determine their efficacy. This strategy may be extremely useful to assess at very early time points positive (or negative) reactions to therapy. The efficacy of an experimental inhibitor drug against human melanoma was examined via MALDI mass spectrometry. The drug was orally administered to mice bearing human melanoma xenograft tumors at low and high physiological doses (16-fold increase). This compound inhibits one of the steps of the ras gene pathway and has been shown to decrease tumor volume at the doses used here. The mice were dosed for 19 days and the tumors were removed 12 hr after the last dose. Multiple mass spectra from several sections from each sample were acquired and averaged. These are shown in Figure 6. Many signals showed differences in intensities between the 2 measurements, indicating changes in the proteome as a response to dosage in the drug treatment. For example, several regions of the spectrum have been expanded to show signals such as m/z 4738 and m/z 8452 which were significantly reduced from low dose to high dose treatment. Other signals such as m/z 8720 show an inverted trend, with a significant increase in expression after high-dose treatment. These data suggest that changes in multiple protein signals can be used to indicate which tumors will be responsive to drug therapy at in some cases, an early stage.

Perspectives

Imaging mass spectrometry is a new technology that is currently undergoing further development to make it more routinely accessible to users. Imaging time depends on several instrumental parameters, namely the laser repetition rate, spot-to-spot sample repositioning and data processing. Lasers with repetition rates at or above 1 kilohertz and improved electronics will considerably reduce acquisition times from hours to minutes. Acquisition algorithms capable of recording high-throughput data and specially targeted data mining tools are also being developed. Imaging resolution, currently in the 50 μm range for tissue level analysis, may be increased to 1–5 μm for applications requiring subcellular analyses. Such developments are ongoing in our laboratory and elsewhere (Spengler and Hubert, 2002). Efforts to improve sample preparation and matrix coating procedures are also being undertaken to provide protocols that more easily achieve high sensitivity and high resolution images.

The potential of such a molecular imaging technology is considerable. The fundamental contributions of the technology are rapidly providing molecular weight specific profiles and images at relatively high resolution and sensitivity. These provide important information in the investigation of cellular processes in both health and disease. Imaging MS is of extraordinary benefit as a discovery tool because one does not need to know in advance the specific proteins that have changed in a comparative study. Furthermore, the cellular origins and relative concentrations of the markers across the section can be assessed. Once a marker of interest is identified, its precise (subcellular) location, concentration, regulation and function may be investigated, to help understand disease progression at the molecular level. Although current imaging MS technology does not allow analysis of individual cells, we anticipate that new developments will allow this application in the near future.

Clinically, IMS can provide a molecular assessment of tumor progression and treatment obtained from biopsies, with the potential to identify tumor subpopulations and predict patient survival that is not evident based on the cellular phenotype determined histologically. Further, assessment of the efficacy of drug treatment through comparative proteomics is feasible. The information obtained by IMS significantly augments, but does not replace, existing molecular diagnostic tools. Together, these tools promise to promote new discoveries in biology and medicine.