Article pubs.acs.org/jpr
Experimental Models to Study Drug Distributions in Tissue Using MALDI Mass Spectrometry Imaging Á kos Végvári,*,†,‡ Thomas E. Fehniger,‡,§ Melinda Rezeli,† Thomas Laurell,† Balázs Döme,∥,⊥ Bo Jansson,# Charlotte Welinder,▽ and György Marko-Varga†,○ †
Clinical Protein Science & Imaging, Biomedical Center, Department of Measurement Technology and Industrial Electrical Engineering, Lund University, BMC C13, SE-221 84 Lund, Sweden ‡ CREATE Health, Lund University, Biomedical Centre D13, Tornavägen 10, SE-221 84 Lund, Sweden § Institute of Clinical Medicine, Tallinn University of Technology, Akadeemia tee 15, 12618 Tallinn, Estonia ∥ Department of Thoracic Surgery, Medical University of Vienna, Spitalgasse 23, 1090 Vienna, Austria ⊥ Department of Tumor Biology, National Korányi Institute of Pulmonology, Budapest, Hungary # BioInvent Int. AB, Sölvegatan 41, SE-223 70 Lund, Sweden ▽ Department of Oncology, Clinical Sciences, Lund University, Barngatan 2B, SE-221 85 Lund, Sweden ○ Department of Surgery, Tokyo Medical University, 6-7-1 Nishishinjiku Shinjuku-ku Tokyo, 160-0023 Japan ABSTRACT: Requirements for patient safety and improved efficacy are steadily increasing in modern healthcare and are key drivers in modern drug development. New drug characterization assays are central in providing evidence of the specificity and selectivity of drugs. Meeting this need, matrix-assisted laser desorption ionization-mass spectrometry imaging (MALDI-MSI) is used to study drug localization within microenvironmental tissue compartments. Thin sections of human lung tumor and rat xenograft tissues were exposed to pharmaceutical drugs by either spotting or submerging. These drugs, the epidermal growth factor receptor antagonists, erlotinib (Tarceva) and gefitinib (Iressa), and the acetylcholine receptor antagonist, tiotropium, were characterized by microenvironment localization. Intact tissue blocks were also immersed in drug solution, followed by sectioning. MALDI-MSI was then performed using a Thermo MALDI LTQ Orbitrap XL instrument to localize drug-distribution patterns. We propose three MALDI-MSI models measuring drug disposition that have been used to map the selected compounds within tissue compartments of tumors isolated from lung cancer patients. KEYWORDS: lung cancer, adenocarcinoma, MALDI-mass spectrometry imaging, erlotinib, gefitinib, tiotropium
1. INTRODUCTION Today, mass spectrometry imaging (MSI) plays an important role in the pharmaceutical industry in the evaluation of candidate drugs in experimental models. Because MSI is not dependent on supplementary labels to detect administered compounds, the chemical structure of drugs is not modified and the detection of compounds within tissue is consistent with de novo uptake and dispersal. Despite its lower resolving power relative to autoradiography in whole tissue studies, MSI has the unique advantage of delivering the absolute mass identity of whole compounds, compound fragments, and compound metabolites not seen in other imaging technologies.1 However, such assays do not necessarily utilize the entire potential of MSI.2,3 New directions and experimental human models, which are dynamic in nature, for example, combinatorial cancer therapies, are areas of intense research.4−6 A key event in the drug © 2013 American Chemical Society
development process involves the linking of specific disease mechanisms to targets, which is interconnected to the targetdrug binding properties and directly related to the pharmacokinetics and the pharmacodynamic properties of any given drug compound. Clearly, the physicochemical properties of a compound and its behavior in a tissue microenvironment would govern the selectivity and specificity of drug binding to the targeted protein. Translational research is vital in addressing and optimizing these properties; underlining the step going from animal disease models to man is probably the most crucial in the drug development process. Analyzing tissues by a direct measurement using matrix-assisted laser desorption ionizationmass spectrometry imaging (MALDI-MSI) is a technology that is developing rapidly. MALDI-MSI provides an accurate Received: June 19, 2013 Published: October 17, 2013 5626
dx.doi.org/10.1021/pr400581b | J. Proteome Res. 2013, 12, 5626−5633
Journal of Proteome Research
Article
sections on a Leica CM1950 cryostat (Leica Microsystems, Wetzlar, Germany) at −20 °C. Xenograft Tumor. The human lung adenocarcinoma cell line A549 was purchased from the ATCC (Rockville, MD) and grown according to manufacturer’s instructions. SCID mice were inoculated subcutaneously on the hind leg with 0.1 mL of cell suspensions of (1 to 2) × 106 viable A549 lung adenocarcinoma cells. Tumors were allowed to grow to the size of ∼8 × 8 mm; then, the animals were sacrificed and tumors were snap frozen. The animal experiments were approved by the local ethics committee.
identification of drugs administered to patients as well as endogenous proteins, peptides, and lipids.7−10 We have chosen to identify specific drug-target protein interactions in tissue sections isolated from disease regions in human as well as in humanized mouse models. Both tyrosine kinase inhibitors (TKIs) and a muscarinic receptor antagonist, which were deposited evenly in confined spots on tissue surfaces, showed unique distributions by MALDI-MSI on the different tumor types.9 In a recent study in human subjects we have also shown that inhaled ipratropium localizes to smooth muscle bundle compartments within airway wall shortly after exposure.11 Together with studies using inhaled tiotropium conducted in vivo, in the experimental models we have now shown that the drug distributions could be mapped within either biopsies or in whole organs using MALDI-MSI.8 Lung cancer patients characteristically overexpress epidermal growth factor receptor (EGFR), which is also associated with mutations and is a major strategy operating personalized medicine treatments.12,13 The somatic mutations of EGFR have a key role in the treatment of nonsmall cell lung cancer (NSCLC) with improved outcomes using TKIs, gefitinib and erlotinib.14,15 While gefitinib is mostly effective in patients with mutations in the EGFR kinase domain and to a lesser extent EGFR amplification, erlotinib is typically given to NSCLC patients without achieving complete response despite an initial response rate of 75%.16 The present study displays three experimental models to investigate localization of drugs upon exposition to tissue, in which EGFR TKIs and a muscarinic receptor antagonist (also known as acetylcholine receptor antagonist) are used in NSCLC. Moreover, we demonstrate high expression of the drug target by immunohistochemistry that correlates with the compound localization.
2.3. Sample Preparation and Histology
Tissue Incubation in Drug Solutions. For the concentration gradient and dispersion models, whole tumors or tissue sections were immersed in a solution of drug compounds. For the concentration gradient model, a solution composed of tiotropium (0.05 mg/mL) and erlotinib (0.4 mg/mL) was used to immerse tissue blocks. For the dispersion model, both tiotropium and erlotinib in 50% MeOH at 0.5 mg/mL concentration was used to immerse freshly cut tissue sections on slides. Following incubation for 150 and 60 min using concentration gradient models and dispersion, respectively, the whole tumors and tissue sections were rinsed three times with 4 mL of PBS and three times with 4 mL of water, consecutively. Drug Deposition. For the directed dosage model, drug compounds were dissolved in pure MeOH at 1 mg/mL stock concentration and stored at 4 °C. Following dilution into 50% MeOH, these solutions were deposited onto tissue sections in several cycles with the aid of an in-house designed microdispenser-based spotter apparatus.17,18 The spotter was set to deposit a single droplet of drug solution at a rate of 50 Hz at each position of a rectangular array with a 200 μm raster size. The signal generator was operated with a piezo actuation signal of 100 μs pulse duration, amplitude 17 V, around 5.7 μs rise time, and 100 μs fall time. Matrix Deposition. The matrix solution, 7.5 mg/mL CHCA in 50% ACN/0.1% trifluoroacetic acid, was sprayed onto the tissue sections spotted with drug compounds using an Aztek airbrush model A4709 (Testor, Rockford, IL). Immunohistochemistry. Human lung adenocarcinomas tumor sections were dried for 15 min at 37 °C and fixed with 100% MeOH. Before staining, sections were rehydrated in TrisHCl, pH 7.4 buffer containing 5% bovine serum albumin (Sigma, Steinheim, Germany). Tissue sections were stained with mouse antihuman-EGFR antibody (clone 2-18C9, PharmDx, Dako Sweden AB, Stockholm), and mouse IgG1 (Dako Sweden AB, Stockholm) was used as a negative control for 1 h at room temperature, followed by EnVision (Dako Sweden AB, Stockholm). Slides were counterstained with hematoxylin before dehydration and mounting. Histological Staining. Tissue sections were also stained with conventional Mayers hematoxylin and eosin (H&E) staining. Images of stained sections were acquired by a digital microscope slide scanner (Zeiss, Germany, Mirax MIDI).
2. MATERIALS AND METHODS 2.1. Chemicals
Methanol (MeOH) at HPLC grade (99% purity, were purchased from LC Laboratories (Woburn, MA). The anticholinergic/antimuscarinic bronchodilator compound, (1α,2β,4β,7β)-7-[(hydroxidi-2thienylacetyl)oxy]-9,9-dimethyl-3-oxa-9-azoniatricyclo[3.3.1.02,4]nonane (CAS number 136310-93-5; tiotropium bromide), was a kind gift of Dr. Lena Gustavsson (AstraZeneca R&D, Lund, Sweden). 2.2. Clinical and Animal Material
Human Lung Adenocarcinoma Tumor. The study was conducted in accordance with the European Communities Council Directive of November 24, 1986 (86/609/EEC) and was approved by the Hungarian Ethics Committee (2521-0 2010-1018EKU). The lung tissues were surgically removed and immediately after slowly frozen by placement for 2 min on a plastic boat floating in a bath of isopentane that was supercooled with dry ice (−70 °C). The lung tissues were then stored at −70 °C until sectioning into 10 μm thick
2.4. MALDI Mass Spectrometry Imaging
Imaging of drug compounds was performed on a MALDI LTQ Orbitrap XL mass spectrometer (Thermo Fisher Scientific, Bremen, Germany). Full mass spectra were obtained by using the Orbitrap mass analyzer, sampling the tissue sections in 150−500 Da mass range in positive mode using 10 laser shots at 10.0 μJ with automatic gain control off. The distance between measuring points was 50 or 100 μm. MS/MS data was 5627
dx.doi.org/10.1021/pr400581b | J. Proteome Res. 2013, 12, 5626−5633
Journal of Proteome Research
Article
Figure 1. Illustration of the principles of the experimental models for use of MALDI-MSI. (The localized approach was introduced previously in ref 9.)
Figure 2. Demonstration of the concentration gradient model using a mixture of two drug compounds. Similar spatial localization of erlotinib (A) and tiotropium (B) was observed within the same tissue section by mapping corresponding parent masses at 100 μm spatial resolution. The IHC section, showing the tumor cell dense area as highlighted, has characteristically lower signal intensities of the drugs (C).
other. Here the potential for MALDI-MSI to play an important role in aiding the selection of the compound, which has the most favorable properties in terms of pharmacokinetics/ pharmacodynamics (PK/PD), as well as selectivity and specificity was demonstrated.
collected in the linear ion trap analyzer utilized to fragmentize the parent ions of erlotinib (m/z 394.177), gefitinib (m/z 447.162), and tiotropium (m/z 392.098) at normal scan rate, isolating the ions in m/z 2.0 width. Normalized collision energy was 50% during an activation time of 30 ms, and activation Q of 0.250 was applied with wideband activation. The minimal signal required for MS/MS spectra generation was 500 counts. The visualization of the drug fragment ions was performed with the ImageQuest software (Thermo Fisher Scientific, San Jose, CA).
3.1. Concentration Gradient Model
Diffusion and other transport mechanisms may play important roles in drug uptake, which was addressed in the development of this in vitro model. Solid xenograft tumor (lung adenocarcinoma A549) tissue was immersed in a solution of drug mixture containing tiotropium and erlotinib for 2.5 h at room temperature. After the tissue blocks were rinsed with PBS and water, followed by freezing the tissue at −20 °C, 10 μm thin sections were sectioned. MALDI-MSI analysis in full mass scan mode with a 100 μm raster size was performed following matrix depositions. Mapping tiotropium and erlotinib m/z
3. RESULTS The concept of the pharmacokinetic models that we present in this study is illustrated in Figure 1 displaying: (1) the concentration gradient model, (2) the dispersion model, and (3) the directed dosage model, the latter which was previously introduced.9 These experimental models were considered as complementary rather than one being in preference of the 5628
dx.doi.org/10.1021/pr400581b | J. Proteome Res. 2013, 12, 5626−5633
Journal of Proteome Research
Article
Figure 3. Dispersion model showed the localization of erlotinib (red) and tiotropium (green) on a human adenocarcinoma tissue section shown by the signal intensity distributions of their most intense fragment ions, m/z 336.2 (red) and 152.1 (green), respectively (A). MALDI combined image of erlotinib (red) and tiotropium (green) before (A) and after washing the section (B) following drug adsorption on the tissue. H&E staining of the same section following the removal of matrix (C). The spatial localization of gefitinib depicted by its parent ion (m/z 447.162) (D) and fragment ion (m/z 128.1) (E) on human adenocarcinoma tissue section.
somatic mutations in the EGFR.14,19 Kinetically, both erlotinib and gefitinib behaved similarly in these studies in resected adenocarcinoma specimens, which were not characterized for mutant phenotyping but had high expression of EGFR in general. In addition, we also investigated tiotropium mixed with TKIs. Tiotropium has an affinity to three of the muscarinic receptor subtypes on specific cell types located within conducting central airways and within alveolar walls of the lung.20,21 Many primary lung cancer patients being treated with anti-EGFR kinase inhibitors are also treated for COPD with these types of bronchodilators. The initial distribution of erlotinib and tiotropium within the tissue section is shown in Figure 3A, displaying the drug localization on the tissue surface before washing. Erlotinib (red) and tiotropium (green) can be seen differentially distributed in the tissue in patterns of the two drugs that partially overlap, as illustrated by the yellow color on the image in Figure 3A. Following washing, the distribution of erlotinib remained consistent with the initial positioning (Figure 3B), localizing well to the area of tumor cells identified by the pathology unit of the hospital. The local distribution of the fragment ion of erlotinib (m/z 336.2) using MS/MS mode of data acquisition provided unambiguous identification of the compound. The tumor region could also be seen to have a high cell density in
signals revealed that the compounds could indeed penetrate into the tissue by passive diffusion. Typically, a gradient-like distribution was observed with significantly higher signal intensities at the edge of the tissue sections, that is, the surface of the tissue block, as shown in Figure 2A,B. Besides the profound differences in signal intensities of tiotropium and erlotinib, similarities in spatial localization of these compounds were also observed. The concentration gradient model could be useful to follow the efficiency of transport of drugs through various cell types. Furthermore, the velocity could be estimated by taking snapshots of drug distributions at various time points. In agreement with our previous findings, the signal intensity of drugs from tumor cells was substantially lower than that from stroma,9 as revealed by the comparison with IHC, with stained adjacent sections (Figure 2C). 3.2. Dispersion Model
The dispersion model utilizes cryostat sections isolated from clinical samples, such as resections and biopsies. The tissue sections of lung tumors used in this model were immersed into the drug solution, where either adsorption or adsorption/ absorption occurs. The time period for drug exposure was typically 1 h. We investigated two different TKIs, erlotinib and gefitinib, which are globally used by NSCLC patients with 5629
dx.doi.org/10.1021/pr400581b | J. Proteome Res. 2013, 12, 5626−5633
Journal of Proteome Research
Article
Figure 4. Directed dosage model demonstrated MALDI-MS readout of erlotinib (m/z 394.178) spotted onto tissue surface and binding to the tumor region of the tissue (A) as compared with the immunohistochemical staining of EGFR (B) and also shown as overlaid image of MALDI and IHC (C). Tiotropium signal (m/z 392.098) was distributed following the areas of tumor cells in the adenocarcinoma section (D). Stained image showing the IgG1 as negative control (E). H&E histology staining (F).
by their ionization properties, and facilitated finding the experimental conditions suitable to our instrumentation. Figure 4 shows examples of gefitinib distribution on human adenocarcinoma tissues as well as it provides an example of how the drug deposition array appeared on the glass slide. It was also clear from these images that the ionization properties of this TKI drug overlapped well with the immunohistochemical images where the EGFR was stained. (See Figure 4C.) In further analyses, both erlotinib and tiotropium in a mixture were spotted on dried tissue sections at 50 and 100 ng/ L, respectively. The spots were evenly distributed with a 200 μm raster size in a rectangular array, typically distributing 10− 20−50 pg (deposition cycles 2, 4, and 10) drug within 150 μm circular spots on the entire tissue surface. The most intense fragment ion signals of erlotinib (m/z 336.2 and 278.2) were identified and used as direct support of the parent ion (m/z 394.178) location. A typical result obtained with erlotinib on adenocarcinoma tissue section collecting MS/MS fragmentation data revealed uneven but matching distributions of both parent and fragment ions. Importantly, in agreement with our previous findings, lower signal intensities were obtained from areas of tumor cells indicated by high EGFR expression. The impact of these two different drug agents appeared to be significantly different, as tiotropium was not more detectable by higher intensities but also more evenly distributed over the tissue section. The variation in the signal intensity of tiotropium seemed to follow the macroscopic structure of tissue, including holes of airways, whereas the localization of erlotinib was dictated by tumorous areas.
relation to the entire section, as H&E histology staining provides evidence using the very same tissue section (Figure 3C). In a similar investigation, gefitinib was sampled on a tumor section isolated after surgery from an adenocarcinoma patient, displaying efficient signals over the entire section (Figure 3D,E). The reproducibility using adjacent sections of the same tissue was good in terms of distribution and signal intensities of drug compounds. 3.3. Directed Dosage Model
The directed dosage model allows for a guided deposition of the drug to any microenvironment area of the patient tissue, as was previously introduced in investigation of drug localization in cancer tissue sections.9 High-frequency piezo-dispensing spotted the drug onto the tissue surface, ejecting 100 pL drug sample droplets. The accuracy of the localization onto a specific region of the tissue could be achieved with
Experimental Models to Study Drug Distributions in Tissue Using MALDI Mass Spectrometry Imaging Á kos Végvári,*,†,‡ Thomas E. Fehniger,‡,§ Melinda Rezeli,† Thomas Laurell,† Balázs Döme,∥,⊥ Bo Jansson,# Charlotte Welinder,▽ and György Marko-Varga†,○ †
Clinical Protein Science & Imaging, Biomedical Center, Department of Measurement Technology and Industrial Electrical Engineering, Lund University, BMC C13, SE-221 84 Lund, Sweden ‡ CREATE Health, Lund University, Biomedical Centre D13, Tornavägen 10, SE-221 84 Lund, Sweden § Institute of Clinical Medicine, Tallinn University of Technology, Akadeemia tee 15, 12618 Tallinn, Estonia ∥ Department of Thoracic Surgery, Medical University of Vienna, Spitalgasse 23, 1090 Vienna, Austria ⊥ Department of Tumor Biology, National Korányi Institute of Pulmonology, Budapest, Hungary # BioInvent Int. AB, Sölvegatan 41, SE-223 70 Lund, Sweden ▽ Department of Oncology, Clinical Sciences, Lund University, Barngatan 2B, SE-221 85 Lund, Sweden ○ Department of Surgery, Tokyo Medical University, 6-7-1 Nishishinjiku Shinjuku-ku Tokyo, 160-0023 Japan ABSTRACT: Requirements for patient safety and improved efficacy are steadily increasing in modern healthcare and are key drivers in modern drug development. New drug characterization assays are central in providing evidence of the specificity and selectivity of drugs. Meeting this need, matrix-assisted laser desorption ionization-mass spectrometry imaging (MALDI-MSI) is used to study drug localization within microenvironmental tissue compartments. Thin sections of human lung tumor and rat xenograft tissues were exposed to pharmaceutical drugs by either spotting or submerging. These drugs, the epidermal growth factor receptor antagonists, erlotinib (Tarceva) and gefitinib (Iressa), and the acetylcholine receptor antagonist, tiotropium, were characterized by microenvironment localization. Intact tissue blocks were also immersed in drug solution, followed by sectioning. MALDI-MSI was then performed using a Thermo MALDI LTQ Orbitrap XL instrument to localize drug-distribution patterns. We propose three MALDI-MSI models measuring drug disposition that have been used to map the selected compounds within tissue compartments of tumors isolated from lung cancer patients. KEYWORDS: lung cancer, adenocarcinoma, MALDI-mass spectrometry imaging, erlotinib, gefitinib, tiotropium
1. INTRODUCTION Today, mass spectrometry imaging (MSI) plays an important role in the pharmaceutical industry in the evaluation of candidate drugs in experimental models. Because MSI is not dependent on supplementary labels to detect administered compounds, the chemical structure of drugs is not modified and the detection of compounds within tissue is consistent with de novo uptake and dispersal. Despite its lower resolving power relative to autoradiography in whole tissue studies, MSI has the unique advantage of delivering the absolute mass identity of whole compounds, compound fragments, and compound metabolites not seen in other imaging technologies.1 However, such assays do not necessarily utilize the entire potential of MSI.2,3 New directions and experimental human models, which are dynamic in nature, for example, combinatorial cancer therapies, are areas of intense research.4−6 A key event in the drug © 2013 American Chemical Society
development process involves the linking of specific disease mechanisms to targets, which is interconnected to the targetdrug binding properties and directly related to the pharmacokinetics and the pharmacodynamic properties of any given drug compound. Clearly, the physicochemical properties of a compound and its behavior in a tissue microenvironment would govern the selectivity and specificity of drug binding to the targeted protein. Translational research is vital in addressing and optimizing these properties; underlining the step going from animal disease models to man is probably the most crucial in the drug development process. Analyzing tissues by a direct measurement using matrix-assisted laser desorption ionizationmass spectrometry imaging (MALDI-MSI) is a technology that is developing rapidly. MALDI-MSI provides an accurate Received: June 19, 2013 Published: October 17, 2013 5626
dx.doi.org/10.1021/pr400581b | J. Proteome Res. 2013, 12, 5626−5633
Journal of Proteome Research
Article
sections on a Leica CM1950 cryostat (Leica Microsystems, Wetzlar, Germany) at −20 °C. Xenograft Tumor. The human lung adenocarcinoma cell line A549 was purchased from the ATCC (Rockville, MD) and grown according to manufacturer’s instructions. SCID mice were inoculated subcutaneously on the hind leg with 0.1 mL of cell suspensions of (1 to 2) × 106 viable A549 lung adenocarcinoma cells. Tumors were allowed to grow to the size of ∼8 × 8 mm; then, the animals were sacrificed and tumors were snap frozen. The animal experiments were approved by the local ethics committee.
identification of drugs administered to patients as well as endogenous proteins, peptides, and lipids.7−10 We have chosen to identify specific drug-target protein interactions in tissue sections isolated from disease regions in human as well as in humanized mouse models. Both tyrosine kinase inhibitors (TKIs) and a muscarinic receptor antagonist, which were deposited evenly in confined spots on tissue surfaces, showed unique distributions by MALDI-MSI on the different tumor types.9 In a recent study in human subjects we have also shown that inhaled ipratropium localizes to smooth muscle bundle compartments within airway wall shortly after exposure.11 Together with studies using inhaled tiotropium conducted in vivo, in the experimental models we have now shown that the drug distributions could be mapped within either biopsies or in whole organs using MALDI-MSI.8 Lung cancer patients characteristically overexpress epidermal growth factor receptor (EGFR), which is also associated with mutations and is a major strategy operating personalized medicine treatments.12,13 The somatic mutations of EGFR have a key role in the treatment of nonsmall cell lung cancer (NSCLC) with improved outcomes using TKIs, gefitinib and erlotinib.14,15 While gefitinib is mostly effective in patients with mutations in the EGFR kinase domain and to a lesser extent EGFR amplification, erlotinib is typically given to NSCLC patients without achieving complete response despite an initial response rate of 75%.16 The present study displays three experimental models to investigate localization of drugs upon exposition to tissue, in which EGFR TKIs and a muscarinic receptor antagonist (also known as acetylcholine receptor antagonist) are used in NSCLC. Moreover, we demonstrate high expression of the drug target by immunohistochemistry that correlates with the compound localization.
2.3. Sample Preparation and Histology
Tissue Incubation in Drug Solutions. For the concentration gradient and dispersion models, whole tumors or tissue sections were immersed in a solution of drug compounds. For the concentration gradient model, a solution composed of tiotropium (0.05 mg/mL) and erlotinib (0.4 mg/mL) was used to immerse tissue blocks. For the dispersion model, both tiotropium and erlotinib in 50% MeOH at 0.5 mg/mL concentration was used to immerse freshly cut tissue sections on slides. Following incubation for 150 and 60 min using concentration gradient models and dispersion, respectively, the whole tumors and tissue sections were rinsed three times with 4 mL of PBS and three times with 4 mL of water, consecutively. Drug Deposition. For the directed dosage model, drug compounds were dissolved in pure MeOH at 1 mg/mL stock concentration and stored at 4 °C. Following dilution into 50% MeOH, these solutions were deposited onto tissue sections in several cycles with the aid of an in-house designed microdispenser-based spotter apparatus.17,18 The spotter was set to deposit a single droplet of drug solution at a rate of 50 Hz at each position of a rectangular array with a 200 μm raster size. The signal generator was operated with a piezo actuation signal of 100 μs pulse duration, amplitude 17 V, around 5.7 μs rise time, and 100 μs fall time. Matrix Deposition. The matrix solution, 7.5 mg/mL CHCA in 50% ACN/0.1% trifluoroacetic acid, was sprayed onto the tissue sections spotted with drug compounds using an Aztek airbrush model A4709 (Testor, Rockford, IL). Immunohistochemistry. Human lung adenocarcinomas tumor sections were dried for 15 min at 37 °C and fixed with 100% MeOH. Before staining, sections were rehydrated in TrisHCl, pH 7.4 buffer containing 5% bovine serum albumin (Sigma, Steinheim, Germany). Tissue sections were stained with mouse antihuman-EGFR antibody (clone 2-18C9, PharmDx, Dako Sweden AB, Stockholm), and mouse IgG1 (Dako Sweden AB, Stockholm) was used as a negative control for 1 h at room temperature, followed by EnVision (Dako Sweden AB, Stockholm). Slides were counterstained with hematoxylin before dehydration and mounting. Histological Staining. Tissue sections were also stained with conventional Mayers hematoxylin and eosin (H&E) staining. Images of stained sections were acquired by a digital microscope slide scanner (Zeiss, Germany, Mirax MIDI).
2. MATERIALS AND METHODS 2.1. Chemicals
Methanol (MeOH) at HPLC grade (99% purity, were purchased from LC Laboratories (Woburn, MA). The anticholinergic/antimuscarinic bronchodilator compound, (1α,2β,4β,7β)-7-[(hydroxidi-2thienylacetyl)oxy]-9,9-dimethyl-3-oxa-9-azoniatricyclo[3.3.1.02,4]nonane (CAS number 136310-93-5; tiotropium bromide), was a kind gift of Dr. Lena Gustavsson (AstraZeneca R&D, Lund, Sweden). 2.2. Clinical and Animal Material
Human Lung Adenocarcinoma Tumor. The study was conducted in accordance with the European Communities Council Directive of November 24, 1986 (86/609/EEC) and was approved by the Hungarian Ethics Committee (2521-0 2010-1018EKU). The lung tissues were surgically removed and immediately after slowly frozen by placement for 2 min on a plastic boat floating in a bath of isopentane that was supercooled with dry ice (−70 °C). The lung tissues were then stored at −70 °C until sectioning into 10 μm thick
2.4. MALDI Mass Spectrometry Imaging
Imaging of drug compounds was performed on a MALDI LTQ Orbitrap XL mass spectrometer (Thermo Fisher Scientific, Bremen, Germany). Full mass spectra were obtained by using the Orbitrap mass analyzer, sampling the tissue sections in 150−500 Da mass range in positive mode using 10 laser shots at 10.0 μJ with automatic gain control off. The distance between measuring points was 50 or 100 μm. MS/MS data was 5627
dx.doi.org/10.1021/pr400581b | J. Proteome Res. 2013, 12, 5626−5633
Journal of Proteome Research
Article
Figure 1. Illustration of the principles of the experimental models for use of MALDI-MSI. (The localized approach was introduced previously in ref 9.)
Figure 2. Demonstration of the concentration gradient model using a mixture of two drug compounds. Similar spatial localization of erlotinib (A) and tiotropium (B) was observed within the same tissue section by mapping corresponding parent masses at 100 μm spatial resolution. The IHC section, showing the tumor cell dense area as highlighted, has characteristically lower signal intensities of the drugs (C).
other. Here the potential for MALDI-MSI to play an important role in aiding the selection of the compound, which has the most favorable properties in terms of pharmacokinetics/ pharmacodynamics (PK/PD), as well as selectivity and specificity was demonstrated.
collected in the linear ion trap analyzer utilized to fragmentize the parent ions of erlotinib (m/z 394.177), gefitinib (m/z 447.162), and tiotropium (m/z 392.098) at normal scan rate, isolating the ions in m/z 2.0 width. Normalized collision energy was 50% during an activation time of 30 ms, and activation Q of 0.250 was applied with wideband activation. The minimal signal required for MS/MS spectra generation was 500 counts. The visualization of the drug fragment ions was performed with the ImageQuest software (Thermo Fisher Scientific, San Jose, CA).
3.1. Concentration Gradient Model
Diffusion and other transport mechanisms may play important roles in drug uptake, which was addressed in the development of this in vitro model. Solid xenograft tumor (lung adenocarcinoma A549) tissue was immersed in a solution of drug mixture containing tiotropium and erlotinib for 2.5 h at room temperature. After the tissue blocks were rinsed with PBS and water, followed by freezing the tissue at −20 °C, 10 μm thin sections were sectioned. MALDI-MSI analysis in full mass scan mode with a 100 μm raster size was performed following matrix depositions. Mapping tiotropium and erlotinib m/z
3. RESULTS The concept of the pharmacokinetic models that we present in this study is illustrated in Figure 1 displaying: (1) the concentration gradient model, (2) the dispersion model, and (3) the directed dosage model, the latter which was previously introduced.9 These experimental models were considered as complementary rather than one being in preference of the 5628
dx.doi.org/10.1021/pr400581b | J. Proteome Res. 2013, 12, 5626−5633
Journal of Proteome Research
Article
Figure 3. Dispersion model showed the localization of erlotinib (red) and tiotropium (green) on a human adenocarcinoma tissue section shown by the signal intensity distributions of their most intense fragment ions, m/z 336.2 (red) and 152.1 (green), respectively (A). MALDI combined image of erlotinib (red) and tiotropium (green) before (A) and after washing the section (B) following drug adsorption on the tissue. H&E staining of the same section following the removal of matrix (C). The spatial localization of gefitinib depicted by its parent ion (m/z 447.162) (D) and fragment ion (m/z 128.1) (E) on human adenocarcinoma tissue section.
somatic mutations in the EGFR.14,19 Kinetically, both erlotinib and gefitinib behaved similarly in these studies in resected adenocarcinoma specimens, which were not characterized for mutant phenotyping but had high expression of EGFR in general. In addition, we also investigated tiotropium mixed with TKIs. Tiotropium has an affinity to three of the muscarinic receptor subtypes on specific cell types located within conducting central airways and within alveolar walls of the lung.20,21 Many primary lung cancer patients being treated with anti-EGFR kinase inhibitors are also treated for COPD with these types of bronchodilators. The initial distribution of erlotinib and tiotropium within the tissue section is shown in Figure 3A, displaying the drug localization on the tissue surface before washing. Erlotinib (red) and tiotropium (green) can be seen differentially distributed in the tissue in patterns of the two drugs that partially overlap, as illustrated by the yellow color on the image in Figure 3A. Following washing, the distribution of erlotinib remained consistent with the initial positioning (Figure 3B), localizing well to the area of tumor cells identified by the pathology unit of the hospital. The local distribution of the fragment ion of erlotinib (m/z 336.2) using MS/MS mode of data acquisition provided unambiguous identification of the compound. The tumor region could also be seen to have a high cell density in
signals revealed that the compounds could indeed penetrate into the tissue by passive diffusion. Typically, a gradient-like distribution was observed with significantly higher signal intensities at the edge of the tissue sections, that is, the surface of the tissue block, as shown in Figure 2A,B. Besides the profound differences in signal intensities of tiotropium and erlotinib, similarities in spatial localization of these compounds were also observed. The concentration gradient model could be useful to follow the efficiency of transport of drugs through various cell types. Furthermore, the velocity could be estimated by taking snapshots of drug distributions at various time points. In agreement with our previous findings, the signal intensity of drugs from tumor cells was substantially lower than that from stroma,9 as revealed by the comparison with IHC, with stained adjacent sections (Figure 2C). 3.2. Dispersion Model
The dispersion model utilizes cryostat sections isolated from clinical samples, such as resections and biopsies. The tissue sections of lung tumors used in this model were immersed into the drug solution, where either adsorption or adsorption/ absorption occurs. The time period for drug exposure was typically 1 h. We investigated two different TKIs, erlotinib and gefitinib, which are globally used by NSCLC patients with 5629
dx.doi.org/10.1021/pr400581b | J. Proteome Res. 2013, 12, 5626−5633
Journal of Proteome Research
Article
Figure 4. Directed dosage model demonstrated MALDI-MS readout of erlotinib (m/z 394.178) spotted onto tissue surface and binding to the tumor region of the tissue (A) as compared with the immunohistochemical staining of EGFR (B) and also shown as overlaid image of MALDI and IHC (C). Tiotropium signal (m/z 392.098) was distributed following the areas of tumor cells in the adenocarcinoma section (D). Stained image showing the IgG1 as negative control (E). H&E histology staining (F).
by their ionization properties, and facilitated finding the experimental conditions suitable to our instrumentation. Figure 4 shows examples of gefitinib distribution on human adenocarcinoma tissues as well as it provides an example of how the drug deposition array appeared on the glass slide. It was also clear from these images that the ionization properties of this TKI drug overlapped well with the immunohistochemical images where the EGFR was stained. (See Figure 4C.) In further analyses, both erlotinib and tiotropium in a mixture were spotted on dried tissue sections at 50 and 100 ng/ L, respectively. The spots were evenly distributed with a 200 μm raster size in a rectangular array, typically distributing 10− 20−50 pg (deposition cycles 2, 4, and 10) drug within 150 μm circular spots on the entire tissue surface. The most intense fragment ion signals of erlotinib (m/z 336.2 and 278.2) were identified and used as direct support of the parent ion (m/z 394.178) location. A typical result obtained with erlotinib on adenocarcinoma tissue section collecting MS/MS fragmentation data revealed uneven but matching distributions of both parent and fragment ions. Importantly, in agreement with our previous findings, lower signal intensities were obtained from areas of tumor cells indicated by high EGFR expression. The impact of these two different drug agents appeared to be significantly different, as tiotropium was not more detectable by higher intensities but also more evenly distributed over the tissue section. The variation in the signal intensity of tiotropium seemed to follow the macroscopic structure of tissue, including holes of airways, whereas the localization of erlotinib was dictated by tumorous areas.
relation to the entire section, as H&E histology staining provides evidence using the very same tissue section (Figure 3C). In a similar investigation, gefitinib was sampled on a tumor section isolated after surgery from an adenocarcinoma patient, displaying efficient signals over the entire section (Figure 3D,E). The reproducibility using adjacent sections of the same tissue was good in terms of distribution and signal intensities of drug compounds. 3.3. Directed Dosage Model
The directed dosage model allows for a guided deposition of the drug to any microenvironment area of the patient tissue, as was previously introduced in investigation of drug localization in cancer tissue sections.9 High-frequency piezo-dispensing spotted the drug onto the tissue surface, ejecting 100 pL drug sample droplets. The accuracy of the localization onto a specific region of the tissue could be achieved with
