Article Cite This: Anal. Chem. 2019, 91, 7819−7827
pubs.acs.org/ac
Spatially Defined Surface Sampling Capillary Electrophoresis Mass Spectrometry Kyle D. Duncan and Ingela Lanekoff* Department of Chemistry-BMC, Uppsala University, Uppsala 751 24, Sweden
Downloaded via KEAN UNIV on July 22, 2019 at 14:04:13 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
S Supporting Information *
ABSTRACT: Capillary electrophoresis mass spectrometry (CE-MS) is an established technique for targeted and untargeted analysis of metabolites from complex biological samples. However, current CE-MS devices rely on liquid sample extracts, which restricts acquisition of spatially defined chemical information from tissue samples. The ability to chemically profile distinct cellular regions in tissue can contribute better understanding to molecular foundations in health and disease. Therefore, we describe the first CE-MS device capable of untargeted metabolite profiling directly from defined morphological regions of solid tissue sections. With surface sampling capillary electrophoresis mass spectrometry (SS-CE-MS), endogenous molecules are sampled and detected from a single defined tissue location. Characterization of SS-CEMS from different locations of the outer epidermal layer of A. Cepa demonstrated reproducible relative migration times and a peak area RSD of 20% (n = 5). Further, relative migration times were conserved for endogenous metabolites in tissues with varying complexities, including brain, spinal cord, and kidney. Results from proof-of-principle experiments from distinct morphological tissue regions reveal simultaneous analysis of small and large biomolecules, confident metabolite annotation, identification of in-source fragmentation interferences, and discrete isomeric abundances related to biological function. We envision that this new tool will provide in-depth chemical profiling and annotation of molecules in distinct cellular regions of tissue for improved biological understanding.
C
disease pathology, effects of pharmaceutical and xenobiotic exposure, and fundamental biological processes. To enable deeper chemical interrogation of morphological features, tissue regions of interest can be microdissected for subsequent chromatographic analysis.10,11 For example, Zimmerman et al. used laser microdissection followed by targeted LC-MS/MS to study the penetration of ethambutol, a drug candidate for tuberculosis treatment, in healthy and necrotic lesions.12 Additionally, Quanico and co-workers applied microdissection LC-MS/MS for multiomic studies of morphological regions in rat brain tissue to elucidate pathways associated with the respective biological functions.13 However, microdissected tissue requires precise dissection and sample preparation procedures that limit method throughput and can introduce a sampling bias for the extracted compounds. An alternative approach to microdissection followed by sample preparation is to sample the molecules directly from defined regions of tissue sections. This can be performed by ionizing all molecules simultaneously from small areas, ranging from submicrometer to several hundred micrometers, on thin tissue sections using mass spectrometry imaging modalities. Mass spectrometry
apillary electrophoresis (CE) is a powerful technique for analyzing molecules in complex biological samples. When coupled to mass spectrometry (MS), hundreds to thousands of analytes can be separated and detected within the same electropherogram. Further, as highlighted in a recent review, CE-MS can be applied to analyze metabolites in a vast array of biological samples, ranging from biofluids to extracted tissue samples.1 A main advantage of CE-MS metabolomic analysis is the ability to analyze very small (e.g., nL) sample volumes. As a result, CE-MS analysis has enabled untargeted metabolomic profiling of extremely low volume biological samples such as small aliquots of infant sweat for the investigation of metabolomic biomarkers of cystic fibrosis,2 low mass skeletal muscle tissue biopsy extracts,3 and the extracts of single cells to explore small molecules involved in embryonic development.4 Additional benefits of CE-MS metabolite profiling include a high separation capacity, an ability to separate both small polar metabolites not amenable to reversed phase liquid chromatography (LC) and charged hydrophobic lipid species, and the potential for increased throughput with multisegment injections.2,5−9 However, all current CE-MS methods require liquid extracts for metabolite profiling, which limits the ability to obtain molecular information from morphologically distinct regions in complex tissue samples. Regionally defined analysis of metabolites, lipids, peptides, and proteins in tissue sections can provide critical insight into © 2019 American Chemical Society
Received: March 26, 2019 Accepted: May 24, 2019 Published: May 24, 2019 7819
DOI: 10.1021/acs.analchem.9b01516 Anal. Chem. 2019, 91, 7819−7827
Article
Analytical Chemistry
Figure 1. Instrumental schematic detailing the setup and operation of the SS-CE-MS device.
metabolites to polypeptides from defined tissue regions. Further, we identify and characterize in-source fragmentation interferences and detail separation of isomeric and isobaric species from morphological regions in kidney and brain tissue. To our knowledge, this is the first technique demonstrating online CE-MS for untargeted metabolite analysis directly from tissue sections.
imaging techniques can be viewed as direct infusion mass spectrometry. While the MS provides specificity, matrix effects and the lack of separation of isobaric and isomeric species can make chemical and biological interpretations challenging. Therefore, several advanced strategies have been implemented to increase the specificity of imaging techniques including the addition of ion mobility spectrometry,14,15 tandem mass spectrometry,16 specific reactive chemistries,17,18 and metal cationization.19 To further increase the specificity for direct tissue measurements, reduce sample-handling, and minimize matrix effects, Kertzes and Van Berkel introduced strategies for extraction of molecules into a liquid microjunction for subsequent reversed phase LC−MS.20−22 Two variants of a liquid microjunction were used for direct tissue sampling, the first based on a static droplet in contact with the tissue21,23 and the second using a continuously flowing solvent20 for subsequent loading into an injection loop and LC-MS analysis. Direct tissue sampling in combination with LC separation enabled unique chemical information to be obtained from a sampling diameter of ∼ 1000 μm. For example, propranolol was detected in all sampled organs from a whole body mouse section of a dosed animal, while two isomeric propranolol metabolites were found to specifically localize to the lung, kidney, and liver.21 However, these studies relied on diffusion of molecules into the extraction solution for sampling. Capillary electrophoresis on the other hand enables analytes to be actively loaded through electrokinetic injection. In addition, CE enables simultaneous analysis of a wide range of endogenous molecules from biological samples. Therefore, this study aims to pair CEMS with direct surface sampling to demonstrate the first example of untargeted metabolomic analysis directly from distinct cellular regions in tissue. Herein, we report the design and function of surface sampling capillary electrophoresis mass spectrometry (SS-CEMS). This unique tool presents a direct way to obtain the chemical profile of specific regions in morphologically complex tissue without any sample preparation or tissue dissection. We show that SS-CE-MS enables untargeted and concurrent screening of endogenous molecular species ranging from small
■
EXPERIMENTAL SECTION Construction and Operation of SS-CE-MS. A schematic representation of the SS-CE-MS device is shown in Figure 1. The device contains a custom 3D printed piece to hold a solvent delivery stainless steel capillary (360 μm OD, 100 μm ID, Kinvall AB, Sparreholm, Sweden) and a fused silica separation capillary with a TaperTip emitter (50 cm, 360 μm, OD 50 μm ID, New Objective, Massachusetts, USA) firmly together at an 85° angle (Figure S1A). This capillary assembly allows for a liquid microdroplet to be suspended between the capillaries for chemical sampling directly from the tissue surface. The CE was interfaced to the MS by coating the emitter of the separation capillary with polyimide graphite slurry (20% w/w)24 and feeding it through an Upchurch PEEK tee (0.04” through hole) connected to a custom metal sleeve. The CE-MS interface was modeled after the design by Guo and co-workers25 and positioned ∼5 mm from the MS inlet. Electrical contact between the metal sleeve and the emitter tip was facilitated by a conductive liquid (10% acetic acid, SigmaAldrich, Darmstadt, Germany) delivered through the top of the tee (Figure S1B,C). The CE circuit for electrokinetic injection and separation consisted of a high voltage power supply (Spellman SL10, New York, USA) connected to a metal union in contact with the sampling solvent delivery stainless steel capillary and to the conductive liquid in contact with the emitter tip for the high voltage load return. Background electrolyte (BGE) contained 0.1 M acetic acid and 50 mM benzoquinone in 80% deionized water (18 MΩ, Milli-Q, Millipore), 20% LC-MS grade methanol (SigmaAldrich). Before instrument operation and between each electrophoretic run, the CE capillary was back flushed by 7820
DOI: 10.1021/acs.analchem.9b01516 Anal. Chem. 2019, 91, 7819−7827
Article
Analytical Chemistry coupling the ESI emitter tip to a 500 μL gastight syringe with an Upcurch PEEK union. A syringe pump set to 15 μL min−1 (Legato 180, KD Scientific) was used to deliver >100 capillary volumes of BGE, >50 capillary volumes of 1 M sodium hydroxide, >50 capillary volumes of deionized water, and finally another 100 capillary volumes of BGE, to remove possible contaminations and regenerate the silica surface. To sample from the tissue surface, a droplet maintained between the capillaries was lowered onto a targeted location on the tissue section using an XYZ micromanipulator (Quarter Research, Oregon, USA). Following, endogenous material was loaded into the capillary electrokinetically by applying 30 kV for 10−30 s. After loading, the capillary assembly was moved to a vial of BGE using the micromanipulator, and electrophoretic separation was enabled using a 30 kV potential. A supplementary voltage (3−4.5 kV) was added to the emitter tip to maintain a stable electrospray during experiments. All presented data, including data for standard solutions and tissue surface sampling were generated with the SS-CE-MS device. The size of the extraction droplet during SS-CE-MS measurements was ∼360 μm, as approximated by the outer diameter of the CE separation capillary. After evaporation of the remaining sampling BGE microdroplet, the diameter of the observed BGE footprint was ∼400−600 μm, indicating that the total tissue area in contact with BGE had a diameter between 360 and 600 μm. Mass Spectrometry and Data Analysis. All data was acquired with a QExactive Orbitrap mass spectrometer (Thermo Fisher Scientific, Bremen, Germany) at a resolution of 140 000 (m/Δm at m/z 200) and a heated capillary temperature of 300 °C. The AGC target was set to 1 000 000 and mass spectra were recorded between m/z 100−1000. For investigation of in-source fragmentation of creatine with conventional electrospray, a HESI 2 ESI source was used with a sample flow rate of 10 μL min−1, capillary voltage of 3.5 kV, and an auxiliary gas flow of 10 arbitrary units. The experimental setups for nanospray desorption electrospray ionization (nano-DESI) and pneumatically assisted nano-DESI experiments mirrored a previous study.26 The acquired Thermo RAW files were converted to centroided mzXML files, and custom Matlab routines were used to generate electropherograms and mass spectra for Figures. Extracted ion electropherograms were generated by selecting the closest peak to the target accurate mass within a 5 ppm tolerance. Relative migration times were calculated from the slope and intercept of a regression line constructed using the smallest and largest migration times for the presented analytes. Electropherogram peak areas were calculated in ThermoFisher Xcalibur QualBrowser. For untargeted profiling of tissue samples Thermo RAW files were uploaded into MZmine27 for CAMERA28 peak selection. The parameters for MZmine electrophoretic peak detection were a minimum span of 0.1 min, a minimum peak height of 5 × 105 counts, and a 0.002 Da or 5 ppm m/z tolerance (whichever was smaller). Following electropherogram building, the peaks were deisotoped with a m/z tolerance of 0.002 Da or 5 ppm. For CAMERA data filtering, the settings were adjusted to a 0.2 fwhm sigma, 5% fwhm percentage, maximum charge of +1, m/z tolerance of 0.002 or 5 ppm, correlation threshold of 0.9, and correlation pvalue of 0.05. The isotope search was performed before shape correlation. Annotations of creatine, creatinine, γ-aminobutyric acid, dopamine, serotonin, betaine, and valine were verified with standards, using relative migration times and accurate
masses. All other analytes were putatively assigned based on accurate mass identification with METLIN (metlin.scripps. edu). Reagents and Standards. Creatine, creatinine, γ-aminobutyric acid, dopamine, serotonin, betaine, and L-valine were purchased from Sigma-Aldrich at least 99% pure. All stock solutions were prepared in 50/50 deionized water/methanol solvent. Working standard solutions were prepared in BGE. Brain homogenate for determining electrokinetic injection reproducibility was prepared by dissolving ∼1 cm3 of a male Sprague−Dawley brain into 50 mL methanol. The homogenate was sonicated for 3 min, and the supernatant was collected after centrifugation. Tissue Samples. Rat spinal cord tissue samples were prepared as previously described.29 Briefly, the spinal cord from a male Sprague−Dawley rat was flushed with saline solution and rapidly frozen on dry ice for subsequent sectioning into 12 μm thick sections on a cryo-microtome (Leica CM3000, Leica Biosystems, Nussloch, Germany). Sections were thaw mounted onto regular glass slides for SSCE-MS analysis. Tissue sections of kidney and brain were obtained from male Sprague−Dawley rats using the same procedure. All animal experiments were approved by the Uppsala animal ethics committee and were performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. For determining the reproducibility of SS-CE-MS extractions from a surface, the outer epidermal layer was peeled from red Allium Cepa and placed on a clean glass slide.
■
RESULTS AND DISCUSSION The SS-CE-MS device was designed to enable direct and untargeted molecular profiling from a defined location of a biological sample surface, such as a tissue section, using CEMS. The motivation for the device included: (i) minimizing sample preparation to ensure tissue and molecular integrity; (ii) comparing molecular profiles of distinct morphological regions in tissue; (iii) increasing the sensitivity over direct infusion by simplifying the chemical matrix during ionization; and (iv) separating isomers and isobars for increased molecular coverage and biological understanding. While these characteristics are highly advantageous, several obstacles had to be overcome to design a successful device. These include (i) design of a 3D printed holder for keeping the capillary assembly firmly together for simple and reproducible positioning and sampling; (ii) use of a stainless steel solvent delivery capillary to transfer the high voltage to the separation capillary for electrokinetic injection and CE separation; (iii) design of a robust CE-MS interface including an electrospray emitter coated with graphite for mechanical durability and addition of a supplemental voltage for stable electrospray; (iv) addition of benzoquinone to the BGE to reduce air bubble formation during CE separation; and (v) incorporation of a micromanipulator for targeted sampling of specific morphological tissue regions. For active sampling, the molecules from the surface were electrokinetically loaded into the CE capillary for subsequent capillary zone electrophoresis and ESI-MS. A schematic of the SS-CE-MS design and steps for sampling and separation are shown in Figure 1, and photographs are displayed in Figure S1. Sampling Reproducibility. For the SS-CE-MS device to be useful in future biological studies it is important that the technical reproducibility in both relative migration time and 7821
DOI: 10.1021/acs.analchem.9b01516 Anal. Chem. 2019, 91, 7819−7827
Article
Analytical Chemistry
Figure 2. Relative migration times for metabolites are consistent over four locations of onion epidermal layer (A), and four different tissue types (B). For (B), tissue types include onion epidermal layer, rat kidney medulla, spinal cord white matter, and brain gray matter. Multiplication factors [displayed on top of (A and B)] were used to display the diverse set of metabolites on the same scale.
tissue samples were analyzed on different days with different separation capillaries and ESI emitters over the span of seven months. The data show that SS-CE-MS provides robust and reproducible relative migration times, facilitating molecular annotation and interrogation of chemically complex biological tissues. Analyte Detection. SS-CE-MS enables separation and detection of numerous molecules sampled directly from the tissue surface. The number of molecular features detected from a single spot on rat brain cortex was investigated using MZmine.27 This resulted in 13 222 distinct deisotoped molecular features detected with migration times between 3 and 20 min. Further data filtering with CAMERA28 and manual inspection reduced this number to a total of 476 detected unique molecules. Interested readers are directed to data files illustrating the immense and molecularly diverse data from this example SS-CE-MS run (for further information see supplementary data files). The majority of the 476 detected molecules were putatively assigned as small structurally diverse polar metabolite species by searching METLIN (metlin.scrips. edu). The respective metabolite groups are highly diverse and include carnitines, proteogenic and nonproteogenic amino acids, small peptides, vitamins, neurotransmitters, sugars, and nucleobases (including nucleotides, nucleosides, and dinucleotides). An example of 17 diverse endogenous metabolites from several molecular classes are displayed in Figure 3. The figure includes amino acids from five categories (including hydrophobic aliphatic, hydrophobic aromatic, positively charged, and acidic amino acids). In addition to those presented in Figure 3, 17 of the 20 amino acids involved in protein synthesis and many biological pathways are shown in Figure S2. Additional metabolites displayed in Figure 3 are carnitines, necessary for oxidative catabolism of fatty acids, and hypoxanthine, a spontaneous deamination product of adenine that resembles guanine. Further, we also detect the dinucleotide NAD+, which
peak area is high. The technical reproducibility of the electrokinetic injection with SS-CE-MS was evaluated by performing replicate injections (n = 4) of a rat brain homogenate diluted in BGE. The mean peak area relative standard deviation (RSD) for four common metabolites was 15.3% (Table S1), which is consistent with literature values for high voltage electrokinetic injection30 and comparable to results achieved with the previously characterized sheathless CE-MS interface from which SS-CE-MS was modeled.25 To further determine the variance of the entire sampling and detection procedure for SS-CE-MS from a surface, the outer epidermal layer of A. Cepa was used. The epidermal cells of A. Cepa are homogeneously distributed, signifying that the metabolic profiles are conserved throughout the tissue. Replicate sampling events from unique locations on the epidermal layer for ten representative metabolites provided a mean peak area RSD value of 20.7% (n = 5, Table S2). These RSD values are comparable to the RSD values obtained from liquid sampling; therefore, we conclude that the bulk of technical variability for SS-CE-MS measurements in peak areas originate from the electrokinetic injection, with only minor contribution from the surface sampling event. The reproducibility of relative migration times is essential for confidence in analyte annotation. Selected ion electropherograms for ten analytes in four SS-CE-MS measurements on A. Cepa epidermis are displayed in Figure 2A. These data show that the obtained relative migration times using SS-CEMS are consistent between sampling events. Moreover, the relative migration times obtained using SS-CE-MS are reproducible in several tissue types of varying complexity and over many months, despite using different CE separation capillaries. Specifically, selected analytes in Figure 2B show comparable relative migration times for ten endogenous metabolites sampled directly from A. Cepa epidermal layer, rat kidney, rat spinal cord, or rat brain tissue sections. These 7822
DOI: 10.1021/acs.analchem.9b01516 Anal. Chem. 2019, 91, 7819−7827
Article
Analytical Chemistry
diverse metabolite groups within a single sampling event of a defined location on chemically complex tissue. Prominent In-Source Fragmentations Are Identified. In-source fragmentations, where molecules spontaneously dissociate during ionization, have been identified as interferences in metabolomic studies.31 Two commonly studied and important metabolites involved in many metabolic processes, including the cellular energy cycle, are creatine and creatinine.32 Both are readily separated and detected from tissue sections using SS-CE-MS. When analyzing a combined standard of creatinine and creatine by SS-CE-MS, creatinine is detected at ∼4 min (Figure 4C for protonated species), while
Figure 4. Extracted ion electropherograms displaying in-source fragmentation of creatine to creatinine. (A, B, C, and D) are obtained after sampling a mixture of creatine and creatinine standards (each 10 μM). (A and B) display the protonated and sodiated mass channel for creatine, and (C and D) show the protonated and sodiated mass channel for creatinine. (E and F) are obtained from the gray matter of rat spinal cord tissue and show the extracted ion electropherograms for protonated and sodiated endogenous creatinine, respectively.
Figure 3. Electropherograms for selected structurally diverse endogenous metabolites detected by SS-CE-MS from the cortex of rat brain tissue. The Met-His-Met-Pro tetrapeptide is putatively assigned by accurate mass and could contain multiple isomers.
is involved in oxidation and reduction processes in living cells. These data show that within one single analysis at a spatially defined location it is possible to sample, separate, and detect structurally diverse metabolite species in an effective separation window of 3 to 20 min with a calculated electroosmotic flow of 0.56 mm s−1 under the current BGE conditions. With the current BGE conditions and CE separation voltage the majority of metabolites were detected with migration times between 5 and 20 min, giving rise to a duty cycle of ∼30 min per analysis. Future generations of SS-CE-MS will include automation of the surface sampling and capillary rinsing procedures to reduce analysis time and enable mass spectrometry imaging applications. Overall, SS-CE-ME provides a plethora of detected metabolites from structurally
creatine migrates slower and is detected at ∼6 min (Figure 4A,B for protonated and sodiated species, respectively). However, the extracted ion electropherogram for sodiated creatinine (m/z 136.0487, Figure 4D) displays two peaks present at 4 and 6 min, suggesting that creatine is decomposing to creatinine via dehydration during ESI. This in-source fragmentation is mainly detected in the sodiated mass channel of creatinine and only to a limited extent in the protonated and potassiated mass channels (Figure S3). Therefore, we draw the conclusion that sodium adduct specific fragmentation of creatine occurs. This spontaneous and exclusive in-source fragmentation, indicating unique gas phase ion energetics for 7823
DOI: 10.1021/acs.analchem.9b01516 Anal. Chem. 2019, 91, 7819−7827
Article
Analytical Chemistry
Figure 5. SS-CE-MS analysis of spatially defined locations from rat kidney. (A and B) Extracted ion electropherograms of m/z 118.0861 show electrophoretically resolved detection of the isomeric species valine and betaine. (A) Electropherograms for individual standards of valine and betaine, respectively (5 μM). (B) Electropherograms of endogenous valine and betaine sampled from rat kidney medulla and cortex, respectively. The insets in (B) display rescaled electrophoretic peaks for valine. (C) Extracted ion electropherograms for adenine (m/z 136.0627), adenosine (m/z 268.1039), and adenosine monophosphate (m/z 348.0703) sampled from the kidney medulla and cortex with SS-CE-MS.
resolving power of 140 000 (m/Δm at m/z 200). Thus, without SS-CE-MS, isobaric interferences could strongly skew the interpreted abundance of targeted molecular species sampled from the tissue surface. Tandem MS has the potential to separate isobars and isomers; however, the presence of common product ions makes differentiation by tandem MS challenging. This of particular importance when isomers or isobars are present at varying physiological concentrations, since the intensity of the fragments will be a result of both species. Valine and betaine are two isomeric species that are difficult to confidently separate by tandem MS due the presence of many common product ions (Table S3). In addition, they have vastly different biological functions; valine is an essential amino acid required for protein synthesis, while betaine is an abundant osmoregulator.34 It is therefore of utmost importance for accurate biological interpretation to distinguish between these species. The extracted ion electropherograms for individual standard solutions of valine and betaine in Figure 5A demonstrate separation of valine and betaine detected at ∼7 min and ∼8 min, respectively. We further employed SS-CE-MS to investigate the abundance of endogenous valine and betaine in the medulla and cortex of a rat kidney tissue section (Figure 5B). The annotations of endogenous valine and betaine are based on migration time agreement with standards and accurate mass measurements (118.0861, mass error = −0.67 ppm). It is apparent from the large difference in endogenous peak areas that the relative abundance of betaine is much higher than valine (∼2% valine to betaine ratio) in both the kidney medulla and cortex. In addition, replicate analyses (n = 3) of the medulla and cortex show that the betaine signal is ∼5 times higher in the medulla than in the cortex (p-value = 0.002, Table S4), which is in accordance with the literature.35,36
each adduct species, is challenging to detect without separation prior to MS. To demonstrate in-source fragmentation for endogenous creatinine, SS-CE-MS was used to analyze the gray matter in rat spinal cord tissue. The dominant peak for protonated creatinine from spinal cord tissue was detected at ∼4 min (Figure 4E), in accordance with the creatinine standard. However, the dominant peak of sodiated creatinine is detected in accordance with the creatine standard migration time at ∼6 min (Figure 4F). Therefore, the majority of signal observed for sodiated creatinine from the tissue actually arises from endogenous creatine. This specific in-source fragmentation is not a consequence of CE-MS analysis. In fact, after identification with SS-CE-MS, the same phenomenon was detected using conventional ESI, nanospray desorption electrospray ionization (nano-DESI),33 and pneumatically assisted nano-DESI26 (Figure S4). All together, these SS-CEMS results demonstrate for the first time that it is essential to monitor protonated creatinine with ESI to ensure minimal interference from creatine due to in-source fragmentation. Isobaric and Isomeric Interferences Are Separated. Interferences in direct infusion MS can also arise from isobaric compounds that are not resolved, even with high mass resolving instruments. Fortunately, isobaric species can be separated with SS-CE-MS. For example, the protonated, sodiated, and potassiated adducts of GABA are all detected at ∼6 min (Figure S4). However, the prevailing peak in the potassiated GABA mass channel is observed at ∼13 min, indicating the presence of an isobaric species. The isobaric interference, detected at ∼13 min (m/z 142.0264 putatively assigned as phosphoethanolamine, metlin.scripps.edu), is indistinguishable from potassiated GABA (m/z 142.0265) by mass spectrometry alone, even at this relatively high mass 7824
DOI: 10.1021/acs.analchem.9b01516 Anal. Chem. 2019, 91, 7819−7827
Article
Analytical Chemistry
Figure 6. SS-CE-MS analysis of three regions from rat brain tissue, the cortex, hippocampus, and thalamus. Extracted ion electropherograms show endogenous neurotransmitters and polypeptide simultaneously sampled and detected in the same electrophoretic run. GABA, serotonin, and dopamine annotations were verified with standards, while acetylcholine and thymosin β-4 are putatively assigned based on accurate mass. The depicted spot size corresponds to the total tissue area in contact with BGE.
cortex is a result of the larger amount of in-source fragmented adenosine. Thus, the added separation of the adenine mass channel at m/z 136.0517 enables in-source fragmentation interferences to be eliminated, revealing twice the peak area for endogenous adenine in the kidney medulla (8.7 × 106) compared to the cortex (4.4 × 106). Overall, these results showcase the capacity for separation of isobars, isomers, and interferences in complex biological samples with SS-CE-MS within spatially defined tissue regions. Spatially Defined Profiling of Diverse Biomolecules. To further demonstrate spatially defined measurements, SSCE-MS was employed to investigate the molecular composition of different regions in brain tissue. Three regions were sampled from rat brain tissue section, including the cortex, hippocampus, and thalamus (Figure 6). These data reveal the detection of small endogenous neurotransmitters sampled directly from brain tissue sections. Previous work has shown that low physiological concentrations and poor ionization efficiencies of small neurotransmitters, such as dopamine and serotonin, restrict their detection by direct infusion MS without targeted derivatization strategies.37−41 However, endogenous acetylcholine, γ-aminobutyric acid (GABA), dopamine, and serotonin were readily detected from tissue sections with SS-CE-MS without any sample preparation or derivatization. Analyte annotations for GABA, serotonin, and dopamine are supported by migration times obtained from standard solutions (Figure S5). It was observed that acetylcholine, GABA, and serotonin are present in the cortex, hippocampus, and thalamus. Conversely, dopamine was detected only in the thalamus. Note that the single sampling event may not allow for comparison of absolute peak areas between locations. The observed neurotransmitter distributions are in agreement with previous work outlining the uniform distribution of acetylcholine and GABA42 and specific localization of dopamine in rat brain tissue sections.41 These
Finally, the results show that betaine is homogeneously distributed in both the cortex and the medulla, while valine is more heterogeneously distributed in these morphological regions, as illustrated by the higher RSD values for valine peak areas found in Table S4. These results signify the importance of spatially resolved detection and separation of isomeric species directly from tissue sections. The capacity for separation and detection of isomeric species localized in morphological regions makes SS-CE-MS a valuable tool for profiling chemically complex tissue. In addition to spatially defined measurements of valine and betaine, SS-CE-MS was employed to investigate the relative abundance of adenine in the kidney medulla and cortex. Adenine is a nucleobase incorporated into DNA and a precursor for the nucleosides adenosine and adenosine monophosphate, which are essential in many cellular metabolic functions. To mimic direct infusion MS without separation, we calculated the total peak area detected at the accurate mass for adenine (m/z 136.0517) to be 1.7 × 108 for the kidney medulla and 2.2 × 108 for the kidney cortex. As such, data obtained with direct infusion MS would suggest that adenine is more abundant in the kidney cortex. However, this is not a true representation of adenine distribution within the kidney. As evidenced by SS-CE-MS separation, there are three peaks detected at m/z 136.0517. These correspond to adenine detected at ∼6 min, adenosine in-source fragmentation detected at ∼7.5 min, and adenosine monophosphate insource fragmentation detected at ∼11 min (Figure 5C). Note that the migration times for the intact molecular ions for adenosine (m/z 268.1039) and adenosine monophosphate (m/z 348.0703) match the in-source fragments. The data show consistent peak areas for intact adenosine monophosphate in the cortex and medulla, while the peak area for adenosine is twice as high in the cortex compared to the medulla. Therefore, the overall higher abundance of adenine in the 7825
DOI: 10.1021/acs.analchem.9b01516 Anal. Chem. 2019, 91, 7819−7827
Article
Analytical Chemistry results demonstrate the high-sensitivity of SS-CE-MS, making it a valuable tool for targeted mapping of neurotransmitter localizations within tissue sections. In addition to low molecular weight compounds, the electropherogram from SS-CE-MS analysis of rat brain tissue contains multiply charged molecular species. For example, an ion at the deconvoluted protonated m/z 4961.4874 was detected in three charge states (Figure S6) in the cortex and hippocampus (Figure 6) and tentatively assigned as Thymosin β-4 (metlin.scripps.edu). This assignment was supported by previous studies identifying Thymosin β-4 in rat brain tissue.43,44 In addition to Thymosin β-4, several other unidentified multiply charged species were detected by SSCE-MS (for further information see the Supporting Information data files). Detection of proteins and peptides directly off the tissue has been previously reported with22 and without separation;15,45−50 however, this is the first report where the tissue was not washed with solvents (e.g., ethanol) prior to analysis to remove abundant lipids and metabolites. While washing enables analysis of large biomolecules, it restricts simultaneous detection of small biomolecules. In addition, it may cause protein modifications and alter molecular localization.51 By concentrating the extracted molecules into discrete migration bands, SS-CE-MS enables simultaneous detection of both polypeptides and small molecules from the same tissue location without any sample treatment. Overall, SS-CE-MS provides direct and spatially defined measurements for a diverse set of molecular species, ranging from small molecules to polypeptides. This proof-of-principle study demonstrates feasibility and promising results for the direct surface sampling of diverse biomolecules from thin tissue sections. However, several challenges remain to be solved before SS-CE-MS can be routinely applied in biological studies. These challenges include: (i) reducing the effective sampling area to facilitate studies of smaller morphological regions; (ii) automating sampling and separation to improve throughput and enable systematic sampling; (iii) redesigning the capillary assembly for introduction of internal standards to allow for relative quantification of extracted metabolites; and (iv) investigation of negative mode separation for acidic anionic metabolites such as sugars. Future studies may also include analysis of single cells, where preliminary results show feasibility for metabolite profiling at the single cell level.
confident molecular annotation and regiospecific molecular profiling of biological systems.
■
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.analchem.9b01516.
■
Photographs of SS-CE-MS capillary device and capillary assembly, electropherograms from amino acids sampled from brain tissue, in source fragmentation of creatine and creatinine, isobaric species detected for GABA with SS-CE-MS, detection of neurotransmitters from rat brain tissue, endogenous polypeptide detected from rat brain tissue by SS-CE-MS, replicate peak areas for injections of rat brain homogenate solution, replicate peak areas for injections from Allium Cepa epithelium layer, product ions for MS/MS analysis of valine and betain, valine and betain peak areas sample from kidney medulla and cortex (PDF) Supporting data files: An example RAW and MZmine file obtained from SS-CE-MS analysis of rat brain tissue and CSV peaks lists for the deisotoped molecular features and pseudo spectra peaks from CAMERA screening are published in a data redepository (doi:10.17632/7kkp29fyr5.2)
AUTHOR INFORMATION
Corresponding Author
*E-mail: Ingela.Lanekoff@kemi.uu.se. ORCID
Kyle D. Duncan: 0000-0003-0575-0858 Ingela Lanekoff: 0000-0001-9040-3230 Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS Funding for this work was provided by the Swedish Foundation for Strategic Research (SSF ICA-6) and the Swedish Research Council (VR 621-2013-4231). The authors thank Dr. Ping Sui for the rat spinal cord sections and Prof. Fredrik Palm for the rat kidney and brain tissue used for this work.
■
■
CONCLUSION We report the first CE-MS device that enables sampling directly from the surface of biological tissue sections without sample preparation of sectioned tissue. SS-CE-MS provides spatially defined untargeted chemical profiling and high sensitivity measurements by concentrating analytes into discrete migration bands. This is exemplified by the direct detection of metabolites and polypeptides from tissue without washing procedures, enabling simultaneous characterization of small and large biomolecules from a single tissue location. Additionally, SS-CE-MS provides reproducible relative migration times over vastly different tissue complexities, allows for differentiation of isobaric and isomeric species, and identifies in-source fragmentation interferences, thereby increasing the confidence in analyte annotations for enhanced biological interpretations for defined morphological regions in tissue sections. We foresee SS-CE-MS as an invaluable tool for
REFERENCES
(1) García, A.; Godzien, J.; López-Gonzálvez, Á .; Barbas, C. Bioanalysis 2017, 9 (1), 99−130. (2) Britz-McKibbin, P.; Keenan, K.; Brick, L.; Hill, S.; Pedder, L.; Macedo, A. N.; Gonska, T.; Mathiaparanam, S. ACS Cent. Sci. 2017, 3 (8), 904−913. (3) Saoi, M.; Percival, M.; Nemr, C.; Li, A.; Gibala, M.; BritzMcKibbin, P. Anal. Chem. 2019, 91 (7), 4709−4718. (4) Onjiko, R. M.; Moody, S. A.; Nemes, P. Proc. Natl. Acad. Sci. U. S. A. 2015, 112 (21), 6545−6550. (5) Barbas, C.; Moraes, E. P.; Villaseñor, A. J. Pharm. Biomed. Anal. 2011, 55 (4), 823−831. (6) Hirayama, A.; Wakayama, M.; Soga, T. TrAC, Trends Anal. Chem. 2014, 61, 215−222. (7) Kuehnbaum, N. L.; Kormendi, A.; Britz-McKibbin, P. Anal. Chem. 2013, 85 (22), 10664−10669. (8) Zhang, W.; Hankemeier, T.; Ramautar, R. Curr. Opin. Biotechnol. 2017, 43, 1−7. 7826
DOI: 10.1021/acs.analchem.9b01516 Anal. Chem. 2019, 91, 7819−7827
Article
Analytical Chemistry (9) Azab, S.; Ly, R.; Britz-McKibbin, P. Anal. Chem. 2019, 91 (3), 2329−2336. (10) Dilillo, M.; Ait-Belkacem, R.; Esteve, C.; Pellegrini, D.; Nicolardi, S.; Costa, M.; Vannini, E.; de Graaf, E. L.; Caleo, M.; McDonnell, L. A. Sci. Rep. 2017, 7 (1), 603. (11) Dilillo, M.; Pellegrini, D.; Ait-Belkacem, R.; de Graaf, E. L.; Caleo, M.; McDonnell, L. A. J. Proteome Res. 2017, 16 (8), 2993− 3001. (12) Zimmerman, M.; Lestner, J.; Prideaux, B.; O'Brien, P.; DiasFreedman, I.; Chen, C.; Dietzold, J.; Daudelin, I.; Kaya, F.; Blanc, L.; Chen, P.-Y.; Park, S.; Salgame, P.; Sarathy, J.; Dartois, V. Antimicrob. Agents Chemother. 2017, 61 (9), e00924−17. (13) Quanico, J.; Franck, J.; Wisztorski, M.; Salzet, M.; Fournier, I. Biochim. Biophys. Acta, Gen. Subj. 2017, 1861 (7), 1702−1714. (14) Trim, P. J.; Henson, C. M.; Avery, J. L.; McEwen, A.; Snel, M. F.; Claude, E.; Marshall, P. S.; West, A.; Princivalle, A. P.; Clench, M. R. Anal. Chem. 2008, 80 (22), 8628−8634. (15) Stauber, J.; MacAleese, L.; Franck, J.; Claude, E.; Snel, M.; Kaletas, B. K.; Wiel, I. M. V. D.; Wisztorski, M.; Fournier, I.; Heeren, R. M. A. J. Am. Soc. Mass Spectrom. 2010, 21 (3), 338−347. (16) Lanekoff, I.; Burnum-Johnson, K.; Thomas, M.; Short, J.; Carson, J. P.; Cha, J.; Dey, S. K.; Yang, P.; Prieto Conaway, M. C.; Laskin, J. Anal. Chem. 2013, 85 (20), 9596−9603. (17) Rao, W.; Pan, N.; Tian, X.; Yang, Z. J. Am. Soc. Mass Spectrom. 2016, 27 (1), 124−134. (18) Tang, F.; Guo, C.; Ma, X.; Zhang, J.; Su, Y.; Tian, R.; Shi, R.; Xia, Y.; Wang, X.; Ouyang, Z. Anal. Chem. 2018, 90 (9), 5612−5619. (19) Duncan, K. D.; Fang, R.; Yuan, J.; Chu, R. K.; Dey, S. K.; Burnum-Johnson, K. E.; Lanekoff, I. Anal. Chem. 2018, 90 (12), 7246−7252. (20) Van Berkel, G. J.; Kertesz, V. Rapid Commun. Mass Spectrom. 2013, 27 (12), 1329−1334. (21) Kertesz, V.; Van Berkel, G. Anal. Chem. 2010, 82 (14), 5917− 5921. (22) Kertesz, V.; Calligaris, D.; Feldman, D. R.; Changelian, A.; Laws, E. R.; Santagata, S.; Agar, N. Y. R.; Van Berkel, G. J. Anal. Bioanal. Chem. 2015, 407 (20), 5989−5998. (23) Kertesz, V.; Weiskittel, T. M.; Vavrek, M.; Freddo, C.; Van Berkel, G. J. Rapid Commun. Mass Spectrom. 2016, 30 (14), 1705− 1712. (24) Nilsson, S.; Wetterhall, M.; Bergquist, J.; Nyholm, L.; Markides, K. E. Rapid Commun. Mass Spectrom. 2001, 15 (21), 1997−2000. (25) Guo, X.; Fillmore, T. L.; Gao, Y.; Tang, K. Anal. Chem. 2016, 88 (8), 4418−4425. (26) Duncan, K. D.; Bergman, H. M.; Lanekoff, I. Analyst 2017, 142 (18), 3424−3431. (27) Pluskal, T.; Castillo, S.; Villar-Briones, A.; Orešič, M. BMC Bioinf. 2010, 11, 1 DOI: 10.1186/1471-2105-11-395. (28) Böttcher, C.; Kuhl, C.; Neumann, S.; Tautenhahn, R.; Larson, T. R. Anal. Chem. 2012, 84 (1), 283−289. (29) Sui, P.; Watanabe, H.; Artemenko, K.; Sun, W.; Bakalkin, G.; Andersson, M.; Bergquist, J. Eur. J. Mass Spectrom. 2017, 23 (3), 105− 115. (30) Schaeper, J. P.; Sepaniak, M. J. Electrophoresis 2000, 21 (7), 1421−1429. (31) Domingo-Almenara, X.; Montenegro-Burke, J. R.; Guijas, C.; Majumder, E. L.-W.; Benton, H. P.; Siuzdak, G. Anal. Chem. 2019, 91 (5), 3246−3253. (32) Wyss, M.; Kaddurah-Daouk, R. Physiol. Rev. 2000, 80 (3), 1107−1213. (33) Roach, P. J.; Laskin, J.; Laskin, A. Analyst 2010, 135 (9), 2233− 2236. (34) Burg, M. B.; Ferraris, J. D. J. Biol. Chem. 2008, 283 (12), 7309− 7313. (35) Bagnasco, S.; Balaban, R.; Fales, H. M.; Yang, Y. M.; Burg, M. J. Biol. Chem. 1986, 261 (13), 5872−5877. (36) Moeckel, G. W.; Lien, Y. H. Am. J. Physiol. Physiol. 1997, 272 (1), F94−F99.
(37) Wu, C.; Ifa, D. R.; Manicke, N. E.; Cooks, R. G. Analyst 2010, 135 (1), 28−32. (38) Fernandes, A. M. A. P.; Vendramini, P. H.; Galaverna, R.; Schwab, N. V.; Alberici, L. C.; Augusti, R.; Castilho, R. F.; Eberlin, M. N. J. Am. Soc. Mass Spectrom. 2016, 27 (12), 1944−1951. (39) Manier, M. L.; Spraggins, J. M.; Reyzer, M. L.; Norris, J. L.; Caprioli, R. M. J. Mass Spectrom. 2014, 49 (8), 665−673. (40) Shariatgorji, M.; Strittmatter, N.; Nilsson, A.; Källback, P.; Alvarsson, A.; Zhang, X.; Vallianatou, T.; Svenningsson, P.; Goodwin, R. J. A.; Andren, P. E. NeuroImage 2016, 136, 129−138. (41) Shariatgorji, M.; Nilsson, A.; Goodwin, R. J. A. J. A.; Källback, P.; Schintu, N.; Zhang, X.; Crossman, A. R. R.; Bezard, E.; Svenningsson, P.; Andren, P. E. E. Neuron 2014, 84 (4), 697−707. (42) Bergman, H.-M.; Lundin, E.; Andersson, M.; Lanekoff, I. Analyst 2016, 141 (12), 3686−3695. (43) Hannappel, E.; Xu, G. J.; Morgan, J.; Hempstead, J.; Horecker, B. L. Proc. Natl. Acad. Sci. U. S. A. 1982, 79 (7), 2172−2175. (44) Vartiainen, N.; Pyykönen, I.; Hökfelt, T.; Koistinaho, J. NeuroReport 1996, 7 (10), 1613−1616. (45) Feider, C. L.; Elizondo, N.; Eberlin, L. S. Anal. Chem. 2016, 88 (23), 11533−11541. (46) Sarsby, J.; Martin, N. J.; Lalor, P. F.; Bunch, J.; Cooper, H. J. J. Am. Soc. Mass Spectrom. 2014, 25 (11), 1953−1961. (47) Wisztorski, M.; Desmons, A.; Quanico, J.; Fatou, B.; Gimeno, J.-P.; Franck, J.; Salzet, M.; Fournier, I. Proteomics 2016, 16 (11−12), 1622−1632. (48) Griffiths, R. L.; Creese, A. J.; Race, A. M.; Bunch, J.; Cooper, H. Anal. Chem. 2016, 88 (13), 6758−6766. (49) Hsu, C.-C.; Chou, P.-T.; Zare, R. N. Anal. Chem. 2015, 87 (22), 11171−11175. (50) Garza, K. Y.; Feider, C. L.; Klein, D. R.; Rosenberg, J. A.; Brodbelt, J. S.; Eberlin, L. S. Anal. Chem. 2018, 90 (13), 7785−7789. (51) Amstalden van Hove, E. R.; Smith, D. F.; Fornai, L.; Glunde, K.; Heeren, R. M. A. J. Am. Soc. Mass Spectrom. 2011, 22 (10), 1885.
7827
DOI: 10.1021/acs.analchem.9b01516 Anal. Chem. 2019, 91, 7819−7827