of 10 ACS Paragon Plus Environment Analytical Chemistry 1 2 3 4 5 6

Subscriber access provided by UNIVERSITY OF THE SUNSHINE COAST

Article

Biocompatible Surface-Coated Probe for In vivo, In situ, and Microscale Lipidomics of Small Biological Organisms and Cells Using Mass Spectrometry Jiewei Deng, Wenying Li, Qiuxia Yang, Yaohui Liu, Ling Fang, Yunhua Guo, Peng-Ran Guo, Li Lin, Yunyun Yang, and Tiangang Luan Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.8b01218 • Publication Date (Web): 30 Apr 2018 Downloaded from http://pubs.acs.org on May 3, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 10 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

TOC 84x29mm (300 x 300 DPI)

ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 10

Biocompatible Surface-Coated Probe for In vivo, In situ, and Microscale Lipidomics of Small Biological Organisms and Cells Using Mass Spectrometry Jiewei Deng†, Wenying Li†, Qiuxia Yang‡, Yaohui Liu‡, Ling Fang†,§, Yunhua Guo§, Pengran Guo‡, Li Lin†, Yunyun Yang*,‡, and Tiangang Luan*,† †

State Key Laboratory of Biocontrol, South China Sea Bio-Resource Exploitation and Utilization Collaborative Innovation Center, School of Life Sciences, Sun Yat-Sen University, 135 Xingangxi Road, Guangzhou 510275, China ‡ Guangdong Engineering and Technology Research Center for Ambient Mass Spectrometry, Guangdong Provincial Key Laboratory of Emergency Test for Dangerous Chemicals, Guangdong Institute of Analysis (China National Analytical Center Guangzhou), 100 Xianlie Middle Road, Guangzhou 510070, China § Instrumental Analysis & Research Center, Sun Yat-Sen University, 135 Xingangxi Road, Guangzhou 510275, China ABSTRACT: Lipidomics is a significant way to understand the structural and functional roles that lipids play in biological systems. Although many mass spectrometry (MS)-based lipidomics strategies have recently achieve remarkable results, in vivo, in situ, and microscale lipidomics for small biological organisms and cells have not yet been obtained. In this article, we report a novel lipidomics methodology for in vivo, in situ, and microscale investigation of small biological organisms and cells using biocompatible surface-coated probe nanoelectrospray ionization mass spectrometry (BSCP-nanoESI-MS). A novel biocompatible surface-coated solid-phase microextration (SPME) probe is prepared, which possesses a probe-end diameter of less than 5 µm and shows excellent enrichment capacity towards lipid species. In vivo extraction of living biological organisms (e.g., zebrafishes), in situ sampling a precise position of small organisms (e.g. Daphnia magna), and even microscale analysis of single eukaryotic cells (e.g. HepG2) are easily achieved by the SPME probe. After extraction, the loaded SPME probe is directly applied for nanoESI-MS analysis, and a high resolution mass spectrometer is employed for recording spectra and identifying lipid species. Compared with the conventional direct infusion shotgun MS lipidomics, our proposed methodology shows similar result of lipid profiles, but with simpler sample pretreatment, less sample consumption, and shorter analytical times. Lipidomics of zebrafish, Daphnia magna, and HepG2 cell populations were investigated by our proposed BSCP-nanoESI-MS methodology, abundant lipid compositions were detected and identified, and biomarkers were obtained via multivariate statistical analysis.

Lipids play crucial roles in biological systems by serving as building blocks of cell membranes, sources for energy storage, and media for signal transduction.1 Because of the biodiversity and individual difference, it is necessary to explore and investigate the heterogeneity of lipids in different biological organisms and even cells. Lipidomics, as a key approach to understand the structural and functional roles that lipids play in biological systems, is thus of great importance. Many analytical techniques including nuclear magnetic resonance (NMR) spectroscopy,2 chromatography,3 electrophoresis,4 and mass spectrometry (MS),5-8 etc., have been developed to characterize lipids in various biological systems and cells. The vast advantages of MS such as high sensitivity, excellent selectivity, desirable specificity, and information-rich, etc., have enabled it a powerful tool for lipidomics investigation of various complex biological systems. Electrospray ionization (ESI)5,6 and matrix-assisted laser desorption/ionization (MALDI)7,8 are the two conventional ionization methods for MS analysis of lipids, and the most commonly used MS-based lipidomics strategies include direct infusion shotgun MS lipidomics, liquid chromatography (LC)-MS platforms, and MALDI-MS approaches.9 Direct infusion shotgun MS lipidomics applies a syringe pump to infuse crude extracts of

biological samples into a mass spectrometer for analysis, thus it may suffer from matrix effect, and signals from lowconcentration lipids might be restrained obviously. Both direct infusion shotgun MS and LC-MS are only allowed to analyse the extracts of biological samples with a relatively large amount of sample consumption, and thus they are difficult to achieve in vivo and in situ lipidomics investigation. In addition, LC-MS method includes time-consuming steps of sample pretreatment and chromatographic separation. Although MALDI-MS can achieve in situ and microscale analysis, it is operated under vacuum condition, restricting its analysis of living organisms and cells. Ambient MS10-12 has been developed and implemented into analytical practices since the early of 21st century, and it gives us the opportunity for in vivo, in situ, and microscale analysis of complex biological samples under ambient and open-air conditions. Till now, a series of ambient MS methods have been explored for lipids analysis of various biological organisms and cells. For instances, Cooks’ group reported the application of desorption electrospray ionization (DESI)-MS to obtain lipid profile for discrimination of human astrocytoma subtypes,13 and 3-dimensional imaging of lipids for mouse brain was also accomplished.14 Hiraoka’s group reported the 1


Page 3 of 10 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry organisms and cells. Subsequently, the lipids enriched on the SPME probe were analyzed directly by nanoESI-MS under ambient and open-air conditions, with the application of a high-resolution mass spectrometer to record spectra and identify lipid species. After that, lipidomics pattern recognition and biomarkers discovery were performed by multivariate statistical analysis. Compared with the existing lipidomics methods, our proposed methodology shows distinctive potentials: (i) in vivo analysis: suitable for in vivo sampling and analysis of living small biological organisms (e.g. zebrafish), and the organisms are still alive after sampling and analysis; (ii) in situ and microscale analysis: the size of SPME probe is small for in situ microsampling a precise position of small organisms with size at dozens-to-hundreds of micrometer; (iii) sensitivity: high enrichment capacity towards lipid species with improved sensitivity of 100-300 folds; (iv) biocompatibility: suitable for direct sampling and analysis of complex biological samples with good repeatability and reproducibility as well as long durability.

application of probe electrospray ionization (PESI)-MS for analysis of lipids from renal cell carcinoma.15 The application of easy ambient sonic-spray ionization (EASI)-MS for investigation of the changes of acidic lipids in living cyanobacteria during their different growth phases has been demonstrated by Liu’s group.16 Subcellular metabolite and lipid analysis of Xenopus laevis eggs was demonstrated by Vertes’s group using laser ablation electrospray ionization (LAESI)-MS.17 Differences in the lipid profiles between healthy and apoptotic HEK cells were observed using laser desorption/ionization droplet delivery (LDIDD)-MS.18 Direct extraction and analysis of triacylglycerols in human adipocytes was reported by Phelps et al. using nanomanipulation-coupled nanospray MS.19 Direct lipidometabolomics was accomplished by Masujima's group using live single-cell MS for analysis of a floating white blood cell within a nanospray tip after addition of ionization solvent followed by super-sonication.20 Although the above ambient MS studies offer new features for rapid and direct characterization of lipids for various biological organisms, they are challenge to establish a systematic and comprehensive lipidomics methodology. For instances, DESI-MS, EASI-MS, LAESI-MS, and LDIDD-MS are surface analysis methods, thus they are difficult to perform in-depth lipidomics investigation. PESI-MS utilizes a fine tungsten probe without adsorbents to sample lipids for analysis, and its sensitivity and selectivity is limited, only a few lipid species is detected. In addition, most of the existing studies are focused on the analysis of non-target metabolites without the extraction and enrichment of lipids, and thus their sensitivity are unsatisfied. Therefore, development of a novel methodology for in vivo, in situ, and microscale lipidomics investigation of small biological organisms and cells with high sensitivity is especially demanded. Coupling solid-phase microextraction (SPME) with ambient MS has emerged as a strategy for highly sensitive analysis of trace analytes in complex biological samples.21,22 Recently, SPME based ambient ionization methods such as surfacecoated wooden-tip ESI,23 coated blade spray ionization,24 and surface-coated probe nanoelectrospray ionization (SCPnanoESI)25 have been developed for rapid analysis of trace compounds in complicated biological samples with high sensitivity and low matrix effect. Among these methods, SCPnanoESI-MS is a microscale analysis method developed by our group recently, in which a surface-coated tungsten SPME probe with tip-end diameters of several-µm was designed for extraction and enrichment of trace compounds from small biological organisms.25 Undoubtedly, SCP-nanoESI-MS is a desirable approach for in vivo, in situ, and microscale analysis of complex biological organisms. However, the previously designed surface-coated SPME probe is not biocompatible, which restricts its application for analysis of biological samples with good reproducibility and long durability. In this article, we furthered our study and developed a novel biocompatible surface-coated probe nanoelectrospray ionization mass spectrometry (BSCP-nanoESI-MS) method for in vivo, in situ, and microscale lipid analysis and lipidomics investigation of small biological organisms and eukaryotic cells. The schematic diagram of BSCP-nanoESIMS lipidomics methodology was shown in Figure 1. First, a biocompatible surface-coated SPME probe toward lipids was designed. Then, the obtained SPME probe was applied for extraction and enrichment of lipids from small biological

EXPERIMENTAL SECTION

Preparation of Biocompatible Surface-Coated SPME Probe. The biocompatible surface-coated SPME probe was prepared with RS-6065 tungsten microdissecting probe (tipend diameter of ~1 µm, Roboz Surgical Instrument Co. Inc., Gaithersburg, MD, USA) as substrate, using a surface modification method similar as our previous study25 with the improvements. In brief, the tungsten probes were oxidated and hydroxylated firstly (details shown in the Supporting Information), and then immersed into n-octadecyldimethyl[3(trimethoxysilyl)propyl]ammonium chloride solution (5% in anhydrous toluene), heating at a temperature of 120 °C for reaction of 12 h (within a reaction kittle at nitrogen atmosphere), for modification of a layer of adsorbent at tungsten probe surface. Afterward, an exterior hydrophilic chitosan polymer was coated on the probe surface by dipping them into a 20 mL of chitosan solution (4 mg/mL, dissolved in 2% acetic acid) for ultrasonic irradiation of 30 min, follow adding a 10 mL of sodium tripolyphosphate solution (0.5 mg/mL) under ultrasonic concussion for 1 h. Sampling and Extraction. All the animal experiments were approved by the Animal Ethical and Welfare Committee of Sun Yat-sen University. A biocompatible surface-coated SPME probe was controlled by a three-dimensional manipulator with smallest microstep size of 1 µm (Fu-LiQian-Tian Opto-Electronics Technology, Beijing, China) to insert into a precise position of a biological organism or cell for sampling. The sampling and extraction time was 60 s. The details for sampling zebrafish, Daphnia magna, and single cells were demonstrated in the Supporting Information. Detection. After sampling, the loaded biocompatible surface-coated SPME probe was inserted into a nanospray tip (BG12-94-4-CE, New objective Inc., MA, USA) prefilled with spray solvent (1 µL for analysis of zebrafish and Daphnia magna, 0.2 µL for analysis of eukaryotic cells) for desorption of 30 s, and then mounted onto a 3-dimensional moving stage (Beijing Zolix instruments Co., Ltd, Beijing, China) and placed pointing to the MS inlet, adjusting to a position of 5 mm away from the MS inlet, and a high voltage of 2.5 kV was applied on the SPME probe for nanoESI-MS analysis. Mass Spectrometry. Mass spectra were acquired on a linear trap quadrupole (LTQ) Orbitrap Elite mass spectrometer 2



(Thermo Fisher Scientific, San Jose, CA, USA) or a Sarix X

Page 4 of 10

7T Fourier transform ion cyclotron resonance (FT-ICR) mass

Figure 1. Schematic diagrams for development of a BSCP-nanoESI-MS lipidomics methodology. a) A biocompatible surface-coated SPME probe toward lipids was designed. b) Extraction mechanism. c) Sampling of an individual zebrafish, Daphina magna, and HepG2 cell. d) Experimental setup of the BSCP-nanoESI-MS analysis. e) Mass data acquisition and pretreatment. f) Pattern recognition and biomarker identification.

spectrometer (Bruker Daltonics, Bremen, Germany), recording an appropriate m/z range with either positive or negative ion detection mode. Accurate mass measurement was

was further modified on the probe surface through ionic gelation (Figures S-2 and 3 in the Supporting Information). Chitosan is a resourceful biopolymer in nature, which has been successfully used in various biomedical applications such as wound healing materials, drug delivery systems, and gene delivery devices due to its favorable biocompatibility, biodegradability, nontoxicity, and non-immunogenicity.27,28 Previous cytotoxicity study has demonstrated that chitosanbased materials were nontoxic, highly biocompatible, and suitable for in vivo investigation.29 The chitosan polymer coating allows lipid species to enter the interior absorbent but prevents biological cells to pass through the coating via size exclusion,28 enhancing the biocompatibility and antiinterference ability of SPME probe for sampling complex biological samples. The obtained biocompatible surfacecoated SPME probe processes a tip-end diameter of less than 5 µm (Figure 2a). In most of cases, it can be easily inserted into small organisms and even single cells for microsampling, because the sizes of eukaryotic cells vary from several to dozens of µm, and many small organisms such as zebrafish and Daphnia magna have sizes of several to dozens of mm. Characterization of the Biocompatible Surface-Coated SPME Probe. To confirm the successful modification of adsorbent and external layer of polymer on the tungsten probe surface, an X-ray photoelectron spectroscopy (XPS) experiment was performed on the biocompatible surfacecoated SPME probe surface, by applying an ESCA LAB 250 XPS instrument (Thermo Fisher Scientific, San Jose, CA). In the obtained XPS spectrum (Figure S-4 in the Supporting Information), carbon-, oxygen-, tungsten-, nitrogen-, phosphorus-, silicon-relative peaks were evidently observed. The content percentages of Si, N, and P were 4.97%, 3.35% and 1.98%, respectively. Deconvolution of the Si 2p peak shows the presence of Si-O (102.2 eV) and Si-CH2 (102.8 eV), and deconvolution of the N 1s peak shows the presence of NCH2 (399.6 eV) and N-CH3 (402.2 eV), verifying the modification of the adsorbent on the tungsten probe surface. Deconvolution of the N 1s peak at 400.6 eV (NH3-CH) as well as P 2p peak at 134.1 eV (P-O) and 136.1 eV (P=O), derived

accomplished by LTQ-Orbitrap-MS with a 120,000 mass resolution mode or by FT-ICR-MS with a 4M recording mode (mass resolution of ~200,000 for lipid species). Multi-stage MS experiments were performed by LTQ-Orbitrap-MS, using high energy collision dissociation (HCD) mode with an appropriate isolation window and collision energy for each precursor ion. Xcalibur 2.2 (Thermo Fisher Scientific, San Jose, CA, USA) and FT-MS control (Bruker Daltonics, Bremen, Germany) software were used for equipment control and data acquisition.

RESULTS AND DISCUSSION

Preparation of the Biocompatible Surface-Coated SPME Probe. In general, a small size of tip-end for SPME probe will benefit its microsampling. Therefore, we selected a tungsten microdissecting probe with tip-end diameter of ~1 µm as substrate to prepare the biocompatible surface-coated SPME probe. Adsorbent makes major contribution to extraction ability for SPME probe, and thus its selection is extremely important. The major lipid species present in biological organisms and possessing important biological activities include glycerolipids, glycerophospholipids, sphingolipids, and fatty acyls, etc (Figure S-1 in the Supporting Information).26 These lipid species are amphipathic molecules contained long carbon chains and acidic/basic units. nOctadecyldimethyl[3-(trimethoxysiyl)propyl]ammonium chloride23,25 is selected as the adsorbent in this study, which shows high enrichment ability towards lipids. The nonpolar alkyl chains of lipids bond to the hydrophobic C18 group upon reversed-phase adsorption, and the phosphate/hydroxyl/carboxyl groups interact with the quaternary ammomium ion via ion-exchange adsorption. In addition, the nitrogen-atoms in lipids can also interact with the hydroxyl groups at tungsten probe surface (Figure 1b). nOctadecyldimethyl[3-(trimethoxysiyl)propyl]ammonium chloride was modified on the tungsten probe surface via silanization, and then an external layer of chitosan polymer 3


Page 5 of 10 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


from sodium tripolyphosphate, confirmed the functionalization

of external polymer layer on the tungsten probe surface.

Figure 2. SEM images from the a) tip (200×) and b) surface (12000×) of biocompatible surface-coated SPME probe, and AFM height image from surface of biocompatible surface-coated SPME probe.

The morphology of the biocompatible surface-coated SPME probe was revealed by scanning electron microscope (Quanta 400 FEG field emission SEM instrument, FEI, The Netherlands). In the obtained SEM images, the tip-end was measured to be ~5 µm (Figure 2a). A high-magnification SEM image showed abundant holes with diameters of hundreds of nm on the probe surface (Figure 2b). Atomic force microscope (AFM) was applied to further investigate the surface morphology of probe surface as well as the sizes of holes, by using a Fastscan Bio AFM instrument (Bruker Daltonics, Bremen, Germany). The obtained height images showed plenty of holes observed on the surface of the developed biocompatible SPME probe (Figure 2c). The root-mean-square roughness, average roughness, and max roughness of the coated material were measured at 18.0, 13.7, and 168 nm, respectively (Figure S-5a in the Supporting Information), and the depths and radii of the holes were measured in the range of 60~90 nm and 120~150 nm level, respectively (Figure S-5b in the Supporting Information). We calculated the molecular lengths of representative lipid species such as phosphatidylcholine (PC) and triradylglycerol (TAG) using Gaussian 09, and the results demonstrated that the maximum

for extraction, and high signal intensities for most lipid species were obtained. A longer extraction time did not increase the signal intensities obviously (Figure S-7 in the Supporting Information), and it might have a negative effect for in vivo sampling of living organisms. Different desorption/spray solvents, i.e., methanol, acetonitrile, methanol/water (v/v=1:1), acetonitrile/water (v/v=1:1), methanol/chloroform (v/v=1:1), and methanol/chloroform (v/v=9:1) were then investigated. Methanol/chloroform (v/v=9:1) gave the most abundant and stable signals, and thus it was selected as the desorption/spray solvent. The amount of desorption/spray solvent was also tested, and the signal duration time was stable for more than 3 minutes when 1 µL of desorption/spray was used (Figure 3a). Different desorption time (i.e., 10, 20, 30, 45, 60, and 120 s) was studied. It was found that the signal intensities increased with the increasing desorption time, and the equilibrium of desorption was reached at 30 s. Different high voltage (i.e., 2.0, 2.5, 3.0, 3.5, 4.0, 4.5, and 5.0 kV) were tested, and the optimized value was selected as 2.5 kV. It should be noted that the signal duration time of lipids decreased dramatically after the high voltage was set higher than 3.0 kV. The distance between the tip-end of nanospray tip and the MS inlet was optimized, and the optimum distance was selected as 5 mm. Both positive and negative ion detection were investigated. Most of the lipids showed their [M+Na]+ and/or [M+K]+ signals in positive ion detection mode, while in negative ion spectrum, only a few signals from fatty acids were obtained (Figure S-8 in the Supporting Information). The proposed BSCP-nanoESI-MS method was compared with conventional direct infusion shotgun MS lipidomics. Firstly, a living zebrafish was sampled and analyzed by BSCPnanoESI-MS, and d7-Rox was added into the desorption/spray solvent as internal standard (IS) with a concentration of 100 ng/mL. Abundant lipid signals at m/z 750-1000 were observed in the obtained positive ion mass spectrum (Figure 3b). After in vivo sampling, the zebrafish was euthanized to death, and 100 mg of muscle was applied for extraction and subsequent direct infusion shotgun MS analysis (d7-Rox was spiked in sample solution to control a similar signal intensities as BSCPnanoESI-MS method, with a concentration of 100 ng/mL). Similar lipid profile at m/z 750-1000 was observed in the obtained positive ion mass spectrum (Figure 3c), although the relative signal intensities for some lipid species were somewhat different. The result supported the feasibility and effectiveness of lipidomics investigation using BSCP-

theoretical lengths of these lipid molecules in all-trans conformation are in the range of 3-5 nm (Figure S-6 in the Supporting Information). Because the molecular sizes of lipid

species are much small than the holes of outer polymer, they can readily transfer through the pores of external coating and then be adsorbed by the internal adsorbent. In addition, the massive holes on the biocompatible surface-coated SPME probe surface can prevent the biological cells to pass through the solid coating via size exclusion, because the holes are small than biological cells (generally in several-µm levels), which provides a better biocompatibility and stability for analyzing complex biological samples. Establishment of BSCP-nanoESI-MS Methodology and Investigation of Extraction Performance. In order to establish a sensitive and effective BSCP-nanoESI-MS lipidomics methodology, experimental parameters including extraction time, type and amount of desorption/spray solvent, desorption time, high voltage, the distance between tip-end and MS inlet, and the positive and negative ion MS detection mode, etc., were all optimized by in vivo lipid analysis of a zebrafish. The extraction time was investigated firstly, and a series of extraction periods including 10, 20, 30, 60, 120, and 300 s were studied. The result revealed that 60 s was sufficient 4



Page 6 of 10

could move through the pores of chitosan layer and then be adsorbed by the internal adsorbent. The coating stability were also investigated, by comparing a biocompatible surface-coated SPME probe and a surfacecoated SPME probe (without the biocompatible chitosan polymer) for 100 times of extraction and analysis of zebrafish. A decreased coating stability and extraction performance with multiple use was observed for the surface-coated probe without biocompatible layer, while the coating property of biocompatible SPME probe was more stable, without obvious loss of extraction performance over 100 times of extraction and analysis (Figure S-9 in the Supporting Information). Note that every time of reuse, the biocompatible surface-coated SPME probe should be washed with methanol/chloroform (v/v=9:1) and pure water successively, and the carryover effect for the SPME probe was negligible (less than 5%) with such appropriate wash steps. SEM was further applied to investigate the morphology of the SPME probe after 100 times of use, by comparing with a new SPME probe without any use. The obtained SEM images clearly showed that their surface morphologies were quite similar, and the multi-used SPME probe did not contain obvious contaminations from complex biological samples (Figure S-10 in the Supporting Information). In addition, zebrafish is alive and recover from the damage easily after several days of sampling. All of these results demonstrated that the developed biocompatible surface-coated SPME probe possessed excellent biocompatibility and superior stability for sampling complex biological samples. In Vivo Lipidomics of Zebrafish and Microscale Lipidomics of Its Single Egg Cells. In vivo lipidomics experiment was performed to investigate zebrafish, a model organism widely used in the fields of ecotoxicology and medical sciences, etc.,30 by sampling its back muscle for BSCP-nanoESI-MS analysis. Abundant lipid signals at m/z 750−1000 was observed in the obtained mass spectrum (Figure 4a), and the lipid species were identified via the database of LIPID MAPS using the accurate mass measurement, fine isotope fingerprint measurement, and multi-stage MS experimental results. A FT-ICR-MS was applied to obtain accurate mass and fine isotope fingerprint simultaneously, which is helpful for molecular formula identification. A LTQ-Obitrap-MS was applied to perform HCD experiment, and accurate mass measurement was also carried out. There are several useful rules for lipid species identification. (i) Most of lipid species show their [M+Na]+ and [M+K]+ signals with high intensities in the MS spectrum, and their signals for [M+H]+ are relatively low, because biological tissues and fluids contain abundant dissociative Na+ and K+. (ii) When a certain amount of formic acid (e.g. 5%) is added into the spray solvent, the signals of [M+H]+ for most of lipid species increased obviously. (iii) When HCD experiment is performed to a precursor ion of [M+Na]+/[M+K]+, two fragment ions (generally with high signal intensities) with an m/z value difference of 22/38 were observed in the obtained MS/MS spectrum. (iv) Information regarding the fatty acid chains can be obtained from the HCD experimental result. For instance, the ion detected at m/z 782.5685 can be calculated as C42H82NO8PNa ([PC(34:1)+Na]+, 1.88 ppm) or C42H82NO8P ([PC(36:4)+H]+, -1.19 ppm) based on the LIPID MAPS searching result. The signal intensity of this ion was relatively high, and there was no obvious change in its signal intensity when 5% formic acid was added into spray solvent. When

nanoESI-MS. In addition, BSCP-nanoESI-MS shows the merits of in vivo sampling and much less sample consumption, and the zebrafish is still alive after sampling and analysis.

Figure 3. a) A signal duration chronogram for a representative BSCPnanoESI-MS analysis. Spectra obtained by b) BSCP-nanoESI-MS, c) direct infusion shotgun MS, d) nanoESI-MS using a fine tungsten probe without surface modification, and e) a chitosan-coated tungsten probe for analysis of a zebrafish. d7-Rox was used as IS with a concentration of 100 ng/mL.

To investigate the extraction capacity, experiment was performed by inserting a biocompatible surface-coated SPME probe and a fine tungsten probe (without the modification of internal adsorbent and external chitosan polymer) into the back muscle of a living zebrafish for extraction and enrichment of lipids simultaneously, and then for nanoESI-MS analysis. Much weaker lipids signals were observed in the obtained MS spectra by the unmodified tungsten probe (Figure 3d) as compared with those obtained by BSCP-nanoESI-MS spectrum (Figure 3b). The enrichment factors for the six dominated lipid signals, i.e., m/z 782.57, 798.54, 828.55, 869.70, 895.71, and 921.73 were calculated as 113 ± 10, 181 ± 17, 296 ± 25, 136 ± 15, 127 ± 12, and 101 ± 14 (n=6), respectively. The finding suggests that the developed biocompatible surface-coated SPME probe possesses high enrichment capacity toward lipid species. We also modified a fine tungsten probe with a chitosan polymer (without the modification of internal adsorbent) to compare its extraction capacity with the biocompatible surface-coated SPME probe. The results suggested that the chitosan-coated tungsten probe also showed certain extraction capacity for lipid species because of its rough surface with relatively large surface area (Figure 3e), with enrichment factors of 5-20 folds when compared with a fine tungsten probe without any surface modification. However, its extraction capacity was not as high as that of biocompatible surface-coated SPME probe (approximately 100-300 folds). The experimental result indicated the internal adsorbent played a prominent role while the external chitosan polymer played an auxiliary role in the lipid enrichment, and the lipid species in the zebrafish muscles 5


Page 7 of 10 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry [M+Na]+ and m/z 921.7327 [M+K]+), were dominantly observed, revealing that PCs and TAGs were the major lipid species present in zebrafish. To investigate the difference of lipid compositions in ovum and adult stages of zebrafish, lipidomics investigation for zebrafish egg cell was performed. In general, a single zebrafish egg cell has a size of 1-2 mm, and it can be easily sampled and analyzed by the proposed BSCP-nanoESI-MS method. A representative lipid profile of single zebrafish egg cell was shown in Figure 4b. It was observed that the dominantly detected lipid species were PC(16:0/18:1) and PC(16:0/22:6), and the signal intensities for many TAGs, i.e., m/z 895.7165 [TAG(16:0/18:1/18:2)+K]+, m/z 921.7324 [TAG(16:0/16:0/22:4)+K]+, and m/z 943.7173 [TAG(14:0/20:0/22:5)+K]+, etc., were dramatically decreased when compared with those of zebrafish muscle.

HCD experiment was upon to the ion, three major fragment ions were observed in the obtained MS/MS spectrum (Figure S-11a in the Supporting Information). The fragment ions at m/z 723.4923 (1.07 ppm) and 599.5003 (0.86 ppm) were corresponded to the successive lost of N(CH3)3 and C2H5PO4, respectively, which suggested the presence of the PC group. The fragment ion at m/z 577.5185 (-0.93 ppm), which showed a mass difference of 22 with m/z 599.5003, was corresponded to the loss of a sodium ion and adduction of hydrogen at m/z 599.5003. Thus, the ion detected at m/z 782.5685 was assigned as [PC(34:1)+Na]+ rather than [PC(36:4)+H]+. In addition, two fragment ions at m/z 467.2530 (-0.64 ppm) and 441.2371 (1.24 ppm), were corresponded to the loss of [N(CH3)3+FA(16:0)] and [N(CH3)3+FA(18:1)], respectively, while the fragment ions at m/z 239.2367 (-1.01 ppm) and 265.2524 (-0.73 ppm) were corresponded to [FA(16:0)-H2O]+ and [FA(18:1)-H2O]+, respectively. Based on the result, the ion detected at m/z 782.5685 was further identified as [PC(16:0/18:1)+Na]+. Another ion detected at m/z 798.5426 with high abundance was also identified analogously. In its HCD spectrum (Figure S-11b in the Supporting Information), the fragment ion peaks at 739.4659 (-2.12 ppm) and 615.4742 (-0.98 ppm) were assigned by the successive loss of N(CH3)3 and C2H5PO4, respectively. The fragment ion at m/z 577.5182 (-1.45 ppm), which showed a mass difference of 38 with m/z 615.4742, was ascribed to the loss of a potassium ion and adduction of hydrogen at m/z 615.4742. The signals of [M+KN(CH3)3-FA(16:0)]+, [M+K-N(CH3)3-FA(18:1)]+, [FA(16:0)H2O]+, and [FA(18:1)-H2O]+, were also observed with the fragmental ions at m/z 483.2268 (-0.90 ppm), 457.2112 (-0.84 ppm), 265.2523 (-1.10 ppm), and 239.2363 (-2.68 ppm), respectively. Thus, the ion at m/z 798.5426 (2.05 ppm) was identified as [PC(16:0/18:1)+K]+. It should be noted that the accurate location of C=C bond in the fatty acid chains could still not be assigned in this study, which is also a big challenge in lipidomics investigation nowadays. In addition, for an ambient mass spectrometric methodology, the isomers could not be differentiated and showed the same m/z values in the mass spectrum, because there is no chromatography separation. For instance, the ion detected at m/z 808.5847 were identified as [PC(36:2)+Na]+ (2.50 ppm) based on the LIPID MAPS searching result. In its HCD spectrum (Figure S-11c in the Supporting Information), three fragment ions at m/z 263.2370, 265.2524, and 267.2684, which correspond to [FA(18:2)-H2O]+, [FA(18:1)-H2O]+, [FA(18:0)-H2O]+, respectively, were observed with ion intensity ration of 1:2:1. Thus, this ion was further identified as the mixture of [PC(18:0/18:2)+Na]+ and + [PC(18:1/18:1)+Na] . Such a problem is also existed in conventional direct infusion shotgun MS lipidomics method. The LIPID MAPS searching result demonstrated a total of 137 lipid species (isomers excluded) including 3 fatty acids (FAs), 2 diradylglycerols (DAGs), 57 TAGs, 33 PCs, 28 phosphoglycerols (PGs), 10 ceramide-1-phosphates (CerPs), 2 cholesteryl esters (CEs), 1 monogalactosyldiacylglycerol (MGDG), and 1 phosphtatidylinositols-ceramide (PI-Cer) were detected and identified (Table S-1 in the Supporting Information). Among the detected lipid species, signals for many PCs and TAGs, i.e., PC(16:0/18:1) (m/z 782.5673 [M+Na]+ and m/z 798.5412 [M+K]+), PC(16:0/22:6) (m/z 828.5513 [M+Na]+ and m/z 844.5248 [M+K]+), + TAG(16:0/18:1/18:2) (m/z 879.7422 [M+Na] and m/z 895.7150 [M+K]+), and TAG(16:0/16:0/22:4) (m/z 905.7584

Figure 4. BSCP-nanoESI-MS spectra obtained by analyzing the a) back muscle and b) one egg cell of zebrafish. c) PCA score and d) loading plots of the investigated zebrafishes and their egg cells.

Subsequently, a total of 10 zebrafishes were in vivo sampling for SCP-nanoESI-MS analysis, and single egg cells from each zebrafish were also analyzed. The obtained MS spectra were imputed into SIMCA-P for principal component analysis (PCA). The obtained PCA score plot (Figure 4c) showed that the data points of zebrafishes and their egg cells were clearly separated into two clusters, which unambiguously revealed the significant difference of lipid profiles for zebrafish and its egg cell. PCA loading plot was further employed to recognize the lipid markers responsible for the variance. From the obtained PCA loading plot (Figure 4d), it was observed that the ions at m/z 895.716, 897.732, 893.701, 921.732, 919.715, 869.758, 871.717, 879.743, and 923.748 (which were corresponded to 6



[TAG(16:0/18:0/18:2)+K]+, [TAG(16:0/18:1/18:2)+K]+, [TAG(16:0/18:2/18:2)+K]+, [TAG(16:0/16:0/22:4)+K]+, + [TAG(16:0/16:0/22:5)+K] , [TAG(16:0/17:0/18:1)+Na]+, + [TAG(16:0/16:0/18:1)+K] , [TAG(16:0/18:1/18:2)+Na]+, and [TAG(16:0/16:0/22:3)+K]+, respectively) showed great positive correlation with zebrafishes while small correlation with zebrafish egg cells, suggesting these lipid species were the potential biomarkers for discriminating zebrafishes and their egg cells. Such a result demonstrated that TAG species increased during the growth process of zebrafish, and the lipid compositions of zebrafish were different from its ovum stage. In Situ Lipidomics of Daphnia Magna. Owing to the micrometer scale tip-end for the biocompatible surface-coated SPME probe, our proposed BSCP-nanoESI-MS lipidomics methodology could be applied to in situ investigation of different parts and organs of individual small organisms. In this study, a model organism, i.e., Daphnia magna (with size of 3-5 mm), was selected for in situ lipidomics investigation. Daphnia magna is an important indicator for aquatic ecosystem health and ecotoxicology, playing prominent roles in transporting energy and nutrients in aquatic ecosystems.31,32 Selectively sampling different parts of Daphnia manga, i.e., head, abdomen, back, and tail, was readily achieved, because the tip-end of the biocompatible surface-coated SPME probe was much smaller than Daphnia magna’s body. Lipidomics investigation of Daphnia magna’s abdomen was performed firstly, and abundant signals for lipid species at m/z 700−950 were obtained (Figure 5a). Then, lipidomics for its back (Figure 5b), head (Figure 5c), and tail (Figure 5d) were investigated consecutively. It was observed that their lipid species were similar but the contents fluctuated dramatically in different parts of Daphnia manga, and the concentration order from high to low was abdomen, back, head, and tail.

Page 8 of 10

Figure 5. BSCP-nanoESI-MS spectra for analysis of the a) abdomen, b) back, c) head, and d) tail of individual Daphnia magna. e) Distribution of PC (16:0/18:3) and TAG (16:0/16:4/18:3) in Daphnia magna. d7-Rox was used as IS, adding into the spray solvent with a concentration of 50 ng/mL.

Identification of the lipid species was also accomplished via the database of LIPID MAPS, and a total of 152 lipid species (isomers excluded) were identified (Table S-2 in the Supporting Information). The identified lipids contained 2 FAs, 28 PCs, 80 TAGs, 6 MGDGs, 11 PGs, 1 CerP, 3 phosphatidylethanolamines (PEs), 15 phosphatidic acids (PAs), 1 sulfatide (SHexCer), and 5 wax esters (WEs). Among the detected lipids, the sodium/potassium adduct signals of TAGs and PCs, i.e., TAG(16:0/16:4/18:3) (m/z 843.6456 [M+Na]+ and m/z 859.6194 [M+K]+), TAG(16:0/16:1/20:5) (m/z 873.6917 [M+Na]+ and m/z 889.6660 [M+K]+), + TAG(18:1/18:3/18:3) (m/z 899.7086 [M+Na] and m/z 915.6817 [M+K]+), PC(16:0/18:3) (m/z 778.5344 [M+Na]+) PC(16:0/18:1) (m/z 782.5657 [M+Na]+), PC(18:2/18:2) (m/z 804.5511 [M+Na]+), and PC(14:0/18:1) (m/z 754.5354 [M+Na]+) were dominantly detected. To clearly visualize the distribution of the lipid species, PC (16:0/18:3) (the highest PC) and TAG (16:0/16:4/18:3) (the highest TAG) were selected as the representative lipids for imaging. A total of 10 points (5 in abdomen, 3 in back, 1 in head, and 1 in tail) of one Daphnia manga were sampled and analyzed, and the obtained values of Im/z778.53/Im/z 844.58 and Im/z843.65/Im/z 844.58 were imputed into Surfer 9 software for imaging (Figure 5e). Obviously, TAG (16:0/16:4/18:3) in abdomen was dramatically higher than that in back, head and tail, which might be attributed to many viscus containing abundant TAGs in the abdomen of Daphnia manga. It could also be observed that the concentration of PC (16:0/18:3) in abdomen was higher than that in back, head and tail, although its concentration fluctuation in different parts was not as obvious as that of TAG (16:0/16:4/18:3). In head and tail, the concentration of PC (16:0/18:3) was even higher than the concentration of TAG (16:0/16:4/18:3). Lipidomics of Eukaryotic Cells. The proposed BSCPnanoESI-MS methodology is also suitable for lipid analysis of single eukaryotic cells and lipidomics investigation of small cell populations. Generally, the sizes of eukaryotic cells could vary from several to dozens of µm. In most cases, the developed biocompatible surface-coated SPME probe can be successfully inserted into a single eukaryotic cell for sampling and enrichment of lipids. Attempt has been made to analyze a single HepG2 cell, and a small quantity of lipid signals at m/z 700−900 was observed (Figure 6a). However, the signal duration time of lipids was very short, remaining only 3-5 s (Figure 6b). Thus, it was difficult for HCD experiments, and even difficult for a signal accumulation experiment by FTICR-MS. Lipidomics was further performed to small HepG2 cell population, and a total of 100 HepG2 cells were analyzed. Abundant lipid signals were obtained (Figure 6c), with a stable signal duration time of approximately 40 s (Figure 6d), and thus the identification of lipid species was easily performed. As a result, a total of 60 lipids (isomers excluded) including 38 PCs, 4 lysophosphatidylcholines (LPCs), 9 PGs, 1 PE, 1 lysophosphatidylethanolamine (LPE), 1 CE, 3 SMs, 1 ShexCer, 1 PA, and 1 phosphatidylethanolamines-ceramide (PE-Cer) were identified (Table S-3 in the Supporting Information). Among them, the signals of 7


Page 9 of 10 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


[PC(16:0/18:1)+Na]+ (m/z 782.5680), [PC(14:0/18:1)+Na]+ (m/z 754.5366), [PC(18:0/18:2)+Na]+ (m/z 808.5830), and [PC(14:0/16:0)+Na]+ (m/z 728.5217) were dominantly detected (Figure 6e). Such strong ions of PCs are consistent with the fact that these species contributes the majority of biological membranes.

2017A030310233), Science and Technology Planning Project of Guangdong Province, China (No. 2016A040403057, 2013B031500003, 2016B020240006, and 2017A070702017), Science and Technology Planning Project of Guangzhou City (No. 201804010298), Fundamental Research Funds for the Central Universities (No.171gpy95), and GDAS’ Special Project of Science and Technology Development (No. 2017GDASCX-0104 and 2018GDASCX-0921).

Figure 6. BSCP-nanoESI-MS spectrum and b) extracted ion chronogram of m/z 782.57 from a single HepG2 cell. c) BSCP-nanoESI-MS spectrum and d) extracted ion chronogram of m/z 782.57 from 100 HepG2 cells. e) Magnified mass spectrum from 100 HepG2 cells.

CONCLUSION

In summary, we have demonstrated the development of a novel BSCP-nanoESI-MS method and its application for in vivo, in situ, and microscale lipidomics investigation of small biological organisms and eukaryotic cells. By using our proposed BSCP-nanoESI-MS method, in vivo lipidomics investigation of living zebrafish was accomplished, distribution imaging of lipid species in different parts of Daphnia magna was obtained, and microscale lipid analysis of a single HepG2 cell was achieved. The capability of in vivo, in situ, and microscale analysis makes the proposed BSCPnanoESI-MS methodology a promising tool for lipidomics investigation.

ASSOCIATED CONTENT

Supporting Information Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.

AUTHOR INFORMATION

Corresponding Author * E-mail: [email protected]. Tel.: +86-20-37656885-823. * E-mail: [email protected]. Tel.: +86-20-84112958. Notes The authors declare no competing financial interest.

REFERENCES

(1) Ma, X.; Xia, Y. Angew. Chem. Int. Ed. 2014, 53, 2592-2596. (2) Gasparovic, C.; Rosenberg, G. A.; Wallace, J. A.; Estrada, E. Y.; Roberts, K.; Pastuszyn, A.; Ahmed, W.; Graham, G. D. Neurosci. Lett. 2001, 301, 87-90. (3) Prache, N.; Abreu, S.; Sassiat, P.; Thiébaut, D.; Chaminade, P. J. Chromatogr. A 2016, 1464, 55–63. (4) Wang, K.; Jiang, D.; Sims, C. E.; Allbritton, N. L. J. Chromatogr. B 2012, 907, 79-86. (5) Kamphorst, J. J.; Fan, J.; Lu, W.; White, E.; Rabinowitz, J. D. Anal. Chem. 2011, 83, 9114–9122. (6) Cífková, E.; Michal Holčapek; Lísa, M.; Ovčačíková, M. n.; Lyčka, A.; Lynen, F. d. r.; Sandra, P. Anal. Chem. 2012, 84, 1006410070. (7) Bail, S.; Stuebiger, G.; Unterweger, H.; Buchbauer, G.; Krist, S. Eur. J. Lipid Sci. Technol. 2009, 111, 170-182. (8) Calvano, C. D.; Monopoli, A.; Ditaranto, N.; Palmisano, F. Anal. Chim. Acta 2013, 798, 56-63. (9) Wang, M.; Wang, C.; Han, R. H.; Han, X. Prog. Lipid Res. 2016, 61, 83-108. (10) Cooks, R. G.; Ouyang, Z.; Takats, Z.; Wiseman, J. M. Science 2006, 311, 1566-1570. (11) Venter, A.; Nefliu, M.; Cooks, R. G. Trends Anal. Chem. 2008, 27, 284-290. (12) Yang, Y.; Huang, Y.; Wu, J.; Liu, N.; Deng, J.; Luan, T. Trends Anal. Chem. 2017, 90, 14-26. (13) Eberlin, L. S.; Dill, A. L.; Golby, A. J.; Ligon, K. L.; Wiseman, J. M.; Cooks, R. G.; Agar, N. Y. R. Angew. Chem. Int. Ed. 2010, 49, 5953-5956. (14) Eberlin, L. S.; Ifa, D. R.; Wu, C.; Cooks, R. G. Angew. Chem. Int. Ed. 2010, 49, 873-876. (15) Mandal, M. K.; Yoshimura, K.; Chen, L. C.; Yu, Z.; Nakazawa, T.; Katoh, R.; Fujii, H.; Takeda, S.; Nonami, H.; Hiraoka, K. J. Am. Soc. Mass. Spectrom. 2012, 23, 2043-2047. (16) Liu, Y.; Zhang, J.; Nie, H.; Dong, C.; Li, Z.; Zheng, Z.; Bai, Y.; Liu, H.; Zhao, J. Anal. Chem. 2014, 86, 7096-7102. (17) Shrestha, B.; Sripadi, P.; Reschke, B. R.; Henderson, H. D.; Powell, M. J.; Moody, S. A.; Vertes, A. PLoS ONE 2014, 9, e115173. (18) Lee, J. K.; Jansson, E. T.; Nam, H. G.; Zare, R. N. Anal. Chem. 2016, 88, 5453-5461. (19) Phelps, M.; Hamilton, J.; Verbeck, G. F. Rev. Sci. Instrum. 2014, 85, 124101. (20) Hiyama, E.; Ali, A.; Amer, S.; Harada, T.; Shimamoto, K.; Furushima, R.; Abouleila, Y.; Emara, S.; Masujima, T. Anal. Sci. 2015, 31, 1215-1217. (21) Deng, J.; Yang, Y.; Wang, X.; Luan, T. Trends Anal. Chem. 2014, 55, 55-67. (22) Fang, L.; Deng, J.; Yang, Y.; Wang, X.; Chen, B.; Liu, H.; Zhou, H.; Ouyang, G.; Luan, T. Trends Anal. Chem. 2016, 85, 61-72. (23) Deng, J.; Yang, Y.; Fang, L.; Lin, L.; Zhou, H.; Luan, T. Anal. Chem. 2014, 86, 11159-11166. (24) Gómez-Ríos, G. A.; Pawliszyn, J. Angew. Chem. Int. Ed. 2014, 53, 14503-14507. (25) Deng, J.; Yang, Y.; Xu, M.; Wang, X.; Lin, L.; Yao, Z. P.; Luan, T. Anal. Chem. 2015, 87, 9923-9930. (26) Li, M.; Yang, L.; Bai, Y.; Liu, H. Anal. Chem. 2014, 86, 161175. (27) Park, B. K.; Kim, M.-M. Int. J. Mol. Sci. 2010, 11, 5152-5164. (28) Zhang, X.; Niu, H.; Pan, Y.; Shi, Y.; Cai, Y. Anal. Chem. 2010, 82, 2363-2371.

ACKNOWLEDGMENT

This research was financially supported by the National Natural Science Foundation of China (No. 21707171, 21625703, 41473092, 21677183, and 21777150), Natural Science Foundation of Guangdong Province, China (No. 8



(29) Henricus, M. M.; Fath, K. R.; Menzenski, M. Z.; Banerjee, I. A. Macromol. Biosci. 2009, 9, 317-325. (30) Hölttä-Vuori, M.; Salo, Veijo T. V.; Nyberg, L.; Brackmann, C.; Enejder, A.; Panula, P.; Ikonen, E. Biochem. J 2010, 429, 235-242. (31) Dai, Z.; Xia, X.; Guo, J.; Jiang, X. Chemosphere 2013, 90, 15891596. (32) Xia, X.; Rabearisoa, A. H.; Jiang, X.; Dai, Z. Environ. Sci. Technol. 2013, 47, 10955-10963.

9


Page 10 of 10

of 10 ACS Paragon Plus Environment Analytical Chemistry 1 2 3 4 5 6

Recommend Documents