Article pubs.acs.org/ac
Pulsed Direct Current Electrospray: Enabling Systematic Analysis of Small Volume Sample by Boosting Sample Economy Zhenwei Wei, Xingchuang Xiong,† Chengan Guo,‡ Xingyu Si, Yaoyao Zhao, Muyi He,§ Chengdui Yang, Wei Xu,§ Fei Tang,‡ Xiang Fang,† Sichun Zhang, and Xinrong Zhang* Beijing Key Laboratory for Microanalytical Methods, Instrumentation, Department of Chemistry, Tsinghua University, Beijing 100084, P. R. China S Supporting Information *
ABSTRACT: We had developed pulsed direct current electrospray ionization mass spectrometry (pulsed-dc-ESIMS) for systematically profiling and determining components in small volume sample. Pulsed-dc-ESI utilized constant high voltage to induce the generation of single polarity pulsed electrospray remotely. This method had significantly boosted the sample economy, so as to obtain several minutes MS signal duration from merely picoliter volume sample. The elongated MS signal duration enable us to collect abundant MS2 information on interested components in a small volume sample for systematical analysis. This method had been successfully applied for single cell metabolomics analysis. We had obtained 2-D profile of metabolites (including exact mass and MS2 data) from single plant and mammalian cell, concerning 1034 components and 656 components for Allium cepa and HeLa cells, respectively. Further identification had found 162 compounds and 28 different modification groups of 141 saccharides in a single Allium cepa cell, indicating pulsed-dcESI a powerful tool for small volume sample systematical analysis.
T
However, the drawback was also obvious since the information on many low-abundance components could be missing. Another feasible solution to this bottleneck problem could be boosting the sample economy of Nano-ESI, elongating the MS signal duration for collecting MS2 data from picoliter volume sample without sample dilution. Actually, Nano-ESI generated continuous electrospray while the mass analyzer of MS worked intermittently. Theoretically, if we could generate pulsed electrospray and perfectly synchronized its frequency with the mass analyzer, we could boost the sample economy of NanoESI by avoiding meaningless sample loss between the timeadjacent MS scan events. Herein, we reported a constant voltage pulsed electrospray ionization mass spectrometry (pulsed-dc-ESI-MS, Figure 1a) to enable systematic profile of components in small volume sample. By simply setting an electrode with dc voltage applied near but apart from the sample solution in the emitter tip, we could generate pulsed electrospray. The polarity of electrospray was controlled by electrode polarity (Figure 1c). By adjusting the electrode voltage, we could precisely control the frequency of pulsed spray (Figure 1b). Different from other pretty proved pulsed electrospray techniques achieved by piezoelectric
he surge of system biology has brought analytical chemistry to “big data” era.1 Nanoelectrospray ionization mass spectrometry (Nano-ESI-MS) was a popular analytical tool for system biology research such as proteomics and metabolomics; however, only suitable for sample with volume larger than nanoliter level.2−6 In the recent years, more and more biological research concerns about small volume samples (picoliter to nanoliter level), especially sample with picoliter volume, such as single cell analysis,7−12 cell niche analysis,13−15 and small volume physiological fluid analysis.16 To perform systematical analysis of such a small volume sample by NanoESI-MS is a nearly impossible task. The biggest obstacle was that only very short or even instantaneous MS signal duration could be acquired from picoliter sample by Nano-ESI17,18 while systematical analysis demanded a period of time to further obtain the MS2 data for the components determination. Besides this, the electrical contact of sample solution and electrode remained a technical problem when using Nano-ESI for small volume samsple analysis. Cooks et al. mentioned that “In ESI, electrical contact with a voltage supply is necessary to generate a continuous spray of charged droplets from a solution. The electrical contact adds dead volume and adsorption surfaces”.19 Therefore, the development of a MS tool for systematic analysis of picoliter sample was urgently demanded. Diluting a small volume sample to the nanoliter or even microliter level for Nano-ESI analysis might be a solution.20,21 © XXXX American Chemical Society
Received: June 4, 2015 Accepted: October 21, 2015
A
DOI: 10.1021/acs.analchem.5b02115 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
DMEM medium with 10% fetal bovine serum and 100 IU/mL penicillin-streptomycin in a 37 °C humidified atmosphere containing 5% CO2. Preparation of Emitters. Borosilicate glass capillaries were pulled by P-2000 (Sutter Instrument) to make the emitters (i.d. of the tip is 1 μm, i.d. = 0.6 mm, o.d. = 1 mm, length = 55 mm) for pulsed-dc-ESI. The parameters of P-2000 were as followed: heat = 345, FIL = 5, VEL = 28, DEL = 128, and PUL = 60. Pulsed-dc-ESI Source. After the sample was loaded into the emitter, the emitter was placed in a holder for CV-PulsedESI analysis. The holder included a straight stainless steel tube (i.d. = 1.2 mm, length = 3 cm). The rear part of the emitter was inserted into this tube thereby the emitter could be fixed. Inside the hold, a steel needle (o.d. = 0.3 mm) was insert into the emitter from its rear part. A dc voltage was applied to this needle to provide the static electrical field. Typically, the distance from the needle tip to the sample was 5 mm. The holder was fixed on the cantilever of a four-dimensional mobile device (x, y, z and θy, Beijing Optical Century Instrument Co., Ltd.). By adjusting the mobile device, the distance between the emitter tip and MS inlet and spray angle could be controlled. Typically, the electrode voltage was +1.5 kV for positive mode analysis and −1.2 kV for negative mode analysis; the distance from emitter tip to MS inlet was 5 mm. Sampling Operation. All the single cell sampling experiments were performed with the aid of a three-dimensional mobile manipulator (MP-225, Sutter Instrument). The procedure was observed by an inverted microscope (DX30, Dayueweijia Science and Technology Co. Ltd., Beijing). Picoliter Level Solution Sampling. The emitter tip was dipped into the solution with a depth about 1 cm for 10 s, and 370 ± 30 pL solution (observed by a microscope, YX20L20, Dayueweijia Science and Technology Co. Ltd., Beijing) was loaded into the emitter Single Allium cepa Cell Sampling. The epidermis of Allium cepa was ripped out by tweezers and placed on a glass slide. A volume of 1 mL of water was added to the slide to wash off the cytoplasm leaked out from broken cells and then drained by filter paper. The glass slide was placed on the single cell manipulator platform. The emitter was controlled by the manipulator to penetrate the cell. Under the surface tension effect, the cell extract was sucked into the emitter tip for Pulsed-dc-ESI-MS analysis. Single HeLa Cell Sampling. The cell culture dish was placed on the single cell manipulator platform. The acupuncture needle was controlled by the manipulator to pick and move the target cell from the cell culture dish. The needle was then inserted and dipped into spray solvent in an emitter preloaded with about 20 nL of acetonitrile in the tip for 30 s to ensure the cell lysis and sufficient dissolution. Special cautions: the acupuncture needle must have been ultrasonically cleaned by acetonitrile for 10 min before use; after the cell was picked, lift up the needle carefully to avoid the cell dropping from needle until the needle tip moved out of the cell culture solution. Mass Spectrometry. All the experiments of analytical performance evaluation was carried out using Thermo LTQ mass spectrometer (Thermo Scientific, San Jose CA). The instrument parameters were as followed: capillary temperature = 275 °C, max injection time = 200 ms. The single cell metabolomics analysis was carried out using Thermo QEOrbitrap mass spectrometer (Thermo Scientific, San Jose CA). The capillary temperature was 320 °C. The analysis methods were full MS and ddMS2. The detailed settings were as
Figure 1. (a) Setup of pulsed-dc-ESI. The electrode with diameter of 0.3 mm was set contactless from the sample solution. The distance between electrode and solution was 5 mm. The distance from emitter tip to MS inlet was 5 mm. The dc voltage of 1.5 kV was applied to the electrode to generate single polarity pulsed electrospray of the sample solution. (b) Adjusting the pulsed spray frequency by electrode voltage. A volume of 1 μL of Somatostatin (10 ppm) was loaded into the emitter. The probe of the oscilloscope was set very near to the emitter tip to detect the pulsed electrospray. (c) To control the polarity of pulsed-dc-ESI by the polarity of power supply. The testing sample was 100 ppm maltoheptaose.
dispensing,22 pulsed voltage23−27 and square wave supply,28−30 Pulsed-dc-ESI had three obvious advantages. (1) Ionization without sample−electrode contact. (2) The pulsed spray could be generated by dc power supply instead of special designed power supply; therefore, no modification to the commercial Nano-ESI was needed. (3) The electrospray in pulsed-dc-ESI was single polarity instead of dual polarity, which could achieve higher sample economy, since usually only one polarity was utilized in systematic analysis.
■
MATERIALS AND INSTRUMENT SETTINGS Materials. Caffeine, maltoheptaose, and cytochrome C were purchased from Sigma; angiotensin II and somatostatin were purchased from Shanghai Chutai Biological Technology Company, Ltd.; indole acetic acid and jasmonic acid were purchased from Beijing Biodee Biotechnology Co. Ltd. The chemicals and medium used for cell culture were all purchased from Corning (NY). The steel electrode (o.d. = 0.3 mm) was purchased from Beijing Siboer Metal Material Co. Ltd. The steel acupuncture needle (o.d. = 0.25 mm, length = 5 mm) for mammalian cell picking was purchased from Beijing Hanyi Material Instrumental Centre. The borosilicate glass capillaries (i.d. = 0.6 mm, o.d. = 1 mm, without filament) were purchased from Vital Sense Scientific Instruments Co. Ltd. The purple Allium cepa bulb was purchased from local market. The HeLa cell lines were purchased from ATCC. Cells were cultured in B
DOI: 10.1021/acs.analchem.5b02115 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry Table 1. Average Flow Rate Achieved by Nano-ESI and Pulsed-dc-ESI for Different Samplesa average flow rate
signal duration (s) compounds indole acetic acid jasmonic acid somatostatin angiotensin II maltoheptaose cytochrome C Allium cepa cell extract
polarity negative negative positive positive positive positive positive
volume (pL) 370 ± 370 ± 370 ± 370 ± 370 ± 370 ± ∼900
30 30 30 30 30 30
nanoESI
pulsed-dcESI
pulsed-dc-ESI/ nano-ESI
nano-ESI (nL/ min)
pulsed-dc-ESI (pL/ min)
pulsed-dc-ESI/nano-ESI (%)
0.9 0.6 1.9 0.7 1.2 1.2 6.6
35.0 25.2 62.7 31.1 57.3 28.6 345.1
38.9 42.0 33.0 44.4 47.8 23.8 52.3
26.7 40.0 12.6 34.3 20.0 20.0 8.2
685.7 952.4 382.8 771.7 418.8 839.2 156.5
2.6 2.4 3.0 2.3 2.1 4.2 1.9
a The volumes of these samples were determined by the method showed in Figures S1 and S6. More detailed information on these samples were listed in Table S1. For the same sample loading amount, the signal duration of pulsed-dc-ESI were 17.4 to 52.3 times than those of nano-ESI. The average flow rates of pulsed-dc-ESI were 157−952 pL/min, which were only 1.9% to ∼5.7% of those of nano-ESI. The total ion chronogram data were listed in Figure S4.
followed. Method duration was 3 min (for Allium cepa cell) or 5 min (HeLa cell). Full MS: resolution = 70 000, AGC = 3 × 106, max injection time = 100 ms, scan range = 134−2000 (Allium cepa cell) or 68−1000 (HeLa cell). The ddMS2 settings: resolution = 35 000, AGC = 1 × 105, ion injection time = 100 ms, loop count = 10, NCE = 50, underfill ratio = 0.1%, dynamic exclusion = 300 s. Metabolome Identification. The spectral data and instrumental information were extracted from the data files (.RAW files) by a MS data translation tool. The data of every scan event was placed in CSV files with the name according to the scan number for further analysis and search. The MS2 data of small molecular weight metabolites (100 to 200) were searched in Massbank database (http://www.massbank.jp) for matching. MS filter search: a program written by MATLAB was used to find all parent ions with typical phospholipid fragments (69.07, 81.07, and 95.09); the threshold was 5 000 (only if the parent ions with the maximum intensity of fragments exceeding 5 000 was considered). Neutral loss search: a program written by MATLAB was used to find all parent ions having the searched neutral loss value in the fragments. All information on parents ions of saccharides and their modified products could be retrieved by using 162 (the m/z of a saccharide group) as the neutral loss search value. The threshold was 0.1 (only if the relative intensity of both the fragments exceeding 0.1 was considered).
long and stable enough for collecting MS/MS information from single cell samples in this study. We further examined the sampling economy by somatostatin (10 ppm, 1 μL, Figure 2). We had obtained similar intensity MS signal by both nano-ESI and pulsed-dc-ESI; however, the spray current of pulsed-dc-ESI was only 1% of nano-ESI (450 pA compared to 40 nA). This result indicated the fact that pulsed-dc-ESI sampled less ions into MS; however, the targeted ions amount was still sufficient to saturate the ion-trap. This
■
RESULTS AND DISCUSSION We had carried out a series of experiments with artificial picoliter solution (Table S1 and Figure S1) to demonstrate that pulsed-dc-ESI strategy could boost the sample economy and significantly elongate the MS signal duration. By comparison to nano-ESI when +1.5 kV dc voltage applied, the average spray frequency of pulsed-dc-ESI for somatostatin solution was 150 Hz (Figure S2) and the flow rate was only 3.0% of nano-ESI (Table 1). We also compared the flow rate of pulsed-dc-ESI and nano-ESI for different samples under this electrode voltage. The flow rate of all the samples reduced to 1.9% to ∼4.2% of nano-ESI (Table 1). Theoretically, the ultimate flow rate could be achieved if precisely synchronizing pulsed electrospray with MS detection (about 10 Hz). However, for convenient consideration, we chose 150 Hz pulsed spray (under 1.5 kV) in the present study. The pulsed spay under this frequency could generate a continuous and stable total ion chronogram with variation of 11.3% (Figure S3), which had been proved
Figure 2. Comparison of sampling efficiency of pulsed-dc-ESI and nano-ESI. A volume of 1 μL of somatostatin (10 ppm) was loaded into emitter. The sample was analyzed by pulsed-dc-ESI followed nano-ESI. The spray current was recorded by micro current monitor couple with a Faraday cup as the probe. For the same sample and similar MS signal intensity (4.4 × 105 by pulsed-dc-ESI versus 5.3 × 105 by nano-ESI), the spray current of pulsed-dc-ESI was 450 pA, only 1% of that of nano-ESI (40 nA). This indicated that actually 450 pA spray current was capable to generate sufficient ions for ion-trap analysis. Admittedly, nano-ESI sampled more ions into MS (spray current was 40 nA); however, this not only consumed more sample but also sampled too much solvent into MS vacuum system, which would lead to relatively poorer solvent desolvation and higher baseline. Therefore, the pulsed sampling strategy did not affect sensitivity. On the contrary, it could help enhance MS sensitivity by improvement of desolvation. C
DOI: 10.1021/acs.analchem.5b02115 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry Table 2. Signal Intensity and SNR Achieved by Nano-ESI and Pulsed-dc-ESI for Different Samplesa signal intensity compounds
polarity
indole acetic acid jasmonic acid somatostatin angiotensin II maltoheptaose cytochrome C
negative negative positive positive positive positive
volume (pL)
nano-ESI
± ± ± ± ± ±
9.2 × 1003 4.50 × 1003 7.9 × 1004 3.9 × 1004 7.0 × 1004 1.0 × 1005
370 370 370 370 370 370
30 30 30 30 30 30
signal to noise ratio
pulsed-dc-ESI
pulsed-dc-ESI/nano-ESI
nano-ESI
pulsed-dc-ESI
pulsed-dc-ESI/nano-ESI
× × × × × ×
6.49 0.56 0.22 4.10 0.60 0.49
71.9 53.5 97.9 21.0 15.7 34.0
350.9 344.8 575.6 624.1 29.2 746.2
4.9 6.4 5.9 29.7 1.9 21.9
6.0 2.5 1.7 1.6 4.2 4.9
1004 1003 1004 1005 1004 1004
a
The volumes of these samples were determined by the method shown in Figure S1. More detailed information on these samples were listed in Table S1. The reduced flow rate of pulsed-dc-ESI would not reduce the SNR of mass spectra. On the contrary, pulsed-dc-ESI could increase the SNR of mass spectra up to 29.7 times. The mass spectra were listed in Figure S5.
result proved that the pulsed electrospray did enhance the sampling economy without loss of MS sensitivity. We also compared the performance of pulsed-dc-ESI and nano-ESI on some small volume sample with a picoliter amount. We used signal-to-noise ratio (SNR) and limit of detection (LOD) as two factors to evaluate the MS sensitivity. We used six compounds (Table S1) for comparison of SNR achieved by nano-ESI and pulsed-dc-ESI (Table 2 and Figure S5). From Table 2, pulsed-dc-ESI achieved better SNR than nano-ESI for all the samples we tested. For example, the SNR of angiotensin II was 21.0 and 624.1 for nano-ESI and pulseddc-ESI, respectively, indicating that 29.7 times enhanced MS sensitivity was achieved by pulsed-dc-ESI. A probable reason to this phenomenon could be that pulsed-dc-ESI achieved zerodead volume analysis, avoiding the sample dilution in NanoESI. Because in the experiment of small volume sample analysis such as single-cell analysis, spray solvent was loaded into the emitter from the rear part to ensure electrical contact of the electrode and sample solution, which could lead to the sensitivity loss. We also compared the LOD of pulsed-dc-ESI and nano-ESI (Figure 3). Taking somatostatin as an example, pulsed-dc-ESI could detect 100 ppb somatostatin while that of nano-ESI was only 1 ppm. Considering the average flow rate, the absolute detection amount of pulsed-dc-ESI for somatostatin could reach 200 zmol, which was 1000 times lower than nano-ESI (240 amol). Therefore, pulsed-dc-ESI was a sensitive analytical tool which was qualified for small volume sample analysis. A theoretical model was proposed to explain the ionization behavior of pulsed-dc-ESI based on our experiment observations and prior work on the electrospray mechanism.31,32 Take the positive mode as an example, the procedure to generate one pulsed spray included four steps: solution polarization, positive electrospray, electrochemical reaction in liquid−gas surface, and discharge between gas and electrode, namelys steps 1−4, respectively (Figure 4). In detail, the solution polarization could be the first step of electrospray. Figure S7 showed the evidence for the solution polarization in step 1. Once the positive charges accumulated enough at the emitter tip, the electrospray in step 2 happened. At the same time, to ensure the charge balance, electrochemical reaction should occur on the surface of rear part solution and gas to generate positive ions in the solution and negative ions in the gas phase. In step 4, the newly generated negative charges in the gas phase was neutralized by the gas electrode discharge. We had found some evidence to support this hypothesis. We measured the waveform of feedback voltage to electrode (Figure S8) and compared this with the waveform of pulsed spray signal measured at the emitter tip. It was apparent that the duration of one pulsed
Figure 3. LOD of nano-ESI and pulsed-dc-ESI. All the spectra showed here were from one scan event during MS analysis. For the nano-ESI, the sample (370 ± 30 pL) was only enough for one scan event (∼0.5 s). Since pulsed-dc-ESI could achieve better SNR, the LOD of pulseddc-ESI for somatostatin was 100 ppb compared to 1 ppm achieved by nano-ESI. Considering the average flow rate of pulsed-dc-ESI was about 380 pL/min (Table 1) and the spectrum of 100 ppb somatostatin analyzed by pulsed-dc-ESI was 0.51 s, we could estimate the absolute detection amount of pulsed-dc-ESI was about 200 zmol, compared to 240 amol achieved by nano-ESI.
Figure 4. Model for the ionization behavior of pulsed-dc-ESI (positive mode). The procedure to generate one pulsed spray had four steps: solution polarization, positive electrospray, electrochemical reaction in liquid−gas surface, and discharge between gas and electrode, namely, step 1 to step 4, respectively.
spray was exactly same as the duration one pulsed discharge between gas and electrode. Moreover, the frequency of pulsed spray and gas electrode discharge could also match. This results indicated the existence of step 2 to 4 in pulsed-dc-ESI. D
DOI: 10.1021/acs.analchem.5b02115 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
Figure 5. Profile of metabolites in single Allium cepa cell by pulsed-dc-ESI-MS. The elongated MS signal duration enabled a full scan−ddMS2 method for data acquisition. This method could record not only the full scan mass spectrum but also the MS2 spectra of the top 1034 components in the full scan of Allium cepa cell. (a) The massive MS2 data of the 1034 components and their accurate m/z values were used to build a 2-D mass spectrum. The full scan axis (y axis) indicated the m/z of the parent ions. At a certain m/z of full scan axis, the color sticks along the MS2 axis (x axis) showed the relative intensity of MS2 fragments of the corresponding m/z. (b) The comparison of components profiled by pulsed-dc-ESI and nanoESI. The upper part showed the projection of the color sticks in part a on the yz plane, which was achieved by pulsed-dc-ESI. One color sticks projection indicated we had obtained the MS2 and accurate m/z information on a component. The black sticks under the full scan axis showed the full scan mass spectrum of the Allium cepa cell, which was obtained by nano-ESI. It was obvious that the full scan spectrum could provide quite limited information since only 23 peaks in the spectrum were with relative intensity more than 5% (valid for analysis). However, the massive MS2 achieved by pulsed-dc-ESI could help to confirm the structures of the 1034 components, avoiding the analysis bias on components with a huge abundance difference in the cell. As a result, pulsed-dc-ESI made it possible to achieve abundant and valid data from single cell sample.
Figure 6. Using neutral loss search to identify all saccharides and their modified products in single cell. The typical neutral loss of saccharides and their modified products in MS/MS spectra was 162, which was the fragments of monosaccharide unit ([C6H12O6 − H2O]). After neutral loss search, we had screened out 505 components belonging to saccharides and their modified products. These components were then used for isotopic exclusion and accurate m/z matching in the database considered all forms of unmodified saccharides with m/z below 2000 (Table S3). Most of the components had a difference value (D-value) with the saccharides in the database. The D-value actually indicated the modified functional group. For example, the D-value of 14 indicated the methylation of saccharides (+ CH2). We finally determined 141 components, of which 17 components (12%) were oligosaccharides and 124 components (88%) were modified products of oligosaccharides in cell. We had figured out 6 modification types in cell, including alkylation, amination, acylation, oxidation, small molecule adducting, and metal ion exchange, which involved 28 chemical groups (Table S4).
(including exact m/z number and MS2 information) of metabolome (Figure 5 and Figure S10). The most outstanding advantage of the “big data” based 2-D mass spectrum was that it could give more information on low-abundance metabolites, even the metabolites submerged in the noise of full scan. For example, we could barely figure out the low-abundance metabolites components in the m/z range of 1000−2000 of Allium cepa cell since many of them were submerged in the noise. However, these components could be clearly visualized in the 2-D mass spectrum (Figure 5). With the abundant MS2 data, we could perform small omics scale analysis, such as database search, typical fragments search, and neutral loss search, to identify the single cell metabolome more accurately.
We used single cell sample as a typical model to evaluate the performance of pulsed-dc-ESI for real biological small volume sample. Figure 5a shows the profile of metabolites in an individual Allium cepa cell with pulsed-dc-ESI-MS and nanoESI-MS. By comparison, 1034 components were profiled by pulsed-dc-ESI while only 23 components were profiled by nano-ESI (Figure 5b). The greatly improved profiling capacity of pulsed-dc-ESI was evidently benefited from the boosting sample economy. Pulsed-dc-ESI could generate about 3 min of stable MS signal from single cell sample, compared with instantaneous MS signal duration achieved by nano-ESI17,18 or other ionization methods.33,34 The long signal duration was essential to the acquisition of “big data”, making it possible to collect abundant MS2 data and build the 2-D mass spectrum E
DOI: 10.1021/acs.analchem.5b02115 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry In Figure 5a, we could intuitively figure out some peaks distributed in specific regions. In the low m/z region of 2-D mass spectrum of Figure 5a, peaks were crowded and irregularly distributed. This region contained abundant information on different kinds of metabolites, which was especially suitable for database search analysis. For example, in the region (full scan, 136 to 210; MS2, 50 to 210; zoomed in Figure S11), we had identified 21 metabolites by matching the MS/MS data in mass bank database (Table S2). Otherwise, if simply identified by accurate m/z data from full scan, only 7 metabolites could be identified in this mass range in the reported work.34 In Figure 5a, we could also intuitively find some peak groups along the MS2 axis of 2-D mass spectrum. For instance, at the m/z of 185.04, 203.05, 347.10, and 365.11 on the MS2 axis, peaks were more crowded than other lines. These regular distributed peaks always indicated a certain kind of metabolites with unique neutral loss or typical fragment value. For example, many saccharides in Allium cepa cell were in modified forms. By searching the neutral loss of 162 (basic saccharide unit [C6H12O6 − H2O]), we could confirm 505 components containing saccharide structures from the 1034 components in a single cell (Figure 6a). We searched all 505 components in a database considering the combination of saccharide oligomerization and different ion adducts (Table S3). We found 141 components, of which 17 components (12%) were oligosaccharides and 124 components (88%) were modified products of oligosaccharides in cell (Figure 6b). Further identification figured out 6 modification types of saccharides in cell, including alkylation, amination, acylation, oxidation, small molecule adduct, and metal ion exchange (Figure 6b), involving 28 chemical groups (Table S4). To our knowledge, this is the first time to examine and identify hundreds of metabolites and their modified products in a single cell, which could provide useful information for deep understanding of single-cell metabolome. These achievements could be obtained not only from Allium cepa cell but also HeLa cell, a typical mammalian cell with a more complex matrix and smaller size35 (Figure S10). The MS/ MS data of 656 components were displayed in the 2-D mass spectrum by pulsed-dc-ESI-MS (Figure S10a,b). The peak groups in the 2-D mass spectrum enabled us to quickly find the typical fragments such as 69.07, 81.07, and 95.09 (Figure S10c). These typical fragments were actually the fragments of phospholipid structure. By searching 69.07, 81.07, and 95.09, we could extract 197 components of phospholipid from components for accurate m/z matching (Figure S10c). Therefore, the above results proved that pulsed-dc-ESI-MS was capable to obtain rich MS data from not only plant cell but also mammalian cell. In conclusion, we had developed pulsed-dc-ESI as an approach for systematical profiling of components in a small volume sample. By boosting sample economy, pulsed-dc-ESI could obtain several minutes long stable MS signal from picoliter sample without loss of MS sensitivity, making it possible to collect abundant data from a picoliter sample for the followed systematical analysis. No additional modification to a commercial MS instrument makes pulsed-dc-ESI a friendly method for wide users. Pulsed-dc-ESI had been showing outstanding performance for single cell metabolomics analysis. We had obtained 2-D profile of metabolites (including exact mass and MS2 data) from single plant and mamallian cell, concerning 1034 components and 656 components for Allium cepa and HeLa cell, respectively. The abundant data helped us to further identify 162 compounds and 28 different
modification groups of 141 saccharides in a single Allium cepa cell. To our knowledge, this was the first time to successfully perform direct systematical analysis for such a small volume sample. We believe that pulsed-dc-ESI could play an important role in systematic analysis of small volume sample in the near future.
■
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.analchem.5b02115. Details on experimental results and single cell metabolome identification (PDF)
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Present Addresses †
X.X. and X.F.: National Institute of Metrology, Beijing 100029, P. R. China. ‡ C.G. and F.T.: Department of Precision Instruments, Tsinghua University, Beijing 100084, P. R. China. § M.H. and W.X.: School of Life Science, Beijing Institute of Technology, Beijing 100081, P. R. China. Author Contributions
Z. Wei, S. Zhang, and X. Zhang conceived of Pico-ESI, applied Pico-ESI-MS for single-cell metabolomics analysis, and wrote the paper. Z. Wei designed and performed the experiments. X. Xiong and X. Fang built the database, worked out the twodimensional mass spectra, and wrote the programs for metabolomics analysis. Z. Wei, C. Guo, and F. Tang constructed the Pico-ESI-MS setup and measured the electrical parameters. X. Si and Y. Zhao helped with the cell culturing and single-cell manipulating. M. He and W. Xu simulated the electricity field of Pico-ESI and helped with the explanation of mechanism. C. Yang helped with the Orbitrap MS analysis. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS X. Zhang and Z. Wei thank the financial support provided by the 973 Program (Grant 2013CB933804) and the National Natural Science Foundation of China (Grant 21390410); S. Zhang, Y. Zhao, and X. Si thank the National Natural Science Foundation of China (Grant 21125525) and the Ministry of Science and Technology of China (Grant 2011YQ6008402). X. Fang, X. Xiong, and X. Zhang thank the Ministry of Science and Technology of China (Grant 2011YQ09000503).
■
REFERENCES
(1) Orchard, S.; Binz, P. A.; Jones, A. R.; Vizcaino, J. A.; Deutsch, E. W.; Hermjakob, H. Proteomics 2013, 13, 2931−2937. (2) Doroghazi, J. R.; Albright, J. C.; Goering, A. W.; Ju, K. S.; Haines, R. R.; Tchalukov, K. A.; Labeda, D. P.; Kelleher, N. L.; Metcalf, W. W. Nat. Chem. Biol. 2014, 10, 963−968. (3) Fiehn, O.; Kopka, J.; Dormann, P.; Altmann, T.; Trethewey, R. N.; Willmitzer, L. Nat. Biotechnol. 2000, 18, 1157−1161. (4) Aebersold, R.; Mann, M. Nature 2003, 422, 198−207. (5) Gstaiger, M.; Aebersold, R. Nat. Rev. Genet. 2009, 10, 617−627. (6) Sabido, E.; Selevsek, N.; Aebersold, R. Curr. Opin. Biotechnol. 2012, 23, 591−597.
F
DOI: 10.1021/acs.analchem.5b02115 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry (7) Irish, J. M.; Kotecha, N.; Nolan, G. P. Nat. Rev. Cancer 2006, 6, 146−155. (8) Newman, J. R. S.; Ghaemmaghami, S.; Ihmels, J.; Breslow, D. K.; Noble, M.; DeRisi, J. L.; Weissman, J. S. Nature 2006, 441, 840−846. (9) Shi, Q. H.; Qin, L. D.; Wei, W.; Geng, F.; Fan, R.; Shin, Y. S.; Guo, D. L.; Hood, L.; Mischel, P. S.; Heath, J. R. Proc. Natl. Acad. Sci. U. S. A. 2012, 109, 419−424. (10) Ibanez, A. J.; Fagerer, S. R.; Schmidt, A. M.; Urban, P. L.; Jefimovs, K.; Geiger, P.; Dechant, R.; Heinemann, M.; Zenobi, R. Proc. Natl. Acad. Sci. U. S. A. 2013, 110, 8790−8794. (11) Rubakhin, S. S.; Romanova, E. V.; Nemes, P.; Sweedler, J. V. Nat. Methods 2011, 8, S20−S29. (12) Zenobi, R. Science 2013, 342, 1243259. (13) Fuchs, E.; Tumbar, T.; Guasch, G. Cell 2004, 116, 769−778. (14) Pietras, A.; Katz, A. M.; Ekstrom, E. J.; Wee, B.; Halliday, J. J.; Pitter, K. L.; Werbeck, J. L.; Amankulor, N. M.; Huse, J. T.; Holland, E. C. Cell Stem Cell 2014, 14, 357−369. (15) Zhou, W. K.; Wei, L. R.; Xu, J. A.; Zhai, Q. Z.; Jiang, H. L.; Chen, R.; Chen, Q. A.; Sun, J. Q.; Chu, J. F.; Zhu, L. H.; Liu, C. M.; Li, C. Y. Plant Cell 2010, 22, 3692−3709. (16) Wei, Z. W.; Han, S.; Gong, X. Y.; Zhao, Y. Y.; Yang, C. D.; Zhang, S. C.; Zhang, X. R. Angew. Chem., Int. Ed. 2013, 52, 11025− 11028. (17) Date, S.; Mizuno, H.; Tsuyama, N.; Harada, T.; Masujima, T. Anal. Sci. 2012, 28, 201−203. (18) Tejedor, M. L.; Mizuno, H.; Tsuyama, N.; Harada, T.; Masujima, T. Anal. Chem. 2012, 84, 5221−5228. (19) Li, A. Y.; Hollerbach, A.; Luo, Q. J.; Cooks, R. G. Angew. Chem., Int. Ed. 2015, 54, 6893−6895. (20) Aerts, J. T.; Louis, K. R.; Crandall, S. R.; Govindaiah, G.; Cox, C. L.; Sweedler, J. V. Anal. Chem. 2014, 86, 3203−3208. (21) Rubakhin, S. S.; Lanni, E. J.; Sweedler, J. V. Curr. Opin. Biotechnol. 2013, 24, 95−104. (22) Berggren, W. T.; Westphall, M. S.; Smith, L. M. Anal. Chem. 2002, 74, 3443−3448. (23) Huang, G. M.; Li, G. T.; Ducan, J.; Ouyang, Z.; Cooks, R. G. Angew. Chem., Int. Ed. 2011, 50, 2503−2506. (24) Huang, G. M.; Li, G. T.; Cooks, R. G. Angew. Chem., Int. Ed. 2011, 50, 9907−9910. (25) Lu, Y.; Zhou, F.; Shui, W. Q.; Bian, L. P.; Yang, P. Y. Anal. Chem. 2001, 73, 4748−4753. (26) Wei, J. F.; Shui, W. Q.; Zhou, F.; Lu, Y.; Chen, K. K.; Xu, G. B.; Yang, P. Y. Mass Spectrom. Rev. 2002, 21, 148−162. (27) Sarver, S. A.; Chetwani, N.; Dovichi, N. J.; Go, D. B.; Gartner, C. A. J. Am. Soc. Mass Spectrom. 2014, 25, 524−529. (28) Qiao, L.; Sartor, R.; Gasilova, N.; Lu, Y.; Tobolkina, E.; Liu, B. H.; Girault, H. H. Anal. Chem. 2012, 84, 7422−7430. (29) Gasilova, N.; Yu, Q. L.; Qiao, L.; Girault, H. H. Angew. Chem., Int. Ed. 2014, 53, 4408−4412. (30) Bushey, J. M.; Kaplan, D. A.; Danell, R. M.; Glish, G. L. Instrum. Sci. Technol. 2009, 37, 257−273. (31) Blades, A. T.; Ikonomou, M. G.; Kebarle, P. Anal. Chem. 1991, 63, 2109−2114. (32) Vanberkel, G. J.; Zhou, F. M. Anal. Chem. 1995, 67, 3958−3964. (33) Amantonico, A.; Urban, P. L.; Fagerer, S. R.; Balabin, R. M.; Zenobi, R. Anal. Chem. 2010, 82, 7394−7400. (34) Stolee, J. A.; Shrestha, B.; Mengistu, G.; Vertes, A. Angew. Chem., Int. Ed. 2012, 51, 10386−10389. (35) Mizuno, H.; Tsuyama, N.; Harada, T.; Masujima, T. J. Mass Spectrom. 2008, 43, 1692−1700.
G
DOI: 10.1021/acs.analchem.5b02115 Anal. Chem. XXXX, XXX, XXX−XXX
