Native State Single-Cell Printing System and Analysis for Matrix

May 31, 2019 - LEARN ABOUT THESE METRICS ... Read OnlinePDF (3 MB) ... Department of Electrical & Computer Engineering, University of Illinois at ...
0 downloads 0 Views 3MB Size
Download PDF
Article Cite This: Anal. Chem. 2019, 91, 8115−8122

pubs.acs.org/ac

Native State Single-Cell Printing System and Analysis for Matrix Effects Qi Li,† Fei Tang,*,† Xinming Huo,† Xi Huang,‡ Yan Zhang,§ Xiaohao Wang,† and Xinrong Zhang∥

Downloaded via UNIV OF SOUTHERN INDIANA on July 17, 2019 at 06:30:16 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

†

State Key Laboratory of Precision Measurement Technology and Instruments, Department of Precision Instrument, Tsinghua University, Beijing 100084, China ‡ Beijing National Laboratory for Molecular Sciences, Key Laboratory of Analytical Chemistry for Living Biosystems, Institute of Chemistry, Chinese Academy of Sciences, Beijing 100190, China § Department of Electrical and Computer Engineering, University of Illinois at Urbana−Champaign, Urbana, Illinois 61801, United States ∥ Department of Chemistry, Tsinghua University, Beijing 100084, China S Supporting Information *

ABSTRACT: Mass spectrometry is subject to matrix effects, which causes severe limitations on the analysis of live single cells in their native state. Here, we propose a three-phase droplet-based single-cell printing analysis system (TPSCP), which can package, extract, separate, print, and analyze live single cells in saline matrixes (such as phosphate buffered saline) with matrix-assisted laser desorption/ionization mass spectrometry. This method can eliminate matrix effects to obtain information on a single cell in their native state. We report that a cell packaging percentage of 44% and single-cell packaging percentage of 88% can be achieved by TP-SCP. The system was capable of processing three to four single cells per second, which was 30 to 40 times higher than the traditional droplet-based microextraction (about 10 s/cell). Additionally, the MCF-7, A2780, 293, and 4T1 cells were screened in our system. The effect of cell viability and heterogeneity analysis was investigated, suggesting that the concentration of monounsaturated phosphatidylinositol and phosphatidylethanolamine both increase in cancer cells. Compared with conventional mass spectrometry, TP-SCP can ensure the accuracy of heterogeneity analysis of live single cells in their native state. Both a principal component analysis and a linear discriminant analysis were used to perform classification and identification of cells with an accuracy of 100%. This method provides an innovative framework for research on cell quality control, cell biology, cancer diagnosis, and prevention.

S

analysis with MALDI MS.11−13 They presented a protocol that combines MALDI-MS with immunocytochemistry to assay predominant lipids over thousands of individual rat brain cells.11 Vertes and co-workers conducted important single-cell and subcellular research with probe MS,14−17 using GeO2 fiber probes for various oligosaccharides in epidermal cells of onion and daffodil bulbs and lipids in sea urchin egg cells.18 Caprioli’s group used transmission geometry MALDI MS to realize the direct imaging of single cells at a subcellular spatial resolution.19−22 However, most MS-based single-cell research is performed in a nonsaline solution environment. Most cells maintain their activity only in isotonic conditions and a fixed pH range. A nonsaline environment will lead to deformation and rupture of cells, thus having a negative impact on the MS analysis of live single cells in their native state. This type of analysis is very

ingle-cell studies reveal cell information in an accurate and comprehensive manner, thus having great potential for the early diagnosis and prevention of disease. As such, these studies have become a pressing topic in cell biology research.1−5 Mass spectrometry (MS) technology, as an unlabeled analysis method, is capable of simultaneously detecting multiple components and even providing unknown molecular structure information. Accordingly, MS is promising for unlabeled multicomponent analysis in research on single cells; some previous work has assessed single-cell studies using MS. For instance, Do and colleagues analyzed the heterogeneity of single neurons with ionic liquid-assisted enhanced secondary ion MS.6 Additionally, Amantonico et al. studied the drug metabolism of unicellular organisms through matrixassisted laser desorption/ionization (MALDI) MS,7 and Nemes and co-workers conducted intensive research on capillary electrophoresis (CE) single-cell MS.8−10 The latter study identified 40 metabolites in three different cell types in distinct tissues with CE-MS, anchoring the interconnected central metabolic networks.10 Furthermore, Sweedler and coworkers made a notable contribution to the field of single-cell © 2019 American Chemical Society

Received: January 19, 2019 Accepted: May 31, 2019 Published: May 31, 2019 8115

DOI: 10.1021/acs.analchem.9b00344 Anal. Chem. 2019, 91, 8115−8122

Article

Analytical Chemistry

cells in the pbs, and the principal component analysis (PCA) and linear discriminant analysis (LDA) algorithm was used to perform classification and identification of cells. The TP-SCP system provides a feasible method for high throughput MS analysis of live single cells in saline matrixes.

important for many kinds of cell research, including quantitative analysis, substructure dynamic analysis, and decomposable functional group analysis.23,24 Phosphate buffered saline (pbs) is the preferred reagent for cell cultures and live cell analysis owing to its effective buffering of salts and pH. However, MS is very sensitive to micromolecule salt matrixes, especially phosphates, which make the target hard to detect with MS due to matrix effects.25−29 While MALDI-MS has an advantage in matrix tolerance, it is still impossible to directly analyze the cells in saline matrixes.30 Meanwhile, the conventional flushing desalination method can only remove salt in the extracellular solution, but intracellular salts cannot be removed, which negatively impacts the MS analysis.31,32 The interference then is more pronounced, especially when small-volume samples of single cells are detected. In that case, the MS peak will be completely covered by the noise, thus restricting the prospect of applying MS to live single-cell analysis. So far, there has been limited research regarding the MS of live single cells (supersmall samples; pL); in those studies, the saline interference matrixes were reported.33,34 Present research on desalination of interferential salt mostly focuses on small-volume samples (μL) such as liquid chromatography (LC),35 capillary electrophoresis (CE),36−38 solid phase microextraction (SPME),39,40 and droplet-based (solid− liquid) microextraction.33,34,41 However, liquid chromatography or capillary electrophoresis may lead to sample dilution during the elution.42 Though the risk of sample dilution can be reduced by certain procedures when solid phase microextraction is used for single-cell analysis, the operation is relatively tedious and complicated. For instance, with dropletbased (solid−liquid) microextraction, it is hard to realize a high throughput analysis of single cells (traditional high precise detection speed is about 10 s/cell) because manual intervention is needed for ensuring experiment precision. Moreover, it is still a major difficulty to accurately control the volume of the extraction phase. Therefore, how to analyze live single cells in saline matrixes and obtain reliable and high throughput information urgently requires a solution. This study proposes a three-phase droplet-based single-cell printing analysis system (TP-SCP), which can package, extract, separate, and print the cell substance of live single cells in saline matrixes (such as pbs) and analyze the cell substance with a MALDI-MS. This method can eliminate matrix effects to obtain single-cell information in the native state. In this system, three-phase droplet technology was first used to assess cell packaging. Then, octyl alcohol/acetonitrile solution was employed to assess the single-cell disruption and cellular substance extraction within the saline matrixes; chemical modification and the pressure balance method were then used to assess the highly effective three-phase separation among the extraction phase of single cells to be analyzed (the aqueous phase of cell residual liquid and the partition phase). Subsequently, a high-speed electro-motive stage was used to print the separated phase (extraction phase of single cells to be analyzed) on a microarray chip to ensure that each hole on the chip contained only the information on the single cells. Finally, MALDI MS was used for analysis. A three-phase droplet model was established as well in this study to provide a theoretical basis for packaging and separation of single cells. The TP-SCP system was applied to perform contrastive analysis on cell heterogeneity in different solutions. Then, the TP-SCP system was used to analyze the lipids in MCF-7, A2780, 293, and 4T1

■

EXPERIMENTAL SECTION Materials and Apparatus. SU-8 photoresist (2075) and polydimethylsiloxane (PDMS, RTV615) were purchased from Suzhou Wenhao Microfluidic Technology Co., Ltd. Polyethylene glycol (PEG; molecular weight 400 g/mol), octadecyltrichlorosilane (OTCS), N-(1-naphthyl)ethylenediamine dihydrochloride powder (nedc), and perfluorotripentylamine were purchased from Beijing J&K Scientific Ltd. Polystyrene fluorescent particles with a diameter of 19 μm were purchased from Tianjin BaseLine Chromtech Research Centre. Phosphatidic acid (PA (34:1)), phosphatidylglycerol (PG (34:1)), and phosphatidylserine (PS (34:1), PS (36:2)) were purchased from Avanti. Superhydrophobic coatings (NC310, NC316) were purchased from Changzhou Nanocoating Co., Ltd. Indium tin oxide (ITO) conductive glasses were purchased from Shenzhen Huanan Xiangcheng Technology Co., Ltd. Mineral oil was purchased from Beijing Solarbio Science & Technology Co., Ltd. Cell balancing liquid (Optiprep) was purchased from Sigma Limited Company). Absolute methanol (99.5%), isopropanol (99.7%), acetonitrile, absolute ethyl alcohol, and pbs were purchased from the platform of Tsinghua University. All reagents were used at their original concentrations without secondary purification. The spin coater (SUSS, Germany), stepper (MA6, SUSS, Germany), and drying machine (SAWA, Switzerland) used for preparing the microchip were all provided by the Institute of Semiconductors, Chinese Academy of Sciences. Pumps (PHD ULTRA Programmable) were purchased from Harvard Apparatus. A fluorescence microscope (Olympus lx73) was used to observe liquid flow. A superhydrophobic target plate (MTP Anchorchip384, Bruker Corp.) was used to test the effect of extracting agent. Electro-motive stages (SGSP26-50, SIGMA Corp.; purchased from Shanghai Boson Technology Co., Ltd. STM32F407) were used to control the movement of the stage. A Dino-Lite digital microscope (AM4815ZT, AnMo Electronics Corporation, Taiwan) was employed to observe the motion of the intracellular fluid drop of single cells. A spray pen was purchased from Shenzhen Crab Kingdom Technology Co., Ltd. MALDI-TOF MS (ultrafleXtreme, Bruker Corp.) and FTICR MS (solariX, Bruker Corp.) were both provided by the Institute of Chemistry, Chinese Academy of Sciences. Cell Culture and Preparation. MCF-7 cells were cultured in a 10 mL DMEM culture medium containing 10% fetal bovine serum (FBS), 100 U/mL penicillin, and 100 μg/mL streptomycin in a humid environment with 5% CO2 at a temperature of 37 °C. Cells were revaccinated once every 2 to 3 days. Before the experiment, trypsin-EDTA was used to treat cells to obtain the cell suspension solution. Cells were centrifuged and suspended again to gain the cell suspension solution with proper cell concentration. A cell suspension solution was divided equally and tested with a live/dead cell detection kit for cell viability. Cell suspension solution used for the experiment contained 98 ± 1% of live cells. The cell suspension solution was then equally divided into four parts and centrifuged at a speed of 1000 rpm. The upper suspension was removed, and pbs, 0.9% aqueous ammonium formate, 8116

DOI: 10.1021/acs.analchem.9b00344 Anal. Chem. 2019, 91, 8115−8122

Article

Analytical Chemistry

Figure 1. TP-SCP system and module design. (a) A schematic diagram of TP-SCP systems. L1, L2, and L3 are the inlets of the partition phase, the extraction phase, and the aqueous phase, respectively. The outlets of the phase to be analyzed and the cell residual liquid phase are O1 and O2, respectively. (b) A schematic diagram of lab-on-a-chip. Microchannels M1, M2, and M3 belong to the single-cell packaging zone; M4 belongs to the microextraction zone, and M5 and M6 belong to the separation zone. (c) A schematic diagram of the laminar flow interface. (d) Hydrophobic modification. (e) A schematic diagram of three-phase droplet separation. (f) A CCD photograph of three-phase droplet separation.

water, and 50% aqueous methanol solution were added successively to prepare the final cell suspension solution with approximately 3.375 × 106 cells/mL. A cell suspension solution of 4T1, 293, and A2780 cells was prepared with the same method. Microfluidic Chip Fabrication. Microfluidic chip molds were prepared by standard soft lithography (Section 1.1, Table S1). Molding technology was utilized to get a PDMS-based microchannel with a depth of 120 μm. The schematic diagram of the microchannel is shown in Figure 1b. Microchannels M1 and M3 were 30 μm wide, and the rest of the microchannels were 60 μm wide. The surfaces of microchannels M1, M2, M3, and M4 in the packaging and extraction zones were provided with a complete hydrophobic treatment. In the separation zone, however, channel M5 was hydrophobic and M6 was hydrophilic through a two-step chemical modification. A schematic diagram of the modification process is shown in Figure 1c,d, and a schematic effect diagram and real effect diagram postmodification are shown in Figure 1e,f, respectively (modification methods are shown in Sections 1.2 and 1.3, Figure S1). Microarray Chip Fabrication. A spin-coating method was used to apply a superhydrophobic coating on the ITO substrate; then, laser processing was applied to remove the coating of partial areas to gain the nonsuperhydrophobic round array. This method was employed to gain a 20 × 20 round array with a diameter of 300 μm and center to center spacing of 600 μm. After the printed analyte was dried, it was dissolved by spraying the nedc methanol−water matrix (1:1, v/v) with a spray pen. After recrystallization, MS was performed. The nedc matrix concentration was 5 mg/mL (fabrication and test are shown in Section 2, Figure S2). System Setup. The TP-SCP system can be divided into three modules: sample injection module, lab-on-a-chip, and

printing module for the substances to be analyzed (Figure 1a; additional image in Figure S3). The microfluidic chip was fixed by a printing module. The microarray chip was placed on electro-motive stages directly below the printing module. The moving speed of the stage was controlled by adjusting the pulse frequency of the drive. Here, the flow rate of the aqueous phase, extract phase, and partition phase were Qw = 0.3 μL/ min, Qe = 0.6 μL/min, and Qp = 1 μL/min, while the controller pulse frequency was 2 kHz. Under these conditions, the difference in hydrophobicity of the microarray chip ensures that the phase droplets enter the pores of the microarray chip upon analysis.43 MS Analysis. A Bruker 9.4 T solarix FTICR MS equipped with an Apollo dual-mode electrospray ionization (ESI)/ MALDI ion source with a 355 nm and 200 Hz solid-state Smartbeam Nd:YAG UV laser (Bruker Daltonics, Bremen, Germany) was employed for recording accurate masses from single cells. Mass spectra were acquired over the mass range from 100 to 1000 Da in negative ion mode. For MALDI-MS profiling, the mass spectra were recorded by accumulating 20 times at 200 laser shots per scan. Instrument calibration used a standard calibration procedure. The sodium trifluoroacetate solution was selected as the calibration solution. Calibration error is less than 2 ppm. Data Analysis. Background subtraction was conducted by extracting and eliminating all peaks that appeared in all blank experiments. Both a principal component analysis and a linear discriminant analysis were obtained to classify and visualize the single-cell data. The identification of metabolites was achieved through matching the accurate mass spectra with the standards in the METLIN (http://metlin.scripps.edu/) and HMDB (http://www.hmdb.ca/) databases. The data were statistically tested with an unpaired t test. Values of P < 0.05 were 8117

DOI: 10.1021/acs.analchem.9b00344 Anal. Chem. 2019, 91, 8115−8122

Article

Analytical Chemistry considered significantly different, while P > 0.05 was considered to not be significantly different.

(Section 5.2). Consequently, the single-cell packaging rate is about 88% of the cell packaging rate. According to the calculations, experimental results, and previous literature, the printing of the single-cell substance to be analyzed has a rate of 3 to 4 single cells per second. This rate is achieved through adjusting the translational speed of the stage to enable the droplets to be entrapped into each hydrophilic hole on the base plate in sequence. System Stability. A series of standard substances with the same concentration (PA (34:1), PG (34:1), and PS (34:1)) were used to test the stability of the TP-SCP system under the same experimental conditions. To eliminate the error in absolute strength caused by the difference in crystalline state of the samples to be analyzed (Section 5.3, Figure S11), two groups of mixed solution, prepared with standard substances PAs/PGs and PGs/PSs, were selected to measure the strength ratio of the two samples in the mixed solution and evaluate the stability of the system accordingly. The results of the experiment are shown in Figure 3. For the mixed solution of

■

RESULTS AND DISCUSSION Overview of the TP-SCP System. The lab-on-a-chip mainly consists of the single-cell packaging zone, the microextraction zone, and the separation zone. The threedimensional diagram of the microchannel is shown in Figure 1a. Microchannels M1 and M3 employ an S-shaped structural design to reduce the disturbance of the injection system and ensure the stability of the sample injection. Microchannel M4 in the microextraction zone also employs an S-shaped structural design to increase the length of the channel. This zone increases the effective time of extraction and accelerates the surface flow velocity of liquid at the bends of the channel for improving extraction efficiency. A three-phase droplet estimation model and a separation model were established to achieve precise regulation and stable separation of three-phase droplets in a microfluidic system (Sections S3 and S4: Figures S4−S9, Tables S2−S5). In this case, an octyl alcohol/ acetonitrile solution (7:3, v/v) was used, while acetonitrile was used to dissolve cell membranes and octanol was used to extract water-insoluble substances in cells (see Section 5.1 for details). Single-Cell Packaging. In the experiment, the flow velocities of the three phases were Qw = 0.3 μL/min, Qe = 0.6 μL/min, and Qp = 1 μL/min, respectively. The lengths of the outlet capillary tubes were Lext,tube = 50 mm and Laqu,tube = 78 mm, respectively. The packaging effect of the polystyrene fluorescent particles is shown in Figure S10. The same experiment was repeated with the same parameters by changing the fluorescent particles to MCF-7 cells and adjusting the density of the solution to a value equal to that of the cells (cell concentration is about 3.375× 106 cells/mL) with Optiprep. During the experiment, no magnetic stirring was conducted because intense mechanical movement may rupture the cell membrane,44 even causing the death of cells. The experimental result of encapsulation of a single cell into a droplet is shown in Figure 2a. Since sample injection is a

Figure 3. Analysis of the stability of TP-SCP system. (a) The MS spectra of the mixed solution of PA (34:1) and PG (34:1); RSD is 3%. (b) The MS spectra of the mixed solution of PG (34:1) and PS (34:1); RSD is 4.2%.

PAs/PGs and PGs/PSs, the values of RSD are 3% and 4.2%, respectively, indicating that the TP-SCP system shows a good stability (RSD < 5%, calculated by a selection of 22 sets of experimental data). The inset figure shows the distribution of changes in strength ratio of lipid standard substances in the 22 test points. A two-tailed Student’s t test was performed to assess the reproducibility of data (Figure S12); the results indicate no significant difference of the ratio between these two groups (P > 0.05). The standard curve of the concentration ratio of phospholipid standard samples was measured by TP-SCP. The PG (34:1) and PS (36:2) values common to cells used in the experiment were selected, and the concentration of PS (36:2) was 1 μM. The concentration of PG (34:1) was 1−5 μM. The R2 was 0.988 (Figure S13). Extracting Agent Analysis. The mixture of octyl alcohol and acetonitrile (7:3) was selected to be the cell lysis/ extracting agent. This mixed solution can evaporate quickly without imposing any influence on the substances to be analyzed. The analysis of an nedc aqueous methanol solution, solution A, solution B, and solution C was conducted with MTP Anchorchip384 target plates (results are illustrated in Figure 4a; preparation steps are listed in Section 5.4). The orange rectangular box corresponds to the spectra within m/z = 700−900. The nedc aqueous methanol solution (blue) does not show obvious characteristic peaks, indicating that it will not cause any interference to the analysis that results during the cell lipid analysis, and it is thus a good matrix solution for lipid analysis. Solution A (green, multicell suspension solution)

Figure 2. Cell distribution of the optimized single-cell packaging. (a) Image of the single-cell packaging results with the red rectangular box indicating a single-cell package unit. (b) Distribution of different cell numbers per droplet (blue bar) and a Poisson fit (red line). Single-cell packaging rate is 39%, and the rate of packaging two or more cells in a droplet is about 5%.

random process, the rate of packaging single cells into droplets can be estimated with a Poisson distribution P(X = x) = e−λ(λx/x!). As such, the number of cells packaged per droplet were counted (Figure 2b). Furthermore, optimizing the experimental conditions achieved a total cell packaging rate of 44%, of which the single-cell packaging rate is 39% and the rate of packaging two or more cells in a droplet is about 5% 8118

DOI: 10.1021/acs.analchem.9b00344 Anal. Chem. 2019, 91, 8115−8122

Article

Analytical Chemistry

Figure 4. Results of MALDI MS analysis. (a) MS spectra in different solutions: blue for matrix solution, green for solution A, purple for solution B, and red for solution C. (b) Single-cell lipid analysis of the MCF-7 cell pbs suspension solution with the TP-SCP system (single droplet packaging single cells). (c) Lipid analysis of droplets without cells packaged with the TP-SCP system (single droplet packaging zero cells). (d) Cell culture medium in the control group with the TP-SCP system. (e) pbs in the control group with the TP-SCP system. (f) Results of cell lipid analysis of the MCF-7 single-cell level pbs solution with MALDI MS.

S15c,d). The characteristic phospholipid of the cell in the onsaline solution represents 77% of the multicell analysis result, while only 50% of the characteristic signal can be detected in the pbs solvent. The result of the 4T1 single cell is shown in Figure S17. The number of phospholipid signals gained in onsaline solution conditions are about 1.7 times of those gained in pbs conditions, which is basically consistent with the analysis result of MCF-7 single cells. The reason may be that cells are vulnerable to swelling, fracture, and rapture in onsaline solution conditions, causing easy interference and stacking of different single cells; the heterogeneity of single cells is thus averaged out of the sample. In pbs conditions, however, cells can effectively maintain their activity and preferably exhibit the heterogeneity of single cells without interference and stacking due to swelling fracture and rapture. Single-Cell MS Analysis. MCF-7 cells were used to test the performance of TP-SCP in a single-cell analysis. In this experiment, the solution (for the cell culture medium and pbs) was selected as the control group to exclude the impact of interference signal on the analysis result. The cell pbs suspension solution (experimental group) was changed to the cell culture medium and pbs in sequence, with the other conditions unchanged, and TP-SCP was used to perform the control experiment. The results of the single-cell analysis with TP-SCP are shown in Figure 4b. This analysis showed obvious phospholipid peaks; empty droplets without cells packaged were analyzed with TP-SCP, and no phospholipid peaks were detected (Figure 4c). The solution of the control group was analyzed with TP-SCP, and there were no detected effective

shows an obvious lipid characteristic. Solution B (pink, extracted extraction liquid) shows peaks in the same position as the pattern of solution A. Solution C (red, extracted cell residual liquid) does not show any similar characteristic with that of solution A. These results imply that this latter extracting agent can separate the cell lipid information from pbs. Discussion on Common Solution for Single-Cell Analysis. The pbs reagent provides an ideal environment for cell survival and is thus preferred for biological assays. In this study, the impact of pbs and other common solutions for MS (such as 0.9% aqueous ammonium formate, water, and an aqueous methanol solution) was discussed. As shown in Section 5.5 and Figure S14, the results reveal that pbs can maintain the osmotic pressure of the cells and buffer the environment for a long time, thus representing the best environment for cell survival. These results ensure that the cell structure has not changed and that cell activity can be maintained. To study the difference in phospholipid analysis of cells in different active states, the comparison was conducted by selecting the common solution (water, on-saline solution) and pbs for MS analysis. A multicell analysis is provided in Section 5.6. The results show that both solutions can detect the characteristic phospholipid of the cell and that the main characteristic peaks are essentially the same (Figure S15a,b). The results additionally imply that cell activity does not have a clear impact on the multicell analysis (Figure S16). Then, TPSCP was used to analyze the phospholipid content of single cells of MCF-7 in both an on-saline solution and pbs (Figure 8119

DOI: 10.1021/acs.analchem.9b00344 Anal. Chem. 2019, 91, 8115−8122

Article

Analytical Chemistry

(34:1)/PS (36:2) in four kinds of cells were obtained (Figure S21). From the results of the t test, there was a significant distinction in the PG/PS ratio between the four kinds of cells. Moreover, differences in concentration ratios between cells of the same species can also be found (P < 0.001), possibly due to cellular heterogeneity. Although the cell types were different, PAs, PEs, PSs, PGs, and PIs were widely widespread in four kinds of cells (Figure 6a). Figure 6b shows the distribution of phospholipid unsaturation in MCF-7 and 4T1 cells. Here PI0, PI1, and PI2 represent saturated, monounsaturated, and polyunsaturated lipids, respectively. The representation principle of PAs, PEs, PGs, and PSs is the same. There was a significant difference in saturated and unsaturated phospholipids. In our observation, the ratio of saturated PAs and PGs in MCF-7 was higher, while the ratio of saturated PEs and PSs in 4T1 was higher; no saturated PIs were detected in either cell type. In MCF-7 cells, the monounsaturated PIs and PEs increased with PI (34:1), PI (36:1), PE (34:1), and PE (36:1). However, the monounsaturated PGs and PSs in 4T1 increased, and there was no significant distinction in monounsaturated PAs. Both cells contained a large number of polyunsaturated phospholipids. The MCF-7 cell contained more polyunsaturated PGs and PSs, while other phospholipids in 4T1 were more abundant. To investigate the origin of the variations, we characterized the correlation of the measured mass spectral peaks with the principal component in the loading plot (Figure 6c). The first five principal components were extracted to obtain more than 50% of the component information. A high loading magnitude indicated a great contribution to the five principal components. Among them, m/z = 748.5140 (PS (33:0)) and m/z = 852.4815 (PS (42:11)) had a higher contribution, manifesting that the PSs may be the main cause of four kinds of cell differences. The difference in PS values between MCF-7, A2780, and 293 is shown in Figure 6d. There was a significant difference in the saturation of PS values among the three cells. The A2780 cells had the highest saturated and monounsaturated PSs, while 293 cells contained a moderate number of these components and MCF-7 cells contained the lowest amount. The differences between the three kinds of cells were significant (P < 0.001), as the polyunsaturated lipid content in MCF-7 cells was higher than in the other cell types. Mitochondria regulate cell energy supply.45 Studies show that, in the mitochondria of eukaryotes, PG is mainly distributed in the structure of the inner and outer membranes. As such, the amount of PG will affect mitochondrial morphology and thus normal physiological functions.46−48 The abnormal energy supply of cancer cells under oxygen-rich conditions reveals serious functional disorders in the mitochondria. Therefore, there is a certain correlation between PG content and cell carcinogenesis. Figure 6e shows the difference in PG content between cells. Polyunsaturated PGs in A2780 and MCF-7 cells significantly increased, and the PG (38:5) and PG (40:5) held the highest abundances. In sum, the less saturated the phospholipid, the stronger is the substance exchange capacity of the corresponding cell membrane. This result is consistent with the abnormal metabolism of cancer cells. PEs play an important role in the cell division,49 especially during cytokinesis. Past literature reports that monounsaturated PEs show an increasing trend in 4T1, such as PE (O34:1) and PE (O-36:1).50 The same trend had also been found

phospholipid peaks (Figure 4d,e). On the basis of these findings, we conclude that the result of single-cell analysis with TP-SCP is indeed the lipid of cells. Accurate molecular weights were determined with FTICR MS to assess the molecular composition (Table S6). To verify the sensitivity of TP-SCP in analyzing single cells in saline matrixes, a traditional analysis method was used to assess the single-cell level in a pbs solution.12 The MS peaks are indistinguishable from the noise (Figure 4f). This finding is because single cells inherently contain very few substances and saline interference matrixes (pbs) may form crystalline complexes on the surface of the analyte, thus complicating the signal analysis. The comparison between Figure 4b,f shows that TP-SCP can accurately analyze liposoluble substances of single cells in interference matrixes to gain single-cell information that cannot be detected due to matrix effects (the result is the same as that for the solvent PG (34:1); Section 5.7 and Figure S18). Identification of Cell Subset. This study also assessed cell classification. Cell suspensions of three kinds of cells (4T1, 293, and A2780) in pbs solutions were selected (Figure 5a−c).

Figure 5. Results of detection, analysis, and classification of the different types of single cells in pbs with TP-SCP. (a) MS spectra of 4T1 single cells. (b) MS spectra of 293 single cells. (c) MS spectra of A2780 single cells. (d) Classification of the four types of cells (MCF7, 4T1, 293, and A2780) with the PCA and LDA algorithms.

The signal-to-noise ratio of partial main phospholipids can be 2 orders of magnitude higher (885.5518, 744.5566, 863.5675, and 887.5668). The information on the composition of phospholipids is shown in Table S6. Both PCA and LDA were used to effectively recognize the type of single cells, according to the MS information. Twenty sets of data for each type of cell (a total of 80 sets of data) were selected (A2780 single-cell analysis with TP-SCP; Figure S19). The 4T1, 293, A2780, and MCF-7 cells were successfully classified as separate populations (Figure 5d). Another ten sets of data for each cell type (a total of 40 sets of data) were randomly mixed and classified using the same PCA and LDA function to verify the accuracy of the classification. The data of the test group can be precisely assigned to the corresponding cell zone with an accuracy rate of 100% (Figure 5d, Section S6, Figure S20). Single-Cell Phospholipid Differential Analysis. After calculation by standard curves, the concentration ratios of PG 8120

DOI: 10.1021/acs.analchem.9b00344 Anal. Chem. 2019, 91, 8115−8122

Article

Analytical Chemistry

Figure 6. Single-cell phospholipid differential analysis. (a) Four kinds of single-cell glycerophospholipid distributions. (b) Differences in phospholipids between MCF-7 and 4T1 cells. (c) Four kinds of cell principal component load maps, in which five principal components were selected for calculation. (d) Phosphatidylserine (PS) differences. (e) Phosphatidylglycerol (PG) differences. (f) Phosphatidylinositol (PI) differences. (g) Phosphatidylethanolamine (PE) differences. *** denotes P < 0.001; n.s means no significant difference.

show that TP-SCP can better ensure the accuracy of single-cell heterogeneity analysis. Subsequently, the system was used to analyze the lipid-based substances in MCF-7, A2780, 293, and 4T1 cells. The PCA and LDA algorithm was used to classify the cells to determine the cell type with an accuracy of 100%. Furthermore, the lipid analysis results show that phosphatidylinositol and phosphatidylethanolamine increase in cancer cells. With negligible user intervention, the system provides a feasible resolution for live single-cell analysis in a pbs solution. The TP-SCP exhibits a potential application prospect in cell quality control and cell biology as well as cancer diagnosis and prevention.

in PIs involved in cell signal regulation and vesicle formation, such as PI (36:1).51 The content of monounsaturated PEs and PIs in A2780 and MCF-7 cells also increases (Figure 6f,g), which may be due to the fact that up-regulation of monounsaturated lipids can provide components for the rapid proliferation of cancer cells, resulting in denser biofilms and changing the fluidity of the membrane to promote cancer.

■

CONCLUSIONS In this study, a TP-SCP system was proposed to analyze the heterogeneity of live single cells in their native state and identify the cell population in saline matrixes. The TP-SCP system employed a three-phase flow technology to achieve the single-cell packaging. A three-phase droplet model was established with the consistency between theoretical and experimental results reaching 0.99, providing a theoretical basis for packaging and subsequent separation of the droplet. This system was able to realize a cell packaging rate of 44% and a single-cell packaging rate of 88%. An octyl alcohol/acetonitrile solution was used to achieve the single-cell disruption and cellular substance extraction. Efficient three-phase separation was conducted by means of adjusting the hydrophilicity/ hydrophobicity of the separation zone and controlling the length of the capillary tube. The TP-SCP was capable of processing three to four single-cell substances per second, which is 30 to 40 times higher than that of a traditional droplet-based microextraction method (about 10 s/cell). This TP-SCP approach also features excellent repeatability (RSD = 3%). The experimental results of a single-cell MS analysis in nonsaline and saline solution environments were compared to

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.analchem.9b00344.

■

TP-SCP system production and test details; model algorithms; data processing methods (PDF)

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Fei Tang: 0000-0001-8155-1178 Notes

The authors declare no competing financial interest. 8121

DOI: 10.1021/acs.analchem.9b00344 Anal. Chem. 2019, 91, 8115−8122

Article

Analytical Chemistry

■

(30) Xu, Y.; Bruening, M. L.; Watson, J. T. Mass Spectrom. Rev. 2003, 22 (6), 429−440. (31) Börnsen, K. O. Methods Mol. Biol. 2000, 146, 387−404. (32) Garden, R. W.; Moroz, L. L.; Moroz, T. P.; Shippy, S. A.; Sweedler, J. V. J. Mass Spectrom. 1996, 31 (10), 1126−1130. (33) Sirikatitham, A.; Yamamoto, T.; Shimizu, M.; Hasegawa, T.; Tsuyama, N.; Masujima, T. Rapid Commun. Mass Spectrom. 2007, 21 (3), 385−390. (34) Zhu, H.; Wang, N.; Yao, L.; Chen, Q.; Zhang, R.; Qian, J.; Hou, Y.; Guo, W.; Fan, S.; Liu, S.; Zhao, Q.; Du, F.; Zuo, X.; Guo, Y.; Xu, Y.; Li, J.; Xue, T.; Zhong, K.; Song, X.; Huang, G.; Xiong, W. Cell 2018, 173 (7), 1716−1727. (35) Fornea, D. S.; Wu, Y.; Marcus, R. K. Anal. Chem. 2006, 78 (15), 5617−5621. (36) Burgi, D. S.; Chien, R. L. Anal. Chem. 1991, 63 (18), 2042− 2047. (37) Johnson, R. D.; Navratil, M.; Poe, B. G.; Xiong, G.; Olson, K. J.; Ahmadzadeh, H.; Andreyev, D.; Duffy, C. F.; Arriaga, E. A. Anal. Bioanal. Chem. 2006, 387 (1), 107−118. (38) Oita, I.; Halewyck, H.; Pieters, S.; Thys, B.; Vander Heyden, Y.; Rombaut, B. J. Virol. Methods 2012, 185 (1), 7−17. (39) Dahlin, A. P.; Bergström, S. K.; Andrén, P. E.; Markides, K. E.; Bergquist, J. Anal. Chem. 2005, 77 (16), 5356−5363. (40) Morales-Cid, G.; Gebefugi, I.; Kanawati, B.; Harir, M.; Hertkorn, N.; Rosselló-Mora, R.; Schmitt-Kopplin, P. Anal. Bioanal. Chem. 2009, 395 (3), 797−807. (41) Zhang, X.-C.; Wei, Z.-W.; Gong, X.-Y.; Si, X.-Y.; Zhao, Y.-Y.; Yang, C.-D.; Zhang, S.-C.; Zhang, X.-R. Sci. Rep. 2016, 6, 24730. (42) Wei, Z.; Han, S.; Gong, X.; Zhao, Y.; Yang, C.; Zhang, S.; Zhang, X. Angew. Chem., Int. Ed. 2013, 52 (42), 11025−11028. (43) Küster, S. K.; Pabst, M.; Jefimovs, K.; Zenobi, R.; Dittrich, P. S. Anal. Chem. 2014, 86 (10), 4848−4855. (44) Mazutis, L.; Gilbert, J.; Ung, W. L.; Weitz, D. A.; Griffiths, A. D.; Heyman, J. A. Nat. Protoc. 2013, 8 (5), 870−891. (45) Pokorna, L.; Cermakova, P.; Horvath, A.; Baile, M. G.; Claypool, S. M.; Griac, P.; Malinsky, J.; Balazova, M. Biochim. Biophys. Acta, Bioenerg. 2016, 1857 (1), 34−45. (46) Chang, S. C.; Heacock, P. N.; Mileykovskaya, E.; Voelker, D. R.; Dowhan, W. J. Biol. Chem. 1998, 273 (24), 14933−14941. (47) Tuller, G.; Hrastnik, C.; Achleitner, G.; Schiefthaler, U.; Klein, F.; Daum, G. FEBS Lett. 1998, 421 (1), 15−18. (48) Jiang, F.; Rizavi, H. S.; Greenberg, M. L. Mol. Microbiol. 1997, 26 (3), 481−491. (49) Gibellini, F.; Smith, T. K. IUBMB Life 2010, 62 (6), 414−428. (50) Doria, M. L.; Cotrim, Z.; Macedo, B.; Simoes, C.; Domingues, P.; Helguero, L.; Domingues, M. R. Breast Cancer Res. Treat. 2012, 133 (2), 635−648. (51) Doria, M. L.; Cotrim, C. Z.; Simoes, C.; Macedo, B.; Domingues, P.; Domingues, M. R.; Helguero, L. A. J. Cell. Physiol. 2013, 228 (2), 457−468.

ACKNOWLEDGMENTS This study was financially supported by the National Natural Science Foundation of China (51575312) and the Fundamental Research Project of Shenzhen (JCYJ20150601165744636).

■

REFERENCES

(1) Mizuno, H.; Tsuyama, N.; Harada, T.; Masujima, T. J. Mass Spectrom. 2008, 43 (12), 1692−1700. (2) Mizuno, H.; Tsuyama, N.; Date, S.; Harada, T.; MASUJIMA, T. Anal. Sci. 2008, 24 (12), 1525−1527. (3) Masujima, T. Anal. Sci. 2009, 25 (8), 953−960. (4) Gholipour, Y.; Erra-Balsells, R.; Hiraoka, K.; Nonami, H. Anal. Biochem. 2013, 433 (1), 70−78. (5) Gong, X.; Zhao, Y.; Cai, S.; Fu, S.; Yang, C.; Zhang, S.; Zhang, X. Anal. Chem. 2014, 86 (8), 3809−3816. (6) Do, T. D.; Comi, T. J.; Dunham, S. J. B.; Rubakhin, S. S.; Sweedler, J. V. Anal. Chem. 2017, 89 (5), 3078−3086. (7) Amantonico, A.; Urban, P. L.; Fagerer, S. R.; Balabin, R. M.; Zenobi, R. Anal. Chem. 2010, 82 (17), 7394−7400. (8) Nemes, P.; Knolhoff, A. M.; Rubakhin, S. S.; Sweedler, J. V. ACS Chem. Neurosci. 2012, 3 (10), 782−792. (9) Onjiko, R. M.; Portero, E. P.; Moody, S. A.; Nemes, P. Anal. Chem. 2017, 89 (13), 7069−7076. (10) Onjiko, R. M.; Moody, S. A.; Nemes, P. Proc. Natl. Acad. Sci. U. S. A. 2015, 112 (21), 6545−6550. (11) Neumann, E. K.; Comi, T. J.; Rubakhin, S. S.; Sweedler, J. Angew. Chem., Int. Ed. 2019, 58 (18), 5910−5914. (12) Si, T.; Li, B.; Comi, T. J.; Wu, Y.; Hu, P.; Wu, Y.; Min, Y.; Mitchell, D. A.; Zhao, H.; Sweedler, J. V. J. Am. Chem. Soc. 2017, 139 (36), 12466−12473. (13) Rubakhin, S. S.; Romanova, E. V.; Nemes, P.; Sweedler, J. V. Nat. Methods 2011, 8 (S4), S20−S29. (14) Shrestha, B.; Vertes, A. Anal. Chem. 2014, 86 (9), 4308−4315. (15) Stolee, J. A.; Vertes, A. Anal. Chem. 2013, 85 (7), 3592−3598. (16) Vaikkinen, A.; Shrestha, B.; Nazarian, J.; Kostiainen, R.; Vertes, A.; Kauppila, T. J. Anal. Chem. 2013, 85 (1), 177−184. (17) Nemes, P.; Woods, A. S.; Vertes, A. Anal. Chem. 2010, 82 (3), 982−988. (18) Shrestha, B.; Vertes, A. Anal. Chem. 2009, 81 (20), 8265−8271. (19) Caprioli, R. M.; Farmer, T. B.; Gile, J. Anal. Chem. 1997, 69 (23), 4751−4760. (20) Zavalin, A.; et al. J. Mass Spectrom. 2012, 47 (11), 1473−1481. (21) Puolitaival, S. M.; Burnum, K. E.; Cornett, D. S.; Caprioli, R. M. J. Am. Soc. Mass Spectrom. 2008, 19 (6), 882−886. (22) Spraggins, J. M.; Rizzo, D. G.; Moore, J. L.; Noto, M. J.; Skaar, E. P.; Caprioli, R. M. Proteomics 2016, 16 (11−12), 1678−1689. (23) Liu, T. L.; Upadhyayula, S.; Milkie, D. E.; Singh, V.; Wang, K.; Swinburne, I. A.; Mosaliganti, K. R.; Collins, Z. M.; Hiscock, T. W.; Shea, J.; Kohrman, A. Q.; Medwig, T. N.; Dambournet, D.; Forster, R.; Cunniff, B.; Ruan, Y.; Yashiro, H.; Scholpp, S.; Meyerowitz, E. M.; Hockemeyer, D.; Drubin, D. G.; Martin, B. L.; Matus, D. Q.; Koyama, M.; Megason, S. G.; Kirchhausen, T.; Betzig, E. Science 2018, 360 (6386), eaaq1392. (24) Demling, P.; Westerwalbesloh, C.; Noack, S.; Wiechert, W.; Kohlheyer, D. Curr. Opin. Biotechnol. 2018, 54, 121−127. (25) Dams, R.; Huestis, M. A.; Lambert, W. E.; Murphy, C. M. J. Am. Soc. Mass Spectrom. 2003, 14 (11), 1290−1294. (26) McCombie, G.; Knochenmuss, R. Anal. Chem. 2004, 76 (17), 4990−4997. (27) Gusev, A. I.; Wilkinson, W. R.; Proctor, A.; Hercules, D. M. Anal. Chem. 1995, 67 (6), 1034−1041. (28) Konry, T.; Dominguez-Villar, M.; Baecher-Allan, C.; Hafler, D. A.; Yarmush, M. L. Biosens. Bioelectron. 2011, 26 (5), 2707−2710. (29) Van Eeckhaut, A.; Lanckmans, K.; Sarre, S.; Smolders, I.; Michotte, Y. J. Chromatogr. B: Anal. Technol. Biomed. Life Sci. 2009, 877 (23), 2198−2207. 8122

DOI: 10.1021/acs.analchem.9b00344 Anal. Chem. 2019, 91, 8115−8122