Concentric Hybrid Nanoelectrospray Ionization‐Atmospheric Pressure Chemical Ionization Source for High‐Coverage Mass Spectrometry Analysis of Single‐Cell Metabolomics

Abstract High‐coverage mass spectrometry analysis of single‐cell metabolomics remains challenging due to the extremely low abundance and wide polarity of metabolites and ultra‐small volume in single cells. Herein, a novel concentric hybrid ionization source, nanoelectrospray ionization‐atmospheric pressure chemical ionization (nanoESI‐APCI), is ingeniously designed to detect polar and nonpolar metabolites simultaneously in single cells. The source is constructed by inserting a pulled glass capillary coaxially into a glass tube that acts as a dielectric barrier layer. Benefitting from the integrated advantages of nanoESI and APCI, its limit of detection is improved by one order of magnitude to 10 pg mL−1. After the operational parameter optimization, 254 metabolites detected in nanoESI‐APCI are tentatively identified from a single cell, and 82 more than those in nanoESI. The developed nanoESI‐APCI is successively applied to study the metabolic heterogeneity of human hepatocellular carcinoma tissue microenvironment united with laser capture microdissection (LCM), the discrimination of cancer cell types and subtypes, the metabolic perturbations to glucose starvation in MCF7 cells and the metabolic regulation of cancer stem cells. These results demonstrated that the nanoESI‐APCI not only opens a new avenue for high‐coverage and high‐sensitivity metabolomics analysis of single cell, but also facilitates spatially resolved metabolomics study coupled with LCM.


Table of Contents
The possible reaction mechanism involved in the ionization process . Effect of plasma gases (N2, air) on the intensity of 9 classes of model compounds with different . Effect of velocity of N2 (0.1 L/min, 0.2 L/min, 0.5 L/min 1.0 L/min) on the intensity of 9 classes of model compounds with different polarities.
. The laser 'cut and drop' sampling process.
. The outcome of t-SNE cluster analysis of metabolites with statistically significant differences (p < 0.05) between 24 CSCs and 18 NSCCs.
Supporting Tables: Table S1.Detailed information of 10 classes of model compounds.
Table S3.Assigned metabolites from a single cell in nanoESI and nanoESI-APCI modes.
Table S4.Performance comparison between related studies and ours.

Cell culture and sampling
According to the optimum culture protocol recommended by ATCC, all cell lines were cultured in a 10 cm diameter dish containing 10% FBS, 1% penicillin-streptomycin solution and DMEM, and placed in an incubator at 37 °C and 5% CO2.For the glucose starvation experiment of MCF7 cells, after removing the DMEM medium, the cells were washed 3 times by PBS, and the cell medium was changed to glucosefree DMEM supplemented with 10% FBS and cultured for 24 h.Cells were washed 3 times by PBS to remove the medium that could interfere with mass spectra.Finally, the dish was filled with PBS and placed under an inverted microscope stage (CKX53, Olympus, Tokyo, Japan) to select targeted cells and monitor the sampling process.
Borosilicate glass capillary tube (I.D. 0.86 mm, O.D. 1.5 mm) was pulled into the micropipette with a tip of ~ 2 μm opening using P-1000 puller (Sutter Instruments, Novato, CA, USA).The parameters are as follows: heat = 546, pull = 0, velocity = 19, delay = 1, and pressure = 600.The pulled glass capillary was fixed on a metal holder, which attached to a 3-dimensional translation manipulator platform (MP-225, Sutter Instrument, Novato, CA, USA).The capillary was carefully inserted into the cell, and negative pressure was applied through an air pump connected to the back end of the metal holder.The cytoplasm was sucked for 90 s.After sampling, the capillary was inserted into the device coaxially to construct the concentric nanoESI-APCI hybrid ionization source.The sampling image of the single cell is shown in Figure S7.A visualized cell sampling process is provided in Video S1.

CSC and NSCC culture
The CSCs were obtained by microsphere culture method.CWR-22Rv1 cells were seeded in a 6-well ultralow attachment plate with serum-free sphere medium, and CSCs would proliferate and form suspended cell spheres.Sphere medium consisted of DMEM/F12 (1:1, v/v), 20 ng/mL bFGF, 20 μL/mL B27 and 20 ng/mL EGF.After 7-10 days, CSCs were filtered and collected with a 100 μm cell sieve.
Then, cells were digested with 1 mL Accutase at 37 °C water bath for 10 min, and 2 mL sphere medium was added to terminate the digestion.Next, cells were transferred to a 5 mL centrifuge tube for a 2 min centrifugation at 100 g.Finally, CSCs were washed twice with PBS and resuspended in PBS for subsequent single-cell sampling.
For comparison, we utilized the regular CWR-22Rv1 cells cultured in DMEM/F12 (1:1, v/v) containing 10% FBS as the model of NSCCs.The culture process was detailed in the "Cell culture and sampling" section.

Human hepatocellular carcinoma (HCC) tissue sample collection and processing
One HCC tissue sample was collected by surgical resection, having been approved by the local Ethical Review Board of the First Affiliated Hospital of Dalian Medical University (Dalian, China, PJ-KS-KY-2023-77).The patient provided written informed consent.The diameter of the tissue sample was 1-2 cm.
The HCC tissue was flash-frozen in liquid nitrogen for 20 s after resection, then transferred to cryogenic vials and stored at -80 °C until sectioned to a 10-μm-thick frozen section at -20 °C using a Leica CM1950 cryostat microtome (Leica Biosystems Nussloch GmbH, Wetzlar, Germany).The 10-μm-thick frozen sections were mounted onto 2.0 μm PEN membrane glass slides.Nine adjacent frozen sections were prepared, one of which was stained by cresyl violet acetate.Before the microdissection, the frozen slices were dried in a vacuum for ~30 min.The Nissl staining image of HCC tissue is illustrated in Figure 5A, it can be observed that HCC tissue section had three histologic types: paracancerous normal tissues, fibroblast tissues and tumor tissues.

Sampling and pretreatment of tissue microregions from tissue sections
The commercial laser capture microdissection (LCM) system (Leica LMD7000, Leica Microsystems, Wetzlar, Germany) was used for microdissection of tissue sections, based on a highly focused and precisely controlled laser beam dissecting microregions of interest and collecting them for later analysis [1] .
After laser beam cutting of the targeted microregions outlines drawn using the LMD7000 mapping tool, microregions were recovered by gravity-assisted dropping of the section into a collection container (the capture cap of a 0.2 mL centrifuge tube), located below the sample substrate.Figure S12 demonstrates the 'cut and drop' sampling process.Five μL extraction solvent (80% MeOH/H2O, 10 mM AmFa, 10% formic acid) was added to the collection tube and ultrasonication was performed for 30 min using a noncontact ultrasonic crusher (MX-96A, XiaoMeiChaoSheng, Kunshan, China) to extract metabolites from the microregions.The concentric nanoESI/APCI hybrid ionization source was then used for subsequent analysis.
The ACQUITY ultra-high performance liquid chromatography system (UHPLC, Waters, Milford, MA, USA) coupled with the Q Exactive-HF MS was used for LC-MS analysis of population cells.The columns for the positive and negative ion modes were ACQUITY UPLC BEH C8 (100 mm × 2.1 mm, 1.7 μm, Waters, Milford, MA, USA) and ACQUITY UPLC HSS T3 (100 mm × 2.1 mm, 1.8 μm, Waters, Milford, MA, USA), respectively.The column temperature was 50 °C, the flow rate was 0.35 mL/min, and the injection volume was set to 5 µL.In the positive ion mode, the mobile phases A and B were H2O with 0.1% formic acid and ACN with 0.1% formic acid, respectively.The elution gradient started with 5% B and maintained for 1 min.Then it was increased linearly to 100% B within 23 min and was kept for 4 min.The gradient was changed back to the initial gradient within 0.1 min and was held for 1.9 min to equilibrate the column.The total run time was 30 min.In the negative ion mode, the mobile phases A and B were H2O and MeOH, respectively, both containing 6.5 mM NH4HCO3.The initial gradient was 2% B and was kept for 1 min, then linearly increased to 100% B within 17 min and maintained for 4 min.The gradient was back to 2% B at 22.1 min and held for 2.9 min.The total run time was 25 min.
The MS was operated with a spray voltage of 3.5 kV in the positive ion mode and 3.0 kV in the negative ion mode.The capillary temperature was 300 °C, the flow rate of the sheath gas and auxiliary gas were set to 45 and 10 (in arbitrary units), respectively.The S-lens RF level was set as 50.The mass range was 85-1250 m/z.The resolutions of 120,000 and 30,000 at m/z 200 were separately set for full MS scan and data dependent MS/MS (ddMS2).The AGC target of 3e6 and maximum IT of 200 ms were applied for full MS scan.While the values were 1e5 and 50 ms in ddMS2 settings.The top 10 ions with the highest abundance per full MS scan were selected for MS/MS acquisition.The normalized collision energies (NCE) were 15, 30, 45 eV, respectively, and the dynamic exclusion time was 7.0 s.Metabolites were identified based on retention time, accurate mass and MS/MS spectra using HMDB, METLIN, mzCloud, MoNA and home-built databases.

Data processing and analysis
A homemade Python script was developed to process single-cell metabolomics raw data, peak alignment, stable feature ion screening, metabolites identification and machine learning.Raw MS data were recorded with Xcalibur software (v2.2, Thermo Fisher Scientific, San Jose, CA, USA), and exported as CSV files including m/z, intensity and S/N.Phosphorylcholine (m/z 184.0733) and creatine (m/z 132.0768) were used as markers for the detection of single-cell events to extract all ion signals associated with the ionization of single-cell contents.Signals with S/N greater than 3 and the detection rate in all cell events greater than 50% were selected as stable characteristic ions for subsequent analysis.After the ion intensities were normalized to total ion intensity, peak alignment was performed between different single cell samples.The metabolites were tentatively assigned according to the accurate mass comparison of the database (mass error < 5 ppm) and MS/MS spectra obtained by liquid chromatography-mass spectrometry (LC-MS) analysis of population cells.Databases include Human Metabolome Database (HMDB, http://www.hmdb.ca),METLIN Metabolite Database (https://metlin.scripps.edu/),mzCloud (https://www.Mzcloud.org/),MassBank of North America (MoNA, https://mona.fiehnlab.ucdavis.edu/)and home-made database OSI-SMMS.A machine learning algorithm based on t-distributed stochastic neighbor embedding (t-SNE) was utilized to reduce the dimensionality of complex metabolic datasets to a two-dimensional plane.Principal component analysis (PCA) was performed using SIMCA-P software (v13.0,Ume trics, Umea, Sweden).The Mann-Whitney U test (M-W test) and Kruskal-Wallis H test (K-W test) were used to evaluate the statistical significance level (p-value) of single-cell metabolic profiles between two or three different cell types.SPSS software (v25.0,IBM, Armonk, USA) was used to conduct the evaluation, and p < 0.05 was considered to be statistically significant.Volcano plots were drawn using Origin software (Origin2023, Northampton, Massachusetts, USA).Heatmap visualization was performed using Multi Experiment Viewer software (MeV, v4.9.0, Dana-Farber Cancer Institute, MA, USA) to show the expression level of single-cell metabolites among different types of cell lines.Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis was conducted on MetaboAnalyst website (https://www.metaboanalyst.ca/).

The possible reaction mechanism involved in the ionization process
The ionization process involves two key reactions, namely the charge transfer reaction and the proton transfer reaction.

Figure S12 .
Figure S12.The laser 'cut and drop' sampling process.(A) Image of target tissue microregion in HCC tissue section obtained by LCM.(B) Image of HCC tissue section after microdissection.(C) Image of target tissue microregion collected in the collection tube.

Figure S13 .Figure S14 .
Figure S13.Discrimination results of cancer cell types.(A) The t-SNE plot with 95% confidence ellipses of three types of cancer cell lines including 43 MCF7 cells, 39 97H cells and 50 PC3 cells.(B) Heatmap of significantly changed metabolites with p < 0.05 for the discrimination of cancer cell types.Color indicates z-scores of metabolites.(C) KEGG metabolic pathways differentially regulated in three different cancer cell types.Each circle was colored by the -log10(p-value) and the size was correlated to matched metabolites

Figure S15 .
Figure S15.The influence of GS treatment on metabolites of MCF7 cells.(A) The t-SNE plot with 95% confidence ellipses of 44 GS-treated cells and 43 MCF7 cells.(B) Volcano plot showing correlations between p-value and fold-change for all metabolites in GS treatment and control groups.Characteristic metabolites are highlighted in red (upregulated) and blue (downregulated).(C) Heatmap of significantly changed metabolites with p < 0.05 for the discrimination of GS treatment and control groups.Color indicates z-scores of metabolites.(D) KEGG metabolic pathways differentially regulated in GS-treated cells compared to MCF7 cells.Each circle was colored by the -log10(p-value) and the size was correlated to the number of matched metabolites.(E) Z-score plot of 17 representative metabolites with differential abundance in GS treatment and control groups.Data are presented as median with interquartile range and points are colored by assigned cell type.*: 0.001 < p < 0.05, **: 0.001 < p < 0.0001, ***: p < 0.0001.

Figure S16 .
Figure S16.The outcome of t-SNE cluster analysis with 95% confidence ellipses of metabolites with statistically significant differences (p < 0.05) between 24 CSCs and 18 NSCCs.

Table S1 .
Detailed information of 10 classes of model compounds.

Table S2 .
LOD and linear range acquired in both modes for different model compounds.

Table S3 .
Assigned metabolites from a single cell in nanoESI and nanoESI-APCI modes.

Table S4 .
Performance comparison between related studies and ours.