Liquid biopsy-based single-cell metabolic phenotyping of lung cancer patients for informative diagnostics

Accurate prediction of chemo- or targeted therapy responses for patients with similar driver oncogenes through a simple and least-invasive assay represents an unmet need in the clinical diagnosis of non-small cell lung cancer. Using a single-cell on-chip metabolic cytometry and fluorescent metabolic probes, we show metabolic phenotyping on the rare disseminated tumor cells in pleural effusions across a panel of 32 lung adenocarcinoma patients. Our results reveal extensive metabolic heterogeneity of tumor cells that differentially engage in glycolysis and mitochondrial oxidation. The cell number ratio of the two metabolic phenotypes is found to be predictive for patient therapy response, physiological performance, and survival. Transcriptome analysis reveals that the glycolytic phenotype is associated with mesenchymal-like cell state with elevated expression of the resistant-leading receptor tyrosine kinase AXL and immune checkpoint ligands. Drug targeting AXL induces a significant cell killing in the glycolytic cells without affecting the cells with active mitochondrial oxidation.


Supplementary Figures
Supplementary Fig. 1 a Fluorescent images of a viable A549 cell, a viable leukocyte and a dead A549 cell (scale bar, 30 μm). b 2-NBDG uptake of viable leukocytes (n=1011), viable A549 cells (n=101), and dead A549 cells (n=124). Dead cells have minimal unspecific 2-NBDG signal (mean ± SD). c Precise single-cell retrieval from a microwell containing multiple cells. Left, before single cell retrieval; Right, after single cell retrieval (scale bar, 30 μm). Source data are provided as a Source Data file. Supplementary Fig. 2 a On-chip metabolic cytometry of C12R signal (left, n=3329 for control, n=3674 for oligomycin) and 2-NBDG (right, n=3784 for control and n=3499 for oligomycin) uptake of A549 cells treated with DMSO control and oligomycin. The data are represented as scatter plots and bar columns (mean ± SD). b Percentages of C12R high /2-NBDG low cells in five perturbed conditions (vehicle, glucose, glutamine, sodium lactate or sodium pyruvate) are represents in bar columns (n=3 independent experiments, mean ± SEM), derived from three independent experiments. * P < 0.05; ** P < 0.01; *** P < 0.001.
c Representative flow cytometry plots report 2-NBDG and C12R fluorescence intensity of A549 cells treated with vehicle, glucose, glutamine, sodium lactate or sodium pyruvate, respectively. It is noted that gating of C12R high /2-NBDG low cells shown in this figure is different from that of metabolically active cells. d Changes of ECAR and OCR of A549 cells in response to 2-DG (5 mM), oligomycin (2 μM), phenformin (25 μM), BPETS (10 μM), and etomoxir (200 μM) by using DMSO as the control (n=3 independent samples for each condition). Each sample represented the average of ten cycles of ECAR and OCR measurements (mean ± SD). Source data are provided as a Source Data file. Supplementary Fig. 3. Validation of mutual interference of the two metabolic probes. Top, box chart of fluorescence intensity of C12R and 2-NBDG uptake in A549 cells under variable C12R concentrations (1 M, 5 M, 10 M and 15 M) and a fixed 2-NBDG concentration (100 M) by on-chip metabolic cytometry; Bottom, box chart of fluorescence intensity of C12R and 2-NBDG uptake in A549 cells under variable 2-NBDG concentrations (100 M, 200 M, 400 M and 600 M) and a fixed C12R concentration (1 M) by on-chip metabolic cytometry. The data was generated from single-cell measurements for each condition (Top, left to right, n=1030, 960, 1110(Top, left to right, n=1030, 960, , 941, 1036(Top, left to right, n=1030, 960, , 987, 1096Bottom, left to right, n=6773, 7401, 6886, 7078, 6862, 7267, 6841, 7058) and represented in box plots (center line, median; square, mean; box limits, upper and lower quartiles; whiskers, 1.5*inter-quartile range). Source data are provided as a Source Data file. Fig. 4 Cut-off determination for separating putative metabolically active tumor cells from leukocytes. Top, scatter plot generated from flow cytometry reports 2-NBDG and C12R fluorescence intensity of leukocytes spiked with (a) A549 and (b) H1650 cells. Leukocytes were isolated from the blood of a healthy donor. Bottom, histograms of CD45, 2-NBDG and C12R levels of leukocytes and spiked tumor cells. The cut-off values of metabolic markers 2-NBDG and C12R is defined as mean plus five standard deviations of leukocytes, as indicated by the black lines.

Supplementary Fig. 5
Percentage of single cells harboring oncogenic driver mutations that determine the malignancy of metabolically active cells. The detected oncogenic driver mutations were consistent with those detected in cell blocks of MPE or primary tumor sites Supplementary Fig. 6 Single-cell CNV profiles and detected EGFR mutation status of metabolically active cells and white blood cells from P1. Fig. 7 Representative flow cytometry plots report 2-NBDG and C12R fluorescence signals of H1975, H1650, A549, MG63, 143B and HCT116 cells that were assayed by 600 μM 2-NBDG and 1 μM C12R. 50,000 events were recorded for each sample. The correlation coefficients (CC) between 2-NBDG and C12R (mean ± SD), derived from four independent experiments, in these cell lines range from 0.54 to 0.64.

Supplementary Fig. 8
Scatter plot of OMC assays on Patient 2 MPE samples. 2-NBDG and C12R fluorescence intensity of all CD45 negative cells (black and red dots) and a portion of CD45 positive cells (blue dots) are shown. The cut-offs for identification of 2-NBDG high and C12R high cells (black dots) are shown in the figure as the black lines. 2-NBDG high and C12R high cells are candidate tumor cells and gated out by five and three standard deviations above mean of CD45 + leukocytes, respectively. Red dots represent CD45 neg /2-NBDG low /C12R low cells and CD45 pos leukocytes are displayed in blue dots. In Patient #2, 11 CD45 neg /2-NBDG high /C12R low cells, 10 CD45 neg /2-NBDG low /C12R high cells and 9 CD45 neg /2-NBDG high /C12R high cells were identified as candidate tumor cells that were found to harbor an EGFR 19Del (p. L747_E749del) and an A750P mutation in exon 19 of EGFR.
Supplementary Fig. 9 The bright field and fluorescence composite images of representative blocks shows both C12R high cells and 2-NBDG high cells. The fluorescence signals of CD45, 2-NBDG and C12R are shown in red, green and blue, respectively (scale bar, 100 μm). The different metabolic phenotypes of MPE samples that are classified by the boundary lines at N/R = 0.5 and 2 can segregate patient response. The green dots denote patients who were on therapies at the MPE draw and metabolic phenotyping. The orchid dots denote newly diagnosed patients who were receiving first-line therapies after the MPE draw and metabolic phenotyping. The blank green dots represent the patients with a PR response to the current therapy at the time of MPE draw and the solid green dots represent those with a SD or PD response at the time of MPE draw. PR: partial response, SD: stable disease, PD: progressive disease.
involved in glutaminolysis between the two metabolic phenotypes (2-NBDG high vs C12R high ) across 5 patients. The glutaminolysis-related genes showed a mixed pattern of change across patients. Supplementary Fig. 13 Relative viability of two metabolically active cell populations in response to 2-DG (5 mM), phenformin (25 μM), BPETS (10 μM), and etomoxir (200 μM) , respectively, with respect to the DMSO control (* P<0.05; ** P<0.005; NS, not significant). Data are presented as the mean ± SD (n=2 independent samples for each condition). Source data are provided as a Source Data file. Supplementary Fig. 14 PD-L1 expression levels, quantified by the single-cell immunofluorescence staining, for the three metabolic phenotypes of CD45 neg cells from MPE sample of P31. All 2-NBDG high /C12 high cells (n=66) and 2-NBDG low /C12R high cells (n=13) on a chip as well as randomly selected double negative cells (n=244) were assayed and shown in the figure (mean ± SD). Source data are provided as a Source Data file. Fig. 15 AXL fluorescence intensity of the three metabolic phenotypes of CD45 neg cells from MPE sample of P29. All 2-NBDG high cells (n=97) and C12R high (n=44) as well as randomly selected double negative cells (n=100) were assayed and shown in the figure (mean ± SD). Source data are provided as a Source Data file. Supplementary Fig. 16 The number and viability of metabolically active cells assayed under DMSO control and AXL inhibition treatment in P3, P9 and P29. Each experiment had three replicates. Supplementary Fig. 17. Evaluation of WGA efficiency using specific primers for 22 chromosomes (see Supplementary Table 5). 22 PCR reactions were performed and products of PCR reactions were separated by gel electrophoresis.

Supplementary Tables
Supplementary Table 1