Nanowire substrate-based laser scanning cytometry for quantitation of circulating tumor cells

Abstract

We report on the development of a nanowire substrate-enabled laser scanning imaging cytometry for rare cell analysis in order to achieve quantitative, automated, and functional evaluation of circulating tumor cells. Immuno-functionalized nanowire arrays have been demonstrated as a superior material to capture rare cells from heterogeneous cell populations. The laser scanning cytometry method enables large-area, automated quantitation of captured cells and rapid evaluation of functional cellular parameters (e.g., size, shape, and signaling protein) at the single-cell level. This integrated platform was first tested for capture and quantitation of human lung carcinoma cells from a mixture of tumor cells and leukocytes. We further applied it to the analysis of rare tumor cells spiked in fresh human whole blood (several cells per mL) that emulate metastatic cancer patient blood and demonstrated the potential of this technology for analyzing circulating tumor cells in the clinical settings. Using a high-content image analysis algorithm, cellular morphometric parameters and fluorescence intensities can be rapidly quantitated in an automated, unbiased, and standardized manner. Together, this approach enables informative characterization of captured cells in situ and potentially allows for subclassification of circulating tumor cells, a key step toward the identification of true metastasis-initiating cells. Thus, this nanoenabled platform holds great potential for studying the biology of rare tumor cells and for differential diagnosis of cancer progression and metastasis.

Figure 1. Fabrication and surface functionalization of quartz nanowire arrays

(a) Spin-coating process of polystyrene (PS) nanoparticles (NPs) on a flat quartz substrate. The size of the PS NPs was about 100 nm. (b) First O₂ plasma etching for size reduction of coated PS NPs. (c) Cr metal deposition (25 nm) using e-beam evaporator and lift-off of PS NPs with N-methy1-2-pyrrolidone (NMP). (d) Ni metal deposition used as an etch mask and Cr metal lift-off process. (e) and (f) Second plasma etching (top-view and tilted-view, respectively). RIE was performed to fabricate the QNW arrays with SF₄/Ar gas for 4 min. After RIE, the remaining Ni metal was completely removed with a wet-etchant (LCE-12K, Cyantek, USA). (g) Depiction of the step-by-step functionalization of nanowire surface with APTES, GA, streptavidin and biotinylated anti-human EpCAM antibody for circulating tumor cell capture.

Figure 2. Imaging and quantitation of lung carcinoma cells captured on nanowire arrays

(a) First and second column: low magnification and enlarged fluorescence images (white box region shown in A, B, C, D, E, and F) of captured human lung tumor cells (A549) on STR-functionalized QNW arrays bound with PDMS wells for different loaded cell populations in the range of 200 to 4,000 cells/well. Third column: CellProfiler^™ generated outlined images of captured cells on STR-QNW arrays for cell counting. Forth and fifth column: size and circularity histogram of immobilized A549 cells on STR-functionalized QNW arrays. The circularity (also known as form factor, ff), which is calculated as [4π (area)/(perimeter)²], represents the criterion of the circularity of the immobilized cells. If the form factor equals to 1, the captured object is a perfect circular object. The solid line represents a Gaussian fitting. (b) Correlation of total captured cells (A549) on STR-functionalized QNW arrays and STR-planar glass substrate versus loaded cells from cell suspension (R²~0.94 and ~0.75 for STR-QNW and STR-planar glass, respectively). Each result and error bar represents an average with standard deviation from three repeats (n=3). (c) Cell capture efficiency (yield) of the captured cells (A549) on two different topographies of substrates, STR-QNW arrays and STR-planar glass substrate as a function of loaded cells in the range of 200 to 4,300 cells/well. The solid-line represents a linear fitting. Each result and error bar represents an average with standard deviation (n=3). (d) Representative fluorescence images of captured A549 cells (human lung tumor cells) on STR-QNW arrays with cell loading in the range of 10 to 64 cells/sample. The captured cells were pre-stained by green-Vybrant DiI and scanned by microarray scanner. The immobilized cell population was then counted manually and also compared to the images from optical and fluorescence microscopy. Yellow-colored numbers (right-top) denote the number of captured cells for each well while red-colored numbers (right-bottom) indicate the total loaded cell population. (e) Correlation of total captured cells on STR-QNW arrays as a function of the loaded cells from cell suspension in the range of 6 to 64 cells/well, indicating a good linear relationship with the loaded cell population. The dotted line represents a linear fitting (R²=0.910). Each result shows an average with standard deviation (n=3). (f) Cell capture yield distribution versus loaded cell population in the range of 6 to 64 cells/well.

Figure 3. Capture of lung carcinoma cells from mixed cell populations

(a) Scanned images of captured cells from the mixture of A549 (green-labeled) and PBMCs (red-labeled) using nanowire arrays and the size distribution in pixels. A549/PBMC on STR-QNW arrays as a function of loaded cells in the range of 1,400 to 3,000 cells/well. (b) Both tumor cells (A540) and monocyte/background cells(U937) captured on nanowire substrate as a function of the ratio A549/U937 when the same amount of A549 cells were spiked in different densities of U93 cells. The result shows the tumor cell capture yield remains relatively constant although the non-specific capture of background cells significantly varies with the PBMC cell density. (c) Capture of rare tumor cells spiked in as-received PBMCs. Scanned images of tumor cells captured from an as-received human PBMC sample spiked with A549 lung cancer cells (15 cancer cells in 1mL PBMC suspension). Tumor cells were pre-stained with membrane dye DiD.

Figure 4. Capture, imaging and quantitation of lung cancer cells spiked in fresh human whole blood samples

(a) Scanned images of captured tumor cells on STR-functionalized NW substrates where only ~ 10 cells spiked into human whole blood sample. Typically, the whole blood sample (3 mL) was collected from either healthy volunteers (V1 and V2) or primary brain tumor patients (P1 and P2). 10 A549 cells were then spiked into 1mL of RBC-lysed blood. The mock metastatic patient blood samples were evenly aliquoted and loaded into 18 microwells on top of a nanowire substrate. To identify captured CTCs from numerous other cellular components of lysed whole blood, the samples were stained with DRAQ-5 (red-665 nm) and fluorescence-conjugated anti-cytokerain (blue-488 nm, eBioscience, USA) for all nuclei and epithelial tumor cells, respectively. Green-colored numbers (right-top) for each microarray scanned image (top of the first row) denote the number of captured cells for each well while red-colored numbers (right-bottom) indicate the microwells numbered 1 to 18. Using dual-color imaging, the tumor cells show purple (dual positive, red-DRAQ-5+/blue-CK-488+) while all the cells including non-specific cells can be identified by nuclear dye (no shown). (b) Microarray scanned images of negative control sample (as-marked NC2) from the same blood samples (V2) used for preparing spiked blood samples. (c) Summary of the captured cells for two peripheral patient samples (P1 and P2) and two volunteer blood samples (V1 and V2), showing the average capture yield to be ~ 67.5 ± 15% (n=4). 10 cells of A549 (CTCs) with final volume of 1 mL were used for the capture experiments.

Figure 5. Quantitation and characterization of tumor cells captured from human blood samples

(a) Tumor cell capture yield (the number of captured tumor cells as visualized by fluorescent marker vs. the nominal number of tumor cells spiked in 1mL of the whole blood sample). P1, P1, V1, and V2 denote samples from brain tumor patient 1, patient 2, healthy volunteers 1 and 2. (2) Cell size quantified for all captured tumor cells from four blood samples (P1, P2, V1, and V2) and one of the negative controls (NC2).