Identification and genetic analysis of cancer cells with PCR-activated cell sorting

Cell sorting is a central tool in life science research for analyzing cellular heterogeneity or enriching rare cells out of large populations. Although methods like FACS and FISH-FC can characterize and isolate cells from heterogeneous populations, they are limited by their reliance on antibodies, or the requirement to chemically fix cells. We introduce a new cell sorting technology that robustly sorts based on sequence-specific analysis of cellular nucleic acids. Our approach, PCR-activated cell sorting (PACS), uses TaqMan PCR to detect nucleic acids within single cells and trigger their sorting. With this method, we identified and sorted prostate cancer cells from a heterogeneous population by performing >132 000 simultaneous single-cell TaqMan RT-PCR reactions targeting vimentin mRNA. Following vimentin-positive droplet sorting and downstream analysis of recovered nucleic acids, we found that cancer-specific genomes and transcripts were significantly enriched. Additionally, we demonstrate that PACS can be used to sort and enrich cells via TaqMan PCR reactions targeting single-copy genomic DNA. PACS provides a general new technical capability that expands the application space of cell sorting by enabling sorting based on cellular information not amenable to existing approaches.


Supplementary
. MATLAB generated scatterplots of drop fluorescence obtained from microscope images demonstrate vimentin probe specificity for DU145 cell lysate. (A) Vimentin fluorescence correlates well with DU145 cell lysate marked by calcein violet stain. (B) Correlation between vimentin RT-PCR and Raji cell lysate marked with calcein green stain is weak and the majority of events detected were due to co-encapsulation of Raji and DU145 cells. In 9 of 16 droplets where vimentin was detected with Raji cell lysate, DU145 lysate was also detected. (C) Raji and DU145 co-encapsulation is driven by Poisson statistics and is quite rare (1.3% of cell containing droplets). (D) Analysis of droplet size distribution indicates a high degree of emulsion stability.
Supplementary Figure S3. PACS droplet sorting efficiency is robust. Scatterplot analysis of fluorescent images taken of presorted (black plots) and PACS sorted droplets (violet plots). As expected, the presorted droplets are mostly empty and lack either calcein or vimentin-positive TaqMan HEX signal. Only 3 of these droplets were positive for both signals (upper right quadrant). In contrast, almost all (95.8% n=359) of the droplets that were PACS workflow sorted based on the presence of both cell lysate (calcein) and TaqMan signal (HEX) are indeed positive. This analysis demonstrates the high fidelity nature of the microfluidic sorting device employed in the PACS workflow.
Supplementary Figure S4. MATLAB generated scatterplots of drop fluorescence obtained from microscope images demonstrate SRY probe specificity for DU145 cell lysate. (A) Correlation between SRY PCR and HEK293 cell lysate, marked with calcein green stain, was extremely rare. Only one double positive was detected from two replicate experiments and this was likely due to coencapsulation with a DU145 cell or a cell-free chromosomal fragment during drop making (B) SRY fluorescence correlates well with DU145 cell lysate, marked by calcein violet stain. (C) HEK293 and DU145 co-encapsulation events. (D) Analysis of droplet size distribution indicates a high degree of emulsion stability.
Supplementary Figure S5. Detection and sorting of DU145 cancer cells using a genomic SRY TaqMan assay. Scatterplot diagram of single-cell RT-PCR sorted droplets showing the calcein violet cell stain fluorescence used to mark HEK293 and DU145 cells on the x axis and HEX (SRY probe) fluorescence from the TaqMan positive reactions on the y axis. Dashed red lines indicate where the sorting thresholds were applied. Only droplets in the upper right quadrant (DU145 cells) were selected for sorting. This PACS data was generated from an initial 90% HEK293 and 10% DU145 heterogeneous cell suspension. Sorted droplets shown in this scatterplot were used for subsequent downstream sequencing to demonstrate successful DU145 lysate enrichment following PACS (Fig. 8).
Supplementary Figure S6. Efficient multiplexing with droplet-based single-cell RT-PCR TaqMan assays. DU145 cells were examined for the simultaneous expression of actin and vimentin transcripts using a single-cell multiplexed TaqMan assay. Data from fluorescent images taken of the droplets following microfluidic single-cell RT-PCR preparation and thermocycling is shown in scatterplots. (A) Two-dimensional plot of actin and vimentin TaqMan probe fluorescence in drops. Drops that also contain DU145 cell lysate, determined by calcein violet staining DU145 cells, are plotted as the darker data points. Drops without calcein fluorescence appear faded. Actin and vimentin were simultaneously detected with the multiplexed reaction in 96.0% (n=273) of all drops containing DU145 lysate. The high detection rate with this multiplex reaction indicates that the reactions are efficient and robust. (B) The same data presented in (A) is shown on a three-dimensional plot with calcein violet fluorescence represented by the vertical axis. Multiplex reactions such as this can be extremely useful in identifying and isolating unique cell types based on the correlation of multiple biomarkers with PACS.