PMC

mRNA expression analysis following PACS enrichment. (A) qRT-PCR amplification curves from GAPDH or CD9 performed on cell line isolated total RNA control samples. CD9 was expressed significantly higher in DU145 cells than in Raji cells. (B) RNA from presorted and vimentin-positive PACS (Vim sorted) droplets was analyzed for CD9 expression following normalization of input levels with GAPDH. CD9 was only detected in the vimentin-positive sorted droplets indicating that PACS enriched for DU145 expressed transcripts. Three replicate amplification curves are shown for each qRT-PCR experiment.

Workflow for PCR-activated cell sorting. (A,B) Raji and DU145 cells are isolated into aqueous microdroplets in an oil-based emulsion and lysed. Only DU145 cells express vimentin mRNA and have genetic mutations in RB1 and CDKN2A genes. (C) A microfluidic chip then processes the cells, readying the lysate for PCR. Single-cell TaqMan PCR reactions targeting vimentin mRNA are thermocycled and droplets are sorted based on positive TaqMan probe fluorescence. (D) Following microfluidic droplet sorting, the cell lysate can be recovered for downstream nucleic acid analysis.

Single-cell vimentin TaqMan assays are specific for DU145 cancer cells. (A) Merged brightfield and fluorescence images showing amplified vimentin TaqMan probe (red), calcein violet DU145 (blue) and calcein green Raji (green) lysate in droplets. The presence of a purple-violet color indicates droplets where both DU145 calcein violet stain and HEX from the vimentin TaqMan probe were detected. Individual fluorescence channels from the dashed region are shown below. Scale bar is 100 μm. (B) Vimentin TaqMan detection rates of individual DU145 and Raji cells processed with the single-cell RT-PCR microfluidic workflow. Data was compiled from replicate experiments analyzed with a MATLAB script.

Quantitative analysis of PACS genome enrichment with next-generation sequencing. Analysis of RB1 and CDKN2A genomic loci for the presence of DU145-specific SNPs. Sequencing libraries from RB1 and CDKN2A amplicons were generated using Nextera XT reagents. (A) Quantitative analysis of RB1 sequence reads demonstrated that the DU145-specific nonsense mutation, AAG to TAG (see Figure ), was found in 6.2% of the sequence reads generated from presorted cell lysate. Following PACS sorting (Vim+ PACS) the presence of this mutation relative to the Raji-specific codon was enriched to 87.7%. (B) Similar data was obtained upon sequence analysis of CDKN2A amplicons generated from presorted and Vim+ PACS sorted cell lysate. The DU145-specific missense mutation, GAC to TAC (see Figure ), went from comprising 13.5% of the sequence reads to 74.2% of the sequence reads upon PACS enrichment. More than 15 000 sequence reads were analyzed for each of the four samples.

Single-cell genomic SRY TaqMan assays are specific for male DU145 cells. (A) Merged brightfield and fluorescence images showing amplified SRY TaqMan probe (red), calcein violet DU145 (blue) lysate and calcein green HEK293 (green) lysate in droplets. The presence of a purple-violet color indicates droplets where both DU145 calcein violet stain and HEX from the SRY TaqMan probe were detected. Individual fluorescence channels from the dashed region are shown below. Scale bar is 100 μm. (B) SRY TaqMan detection rates of individual DU145 and HEK293 cells processed with the single-cell PCR microfluidic workflow. The SRY reaction detection rate for DU145 cells was 45.6% (+/− 0.4%) and the detection rate for HEK293 cells was 0.2% (+/− 0.3%). Data was compiled from replicate experiments analyzed with a MATLAB script.

Sanger sequencing of DU145 enriched genomic DNA following sorting based on vimentin mRNA detection. (A) A portion of the RB1 locus was amplified from genomic DNA isolated from individual cell lines and Sanger sequenced as a control (top two sequences). Raji cell RB1 encodes for a lysine at amino acid position 715 (black box). DU145 genomic DNA has a nonsense mutation at this position (red box). Sequencing of genomic DNA amplified from droplets prior to vimentin-positive PACS sorting (Presorted) produces a Raji cell sequence with lysine at position 715. This is expected based on the initial encapsulation of a 90% Raji cell and 10% DU145 cell suspension. Following vimentin-positive droplet sorting (Vimentin+ PACS), sequencing shows that the genomic DNA is dramatically enriched for the DU145-specifc stop codon. (B) Sequencing of CDKN2A amplicons from control cell genomic DNA (top two sequences). Amino acid 84 is mutated from an aspartic acid to a tyrosine in DU145 cells (red box). Sequencing of the presorted single-cell RT-PCR emulsion DNA yields a Raji-specific aspartic acid. Following vimentin-positive PACS sorting, the genomic DNA is enriched for tyrosine encoding sequences. Arrows indicate the position of the SNPs.

Ultrahigh-throughput detection and sorting with single-cell vimentin RT-PCR reactions. (A) Photograph of the dielectrophoretic microfluidic sorter. Reinjected emulsion entered the device from the left and was interrogated for fluorescence at the laser spot. A voltage was applied to the sorting electrode when a droplet was positive for both calcein and HEX above the specified thresholds. This pulled the specified droplets into the lower channel for collection. Scale bar is 100 μm. (B) Scatterplot diagram of single-cell RT-PCR sorted droplets showing the calcein violet cell stain fluorescence used to mark Raji and DU145 cells on the x axis and HEX 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 were selected for sorting. This PACS data was generated from an initial 80% Raji and 20% DU145 heterogeneous cell suspension. (C) Presorted and sorted droplets were imaged to evaluate sorting efficiency. Arrowheads in the presorted emulsion image point to two droplets positive for both calcein and HEX. The rest of the droplets are either empty or only calcein positive. Nearly all of the droplets following sorting are positive for both calcein and HEX.

Enrichment of DU145-specific genomic SNPs following sorting based on Y chromosome DNA detection. (A) A portion of the RB1 locus was amplified from genomic DNA isolated from individual female HEK293 and male DU145 cell lines and Sanger sequenced as a control (top two sequences). Similar to the results shown in Figure , DU145 genomic DNA contains a nonsense mutation at amino acid position 715 (red box). Sequencing of genomic DNA amplified from droplets prior to SRY-positive PACS sorting (Presorted) produces a HEK293 cell sequence with a lysine at position 715. This lysine codon predominates due to the fact that the initial cell suspension contained 90% HEK293 cells. Following SRY-positive droplet sorting (SRY+ PACS), sequencing shows that the genomic DNA is enriched for the DU145-specifc stop codon. (B) Sequencing of CDKN2A amplicons from control cell genomic DNA (top two sequences). Amino acid 84 is mutated from an aspartic acid to a tyrosine in DU145 cells (red box). Sequencing of the presorted single-cell RT-PCR emulsion DNA yields a HEK293-specific aspartic acid. Following SRY-positive PACS sorting, the genomic DNA is enriched for tyrosine encoding sequences. Arrows indicate the position of the SNPs.