Copolymerization of single-cell nucleic acids into balls of acrylamide gel

Abstract

We show the use of 5'-Acrydite oligonucleotides to copolymerize single-cell DNA or RNA into balls of acrylamide gel (BAGs). Combining this step with split-and-pool techniques for creating barcodes yields a method with advantages in cost and scalability, depth of coverage, ease of operation, minimal cross-contamination, and efficient use of samples. We perform DNA copy number profiling on mixtures of cell lines, nuclei from frozen prostate tumors, and biopsy washes. As applied to RNA, the method has high capture efficiency of transcripts and sufficient consistency to clearly distinguish the expression patterns of cell lines and individual nuclei from neurons dissected from the mouse brain. By using varietal tags (UMIs) to achieve sequence error correction, we show extremely low levels of cross-contamination by tracking source-specific SNVs. The method is readily modifiable, and we will discuss its adaptability and diverse applications.

Figure 1.

Schematic of single-cell DNA or RNA BAG-seq workflow. The star between first split and second split indicating the place where more cycles of split-and-pool can be added.

Figure 2.

Copy number variation analysis of single-nucleus DNA (snDNA) BAG in cell lines and frozen prostate tumor. (A) Hierarchical clustering of four cell types SKN1, SK-BR-3, MCF-7, and BT-20 at a resolution of 20,000 bins (150 kbp per bin). Red indicates amplification, whereas blue indicates deletion. (B) The 20,000-bin copy number profiles from each of the four clusters in A. (C) Pathology image showing the region of Gleason 9 prostate cancer, which was estimated by pathologist as 60% tumor. Scale bar, 100 µm. (D–F) Representative snDNA BAG copy number profiles from this region: (D) a representative normal copy number profile; (E) a representative diploid tumor profile; and (F) a representative tetraploid tumor profile. (G) Hierarchical clustering of this region by combining data from both the BAG method and 96-well WGA method.

Figure 3.

CNV study of prostate tumor biopsy wash samples from a benign region and a Gleason 6 cancer region. (A) A 20× magnification pathology image of a benign region of the prostate. (B) Pathology image of a Gleason 6 cancer region from the same patient at the same resolution. (C) Hierarchical clustering of biopsy wash sample from the benign region. (D) Hierarchical clustering of biopsy wash sample from the Gleason 6 region showing a normal clone and two tumor clones based on CNV patterns. Red arrows indicate the major (clone 1) and minor (clone 2) tumor clones. (E) A representative normal single-nucleus copy number profile from the biopsy wash of this benign region. (F) Representative single-nucleus copy number profiles from one normal clone and two tumor clones from the biopsy wash of the Gleason 6 cancer region.

Figure 4.

Sequence error correction (ec) and analysis of cross-contamination using error-corrected SNVs. (A) Comparison of error rates between random sampling and ec in trinucleotide context. The number in each box indicates the error rate and is colored by its intensity. The middle base in the trinucleotide context is the “source” base, and the single base on top of each column is the “destination” base. For each “destination” base, the first column corresponds to random sampling method, and the second column corresponds to the ec method. (B) Minority SNV ratios of SK-BR-3 nuclei and SKN1 nuclei from the four-nuclei mixing experiment using the ec method showing very low contaminations between BAGs.

Figure 5.

Single-cell RNA (scRNA) BAG showing high yield, low contamination, and consistent expression profiles. (A–E) A two-cycle split-pool experiment including 235 cells. (F,G) A three-cycle split-pool experiment including 2875 cells. (A) Scatter plot showing the number of SKN1-specific and SK-BR-3–specific SNVs found in exons for each cell. BAGs with majority SKN1 or SK-BR-3 SNVs are colored blue or green. Two (0.85% of total) BAGs without clear majority SNVs are labeled as red. (B) Boxplot showing the number of genes detected per cell. (C) Boxplot showing the number of unique templates captured per cell. (D) Scatter plot of PC1 versus PC2. The scRNA BAGs are colored by their majority SNVs defined in A. Two bulk RNA data sets for SKN1 and SK-BR-3 clusters with their respective single-cell data. The contribution of PC1 is more than eight times that of PC2 (25.2%/3.0%). (E) Heatmap based on 40 (20+, 20−) genes with the most positive and negative correlations to PC1. (F) Scatter plot showing the number of SKN1-specific and SK-BR-3–specific SNVs found in exons for each cell in the three-cycle split-pool experiments. Nineteen (0.66% of total) cells without clear majority SNVs are labeled as red. (G) PC1 versus PC2 from the 2875 cells in the three-cycle split-pool experiment illustrated in F.

Figure 6.

Comparison of single-nuclei RNA clusters distinguishing sexes. (A) UMAP clustering of 860 nuclei from brain BNSTp region, and colored by sex. (B) Eight clusters in A are distinguished and labeled using different colors. (C) Nuclei are split by sex. There are 540 nuclei from males and 320 nuclei from females. (D) Dotplot showing features expression across all clusters. The size of the dot indicates the percentage of cells within a cluster, and the brightness of color indicates the expression level in a cluster.