Clonal tracking using embedded viral barcoding and high-throughput sequencing

Abstract

Embedded viral barcoding in combination with high-throughput sequencing is a powerful technology with which to track single-cell clones. It can provide clonal-level insights into cellular proliferation, development, differentiation, migration, and treatment efficacy. Here, we present a detailed protocol for a viral barcoding procedure that includes the creation of barcode libraries, the viral delivery of barcodes, the recovery of barcodes, and the computational analysis of barcode sequencing data. The entire procedure can be completed within a few weeks. This barcoding method requires cells to be susceptible to viral transduction. It provides high sensitivity and throughput, and enables precise quantification of cellular progeny. It is cost efficient and does not require any advanced skills. It can also be easily adapted to many types of applications, including both in vitro and in vivo experiments.

Fig. 1 |. Experiment workflow.

a, Synthesized semi-random barcode oligos (Table 1) are cloned into plasmids before packaging into a lentiviral vector. Cells of interest are then transduced. To retrieve barcodes, genomic DNA is extracted before qPCR amplification and high-throughput sequencing. Raw sequencing data are processed by a custom data analysis pipeline to quantify the abundance of each barcode. b, PCR strategy. The 33-bp cellular barcode, comprising a 6-bp library ID and a random 27-bp barcode, is flanked by an Illumina TruSeq read1 sequence and a custom read2 sequence so that a single PCR reaction can add the Illumina P5 and P7 adaptors to the ends of each barcode. See Table 2 for primer sequences. RE, restriction enzyme.

Fig. 2 |. Comparing barcode extraction replicates.

Primary mouse hematopoietic stem cells were barcoded and transplanted into recipient mice. Four months after transplantation, the mice were bled, and white blood cells were collected and processed according to Steps 69-126. Cell lysates were divided into two replicate samples and processed separately for genomic DNA extraction, barcode amplification, and sequencing. Each dot represents a barcode. Barocde abundance is highly consistent between the two replicated samples. Pearson correlation: 0.99; P = 5.3 ×10⁻¹⁴⁴. Animal procedures were approved by the Institutional Animal Care and Use Committee of the University of Southern California, and the mice were maintained at USC’s Research Animal Facility.

Fig. 3 |. qPCR amplification of barcode.

BB88 cells were barcoded and 50,000 GFP⁺ cells were sorted via FACS 1 week after transduction. gDNA was isolated and amplified using primers from Table 2. A multi-component plot of barcode amplification is shown. EvaGreen fluorescent dye (green lines) was used to quantify DNA amounts; thus no Rox signal was observed (red lines). a, Two samples with similar amounts of genomic DNA were amplified together, and their exponential curves emerged at similar numbers of PCR cycles. We stopped the reaction at cycle 25, which is about halfway through the exponential phase. This was to avoid over-amplification and to reduce background signals. No-DNA template control samples showed no amplification (two flat green lines). b, One sample was amplified to saturation. This is an example of over-amplification. a.u., arbitrary units.

Fig. 4 |. Optimizing the edit distance thresholds.

Histograms show the distances between unique sequences and their corresponding master barcodes (red), as well as the distances between different master barcodes (blue). Each row shows one edit distance threshold. Data from two independent samples are shown in the two columns. The threshold of edit distance of 4 was chosen as the point at which the distances between master barcodes are higher than and separated from the distances between unique sequences and their master barcodes.

Fig. 5 |. Python pipeline outputs.

Primary human acute lymphoblastic leukemia (ALL) cells were barcoded and transplanted into non-obese diabetic scid-gamma (NSG) mice. Two months after transplantation, the mice were bled, and ALL cells were collected and processed according to Steps 69–126. ALL cells barcoded with virus libraries 8 and 9 were used for this example. a, Custom algorithms written in Python code group reads on the basis of their library IDs. b, The Python algorithm quantifies each barcode with consideration to sequencing errors. Each color represents a unique barcode, and the size represents its relative abundance. Shown are data from library 9 in Fig. 5a. Animal procedures were approved by the Institutional Animal Care and Use Committee of the University of Southern California, and the mice were maintained at USC’s Research Animal Facility.