Many tumors arise from cells or clones with distinct genetic profiles with distinct genetic profiles which contribute to genetic diversity of a tumor, called intratumoral heterogeneity (ITH). Characterizing tumor ITH is important for treatment efficiency as the targeting of a single type of genetic mutation or pathway might be efficient on one clone but not on other clones, leading to treatment resistance and relapse. Analyzing genetics at the single cell level, provides valuable information about the tumor composition and whether there may be a competition between tumor clones during therapy. However, single-cell genetic DNA analyses are time consuming and difficult to validate in part because the analyses are performed on a single cell and no replicate is available. Additionally, compared to genetic analyses on bulk tumor samples, single-cell analyses are extremely sensitive to errors including allele dropout, false positives, false negatives or reading of doublets instead of a single cell and decrease the reliability of the method.
Drs. Alison Thompson and Amy Paguirigan, postdoc and staff scientist, respectively, in Dr. Jerald Radich’s lab (Clinical Research Division), led a study recently published in Plos One reporting the development of a microfluidic chip device allowing single-cell genotyping of a defined genomic region of interest. The new device generates higher quality data by integration of quality controls in the read-out. Dr. Thompson explained, “Dr. Paguirigan’s previous paper in Science Translational Medicine raised some interesting questions about how complex intratumoral heterogeneity in adult myeloid leukemia (AML) can be missed or misinterpreted when solely relying on bulk sequencing. In AML there are a relatively limited number of mutations that occur in a given patient as compared with other more genetically diverse cancer types, making a targeted approach feasible. Our goal was to develop a targeted, single-cell genotyping platform that would make these types of assays much easier, cheaper and quicker than the work we did previously. We also wanted to develop a transparent method where we could quantify error rates for every run and ensure that genotypes were only called on visually confirmed single cells so that we could have better statistics. Microfluidics is potentially a good fit for this type of problem”.
The team developed a “targeted single-cell genotyping platform using the microfluidic self-digitization chip”, or SD chip, to load and analyze samples at the single-cell level. Each of the 1024 wells on the array can be loaded with a 5 nanoliters of PCR mix containing three hydrolysis probes coupled to different fluorochromes to distinguish wild-type (WT) or mutated alleles of the gene of interest as well as a positive control probe specific to the same gene but outside the mutated region. Prior to running the PCR, imaging of the chip identifies empty wells, or wells containing one or two cells. Single cell analyses were performed only for wells with confirmed amplification of both control probe and WT or mutated allele along with a single cell signal detected pre-PCR.
More than one third of adult AML patients have mutations in NPM1 (Nucleophosmin 1), a gene involved in protein chaperoning and cell proliferation. NPM1 was selected as the gene of interest as nearly all AML patients that have a mutation in NPM1 have a 4bp insertion at a particular genomic coordinate. The chip, primers and PCR conditions were first optimized and validated using plasmids and cell lines containing WT or mutated NPM1 gene. The rates of mutant false positives were minimized down to 0.3% (2 false positive mutants out of 610 WT plasmids runs) and WT false positives down to 0.3% (2 false positive WT out of 885 mutated plasmids runs). Allele dropout, characterized by successful PCR but unsuccessful amplification of both WT and mutated alleles, was also reduced using PCR optimized on a heterozygous plasmid template. The allele dropout rates were 5.4% for the WT allele and 8.5% for the mutated allele. Interestingly, validation in cell lines KG1a (WT) and OCI-AML3 (heterozygous) different clonalities of the AML cell line than expected based on previous bulk cell samples data (p-value<1e-5). Such results point out the potential inaccuracy of assumptions made based on bulk single cell DNA sequencing analyses relative single cell analyses such as the SD chip.
One SD chip can analyze a single locus per run for each sample, and the group envisions further developments to improve utility. Dr. Thompson highlights the cost and time savings in moving to the microfluidic system, “To genotype about 300 single cells for a single locus, the cost for reagents and instrument time went from over $1000 to less than $10. The time from sample to answer went from about a week to a single day and now we have much more informative statistics (for example, using the previous method, we could never truly get the cell number per reaction so we had no false positive/doublet+ rates)”. Such improvements are promising for future applications. “We are developing assays for additional loci of interest in AML and would like to develop assay suites to look for mutation co-occurrence in a single-cell. We are also want to further develop the idea of using the device to test predictions from bulk sequencing about the genetic diversity in a tumor”, concluded Dr. Thompson.
Funding for this study was provided by the National Institutes of Health.
Thompson AM, Smith JL, Monroe LD, Kreutz JE, Schneider T, Fujimoto BS, Chiu DT, Radich JP, Paguirigan AL. 2018. Self-digitization chip for single-cell genotyping of cancer-related mutations. PLoS One. 13(5), e0196801.