Therapies targeting signaling molecules mutated in cancers can often have striking short-term effects, but the emergence of resistant cancer cells is a major barrier to full cures 1,2 . Resistance can result from a secondary mutations 3,4 , but other times there is no clear genetic cause, raising the possibility of non-genetic rare cell variability [5][6][7][8][9][10][11] . Here, we show that melanoma cells can display profound transcriptional variability at the single cell level that predicts which cells will ultimately resist drug treatment. This variability involves infrequent, semi-coordinated transcription of a number of resistance markers at high levels in a very small percentage of cells. The addition of Users may view, print, copy, and download text and data-mine the content in such documents, for the purposes of academic research, subject always to the full Conditions of use: http://www.nature.com/authors/editorial_policies/license.html#terms Author contributions: SMS, AR designed the study. SMS performed all experiments and analysis except: MD, ST assisted with fluctuation analysis and RNA-sequencing; EAT, BE performed NGFR and AXL sort experiments; CK, MB, KS performed PDX experiments; PB, MH provided cell lines; MX performed WM989-A6 characterization; EE developed iterative RNA FISH protocol; INA, KN performed DNA sequencing. MH provided guidance. SMS, AR wrote the paper. Author information:AR receives consulting income and AR and SMS receive royalties related to Stellaris™ RNA FISH probes.

In single-cell RNA sequencing (scRNA-seq) studies, only a small fraction of the transcripts present in each cell are sequenced. This leads to unreliable quantification of genes with low or moderate expression, which hinders downstream analysis. To address this challenge, we developed SAVER (single-cell analysis via expression recovery), an expression recovery method for unique molecule index (UMI)-based scRNA-seq data that borrows information across genes and cells to provide accurate expression estimates for all genes.

Molecular differences between individual cells can lead to dramatic differences in cell fate, such as death versus survival of cancer cells upon drug treatment. These originating differences remain largely hidden due to difficulties in determining precisely what variable molecular features lead to which cellular fates. Thus, we developed Rewind, a methodology that combines genetic barcoding with RNA FISH to directly capture rare cells that give rise to cellular behaviors of interest. Applied to BRAF V600E melanoma, we trace drug-resistant cell fates back to single-cell gene expression differences in their drug-naive precursors (initial frequency of ~1:1000–1:10,000 cells) and relative persistence of MAP-kinase signaling soon after drug treatment. Within this rare subpopulation, we uncover a rich substructure in which molecular differences between several distinct subpopulations predict future differences in phenotypic behavior, such as proliferative capacity of distinct resistant clones following drug treatment. Our results reveal hidden, rare-cell variability that underlies a range of latent phenotypic outcomes upon drug exposure.

Non-genetic factors can cause individual cells to fluctuate substantially in gene expression levels over time. Yet it remains unclear whether these fluctuations can persist for much longer than the time for a single cell division. Current methods for measuring gene expression in single cells mostly rely on single time point measurements, making the time of a fluctuation of gene expression or cellular memory difficult to measure. Here, we report a method combining Luria and Delbrück's fluctuation analysis with population-based RNA sequencing (MemorySeq) for identifying genes transcriptome-wide whose fluctuations persist for several cell divisions. MemorySeq revealed multiple gene modules that express together in rare cells within otherwise homogeneous clonal populations. Further, we found that these rare cell subpopulations are associated with biologically distinct behaviors in multiple different cancer cell lines, for example, the ability to proliferate in the face of anti-cancer therapeutics. The identification of non-genetic, multigenerational fluctuations has the potential to reveal new forms of biological memory at the level of single cells and suggests that non-genetic heritability of cellular state may be a quantitative property. Main text:Cellular memory in biology, meaning the persistence of a cellular or organismal state over time, occurs over a wide range of timescales and can be induced by a variety of mechanisms. Genetic differences are one form of memory, encoding variation between organisms on multi-generational timescales. Within an organism, epigenetic mechanisms encode the differences between cell types in different tissues, with cells retaining memory of their state over a large number of cell divisions ( 1 ) . In contrast, recent measurements suggest that expression of many genes in single cells may have very little memory, displaying highly transient .

Although single-cell RNA sequencing can reliably detect large-scale transcriptional programs, it is unclear whether it accurately captures the behavior of individual genes, especially those that express only in rare cells. Here, we use single-molecule RNA fluorescence in situ hybridization as a gold standard to assess trade-offs in single-cell RNA-sequencing data for detecting rare cell expression variability. We quantified the gene expression distribution for 26 genes that range from ubiquitous to rarely expressed and found that the correspondence between estimates across platforms improved with both transcriptome coverage and increased number of cells analyzed. Further, by characterizing the trade-off between transcriptome coverage and number of cells analyzed, we show that when the number of genes required to answer a given biological question is small, then greater transcriptome coverage is more important than analyzing large numbers of cells. More generally, our report provides guidelines for selecting quality thresholds for single-cell RNA-sequencing experiments aimed at rare cell analyses.