Large-scale surveys of single-cell gene expression have the potential to reveal rare cell populations and lineage relationships, but require efficient methods for cell capture and mRNA sequencing1–4. Although cellular barcoding strategies allow parallel sequencing of single cells at ultra-low depths5, the limitations of shallow sequencing have not been directly investigated. By capturing 301 single cells from 11 populations using microfluidics and analyzing single-cell transcriptomes across downsampled sequencing depths, we demonstrate that shallow single-cell mRNA sequencing (~50,000 reads per cell) is sufficient for unbiased cell-type classification and biomarker identification. In developing cortex we identify diverse cell types including multiple progenitor and neuronal subtypes, and we identify EGR1 and FOS as previously unreported candidate targets of Notch signaling in human but not mouse radial glia. Our strategy establishes an efficient method for unbiased analysis and comparison of cell populations from heterogeneous tissue by microfluidic single-cell capture and low-coverage sequencing of many cells.

No abstract

No abstract

SummarySenescent cells play important roles in both physiological and pathological processes, including cancer and aging. In all cases, however, senescent cells comprise only a small fraction of tissues. Senescent phenotypes have been studied largely in relatively homogeneous populations of cultured cells. In vivo, senescent cells are generally identified by a small number of markers, but whether and how these markers vary among individual cells is unknown. We therefore utilized a combination of single‐cell isolation and a nanofluidic PCR platform to determine the contributions of individual cells to the overall gene expression profile of senescent human fibroblast populations. Individual senescent cells were surprisingly heterogeneous in their gene expression signatures. This cell‐to‐cell variability resulted in a loss of correlation among the expression of several senescence‐associated genes. Many genes encoding senescence‐associated secretory phenotype (SASP) factors, a major contributor to the effects of senescent cells in vivo, showed marked variability with a subset of highly induced genes accounting for the increases observed at the population level. Inflammatory genes in clustered genomic loci showed a greater correlation with senescence compared to nonclustered loci, suggesting that these genes are coregulated by genomic location. Together, these data offer new insights into how genes are regulated in senescent cells and suggest that single markers are inadequate to identify senescent cells in vivo.

Circulating tumor cells (CTCs) and circulating tumor DNA (ctDNA) have been recently investigated in several cancer types, but their respective clinical significance remains to be determined. In our prospective study, we compared the detection rate and the prognostic value of these two circulating biomarkers in patients with metastatic uveal melanoma. GNAQ=GNA11 mutations were characterized in archived tumor tissue. Using a highly sensitive and mutation-specific bidirectional pyrophosphorolysisactivated polymerization (bi-PAP) technique, GNAQ c.626A>T, GNAQ c.626A>C and GNA11 c.626A>T copy numbers were quantified in plasma from 12 mL of blood. CTCs were detected at the same time in 7.5 mL of blood by the CellSearchV R technique. Patient characteristics and outcome were prospectively collected. CTCs (1) were detected in 12 of the 40 included patients (30%, range 1-20). Among the 26 patients with known detectable mutations, ctDNA was detected and quantified in 22 (84%, range 4-11,421 copies=mL). CTC count and ctDNA levels were associated with the presence of miliary hepatic metastasis (p 5 0.004 and 0.03, respectively), with metastasis volume (p 5 0.005 and 0.004) and with each other (p < 0.0001). CTC count and ctDNA levels were both strongly associated with progression-free survival (p 5 0.003 and 0.001) and overall survival (p 5 0.0009 and <0.0001). In multivariate analyses, ctDNA appeared to be a better prognostic marker than CTC. In conclusion, ctDNA and CTC are correlated and both have poor prognostic significance. CTC detection can be performed in every patient but, in patients with detectable mutations, ctDNA was more frequently detected than CTC and has possibly more prognostic value.Uveal melanoma is a rare cancer, with a reported incidence of two to eight new cases per million per year in Europe 1 and the United States.2 Specific mutations are found in this particular type of melanoma: >80% of uveal melanoma express mutually exclusive somatic mutations in two paralog proto-oncogenes, GNAQ 3 and GNA11, 4 which encode a-subunits of heterotrimeric G-proteins involved in the MEK-ERK signaling pathway. 5 In both genes, most mutations occur at nucleotide 626 encoding a glutamine at codon 209 (Q209). The presence of these mutations does not influence the risk of metastasis in patients. 4 Uveal melanoma metastases develop mostly in the liver through hematogenous spread of cancer cells. Despite improvement of diagnosis and treatment of the primary eye tumor, there is no effective treatment of metastatic disease; the prognosis of patients with metastatic uveal melanoma is limited, and although a few patients experience extended survival, the median overall survival (OS) following metastases detection is less than 1 year. 6 Circulating tumor cells (CTCs), which are cancer cells detected in patient blood, may correspond to cancer "seeds" that initiate metastatic relapse.7 Over the past two decades, both molecular and cytological detection techniques have