WILD-seq: Clonal deconvolution of transcriptomic signatures of sensitivity and resistance to cancer therapeutics in vivo.

Tumour heterogeneity is thought to be a major barrier to successful cancer treatment due to the presence of drug resistant clonal lineages. However, identifying the characteristics of such lineages that underpin resistance to therapy has remained challenging. Here we present WILD-seq; Wholistic Interrogation of Lineage Dynamics by sequencing, a platform that leverages expressed barcodes to simultaneously map clonal identities and transcriptional states at single cell resolution. Our optimised pipeline ensures recurrent representation of clonal lineages across animals and samples, facilitating analysis of clonal dynamics under perturbation. Application of WILD-seq to two triple negative mammary carcinoma mouse models, identified changes in clonal abundance, gene expression and microenvironment in response to JQ1 or taxane chemotherapy. WILD-seq reveals oxidative stress protection as a major mechanism of taxane resistance that renders our tumour models collaterally sensitive to non-essential amino acid deprivation. In summary, WILD-seq enables facile coupling of lineage and gene expression in vivo to elucidate clone-specific pathways of resistance to cancer therapies.

Somatic mosaicism in the mature brain reveals clonal cellular distributions during cortical development

The structure of the human neocortex underlies species-specific features and is a reflection of intricate developmental programs. Here we analyzed neocortical cellular lineages through a comprehensive assessment of brain somatic mosaicism—which acts as a neutral recorder of lineage history. We employed deep whole genome and variant sequencing in a single postmortem neurotypical human brain across 25 anatomic regions and three distinct modalities: bulk geographies, sorted cell types, and single nuclei. We identified 259 mosaic variants, revealing remarkable differences in localization, clonal abundance, cell type specificity, and clade distribution. We identified a set of hierarchical cellular diffusion barriers, whereby the left-right axis separation of the neocortex occurs prior to anterior-posterior and dorsal-ventral axis separation. We also found that stochastic distribution is a driver of clonal dispersion, and that rules regarding cellular lineages and anatomical boundaries are often ignored. Our data provides a comprehensive analysis of brain somatic mosaicism across the human cerebral cortex, deconvolving clonal distributions and migration patterns in the human embryo.One Sentence SummaryComprehensive evaluation of brain somatic mosaicism in the adult human identifies rules governing cellular distribution during embryogenesis.

Machine Learning-Guided Prediction of Antigen-Reactive In Silico Clonotypes Based on Changes in Clonal Abundance through Bio-Panning

c-Met is a promising target in cancer therapy for its intrinsic oncogenic properties. However, there are currently no c-Met-specific inhibitors available in the clinic. Antibodies blocking the interaction with its only known ligand, hepatocyte growth factor, and/or inducing receptor internalization have been clinically tested. To explore other therapeutic antibody mechanisms like Fc-mediated effector function, bispecific T cell engagement, and chimeric antigen T cell receptors, a diverse panel of antibodies is essential. We prepared a chicken immune scFv library, performed four rounds of bio-panning, obtained 641 clones using a high-throughput clonal retrieval system (TrueRepertoireTM, TR), and found 149 antigen-reactive scFv clones. We also prepared phagemid DNA before the start of bio-panning (round 0) and, after each round of bio-panning (round 1–4), performed next-generation sequencing of these five sets of phagemid DNA, and identified 860,207 HCDR3 clonotypes and 443,292 LCDR3 clonotypes along with their clonal abundance data. We then established a TR data set consisting of antigen reactivity for scFv clones found in TR analysis and the clonal abundance of their HCDR3 and LCDR3 clonotypes in five sets of phagemid DNA. Using the TR data set, a random forest machine learning algorithm was trained to predict the binding properties of in silico HCDR3 and LCDR3 clonotypes. Subsequently, we synthesized 40 HCDR3 and 40 LCDR3 clonotypes predicted to be antigen reactive (AR) and constructed a phage-displayed scFv library called the AR library. In parallel, we also prepared an antigen non-reactive (NR) library using 10 HCDR3 and 10 LCDR3 clonotypes predicted to be NR. After a single round of bio-panning, we screened 96 randomly-selected phage clones from the AR library and found out 14 AR scFv clones consisting of 5 HCDR3 and 11 LCDR3 AR clonotypes. We also screened 96 randomly-selected phage clones from the NR library, but did not identify any AR clones. In summary, machine learning algorithms can provide a method for identifying AR antibodies, which allows for the characterization of diverse antibody libraries inaccessible by traditional methods.

Small Subclones Harboring NOTCH1, SF3B1 or BIRC3 Mutations Are Clinically Irrelevant in Chronic Lymphocytic Leukemia

Introduction. Ultra-deep next generation sequencing (NGS) allows sensitive detection of mutations and estimation of their clonal abundance in tumor cell populations. TP53 mutations identified by ultra-deep NGS significantly impact on survival of chronic lymphocytic leukemia (CLL) patients, independent of their representation in the tumor clone and even if restricted to a small fraction of leukemic cells. Purpose. Here we aim at assessing the frequency, prognostic impact and evolution during disease course of NOTCH1, SF3B1 and BIRC3 mutations identified by ultra-deep NGS in newly diagnosed CLL. Methods. The study was based on a consecutive series of 304 newly diagnosed and previously untreated CLL (median age: 71 years; Binet A/B/C: 79/12/9%; unmutated IGHV genes: 35%; del13q/+12/del11q/del17p: 51/21/8/9%; median follow-up: 8.1 years). By ultra-deep NGS, TP53 mutations (exons 4-8) occurred in 15% of CLL with a median clonal abundance of 18% (range 0.2-95%). NOTCH1 (hg19 g.chr9:139390596-139390829), SF3B1 (exons 14, 15) and BIRC3 (exons 6, 9) mutations were screened on peripheral blood samples by amplicon-based ultra-deep NGS (454 Life Sciences) (average depth: 2437). A bioinformatic algorithm was applied to call non-silent variants out of background NGS noise. Variant calling by the algorithm was validated by Sanger sequencing or, if the variant was below the sensitivity threshold of Sanger sequencing, by both duplicate ultra-deep NGS and allele specific PCR (AS-PCR). Variant allele frequency (VAF) was corrected for tumor representation. Results. Ultra-deep NGS identified 46 NOTCH1 mutations (median VAF 24%; range: 1.4-64%) in 14% (43/304) CLL, 43 SF3B1 mutations (median VAF 16%; range: 0.5-48%) in 11% (35/304) CLL and 37 BIRC3 mutations (median VAF 5.6%; range: 0.2-47%) in 8% (26/304) CLL. In cases harboring more than one mutation (NOTCH1=3; SF3B1=2; BIRC3=5), the variants mapped on distinct sequencing reads from the same amplicon suggesting that they belonged to different CLL subclones. Out of the variants identified by ultra-deep NGS, a significant fraction of NOTCH1 (n=14; median VAF: 3.9%; range: 1.4-9%), SF3B1 (n=25; median VAF: 4.3%; range: 0.5-17%) and BIRC3 (n=30; median VAF: 1.4%; range: 0.2-6%) mutations were missed by Sanger sequencing because their VAF was below the sensitivity threshold of the assay (~10%). The molecular spectrum of these small subclonal mutations was highly consistent with that of mutations that were visible by Sanger sequencing. The overall survival of cases harboring solely small subclonal mutations of NOTCH1, SF3B1 and BIRC3 was similar to that of wild type cases and longer than that of cases with clonal lesions (Fig. 1). Outcome-driven statistical approaches (maxstat and recursive partitioning) were applied to identify the optimal cut-off of the VAF in order to test whether the clonal representation of mutations correlated with CLL survival. Application of these approaches to TP53 mutations failed in identifying a clear cut-off, suggesting that TP53 variants are relevant to the outcome of CLL independent of their VAF. Conversely, in the case of NOTCH1, SF3B1 and BIRC3 mutations, these approaches consistently identified 25%, 35% and 1%, respectively, as the best VAF cut-off values for outcome prediction (Fig. 1), thus suggesting that small subclones harboring NOTCH1, SF3B1 or BIRC3 mutations are clinically irrelevant. By multivariate analysis, NOTCH1 mutations >25% of VAF (HR: 1.8; p=.038), SF3B1 mutations >35% of VAF (HR: 2.9; p=.005) and BIRC3 mutations >1% of VAF (HR: 1.8; p=.044) were significantly associated with an increase in the hazard of death, after adjusting for TP53 lesions (HR: 2.9; p<.001), del11q (HR: 2.2; p=.007), +12 (HR: 1.3, p=.174) and del13q (HR: 0.8; p=.512). Among cases harboring small mutated subclones, longitudinal deep NGS of sequential samples documented the outgrowth of the variants to a clonal level in 1/5 cases for NOTCH1, in 5/12 for SF3B1 and in 3/10 for BIRC3. While selection of NOTCH1 and BIRC3 variants did not associate neither with disease feature nor with treatment, the outgrowth of subclonal SF3B1 mutations was restricted to cases showing unmutated IGHV genes (p=.015). Conclusions. Small mutated subclones of the NOTCH1, SF3B1 and BIRC3 genes appear to be clinically irrelevant. Mutations of NOTCH1 and SF3B1 require a substantial clonal representation to be harmful. Figure 1 Figure 1. Disclosures No relevant conflicts of interest to declare.