Massively parallel single cell lineage tracing using CRISPR/Cas9 induced genetic scars

Abstract

2 equal contributions 3 correspondence: janphilipp.junker@mdc-berlin. de (JPJ) A key goal of developmental biology is to understand how a single cell transforms into a full-grown organism consisting of many different cell types. Single-cell RNAsequencing (scRNA-seq) has become a widely-used method due to its ability to identify all cell types in a tissue or organ in a systematic manner 1-3 . However, a major challenge is to organize the resulting taxonomy of cell types into lineage trees revealing the developmental origin of cells. Here, we present a strategy for simultaneous lineage tracing and transcriptome profiling in thousands of single cells. By combining scRNAseq with computational analysis of lineage barcodes generated by genome editing of transgenic reporter genes, we reconstruct developmental lineage trees in zebrafish larvae and adult fish. In future analyses, LINNAEUS (LINeage tracing by Nuclease-Activated Editing of Ubiquitous Sequences) can be used as a systematic approach for identifying the lineage origin of novel cell types, or of known cell types under different conditions. Measuring lineage relationships between cell types is crucial for understanding fundamental mechanisms of cell differentiation in development and disease 4,5 . In early development and in adult systems with a constant turnover of cells, short-term lineage predictions can be computed directly on scRNA-seq data by ordering cells along pseudo-temporal trajectories according to transcriptome similarity 6-8 . However, for identifying the developmental origin of cells in the adult body, additional information is required. Genetically encoded fluorescent proteins are widely used as lineage markers 9,10 , but due to limited spectral resolution, optical lineage tracing methods have mostly been restricted to relatively small numbers of cells. Pioneering studies based on viral barcoding 11,12 , transposon integration sites 13 , microsatellite repeats 14 , somatic Author contributions JPJ, BS and BH conceived and designed the project. BH developed the experimental protocol, and BH and NM performed experiments. BS developed computational methods and analyzed the data. JPJ guided experiments and analysis. JPJ and BS wrote the manuscript, with input from all other authors. All authors discussed and interpreted results.

Is it possible to reconstruct an accurate cell lineage using CRISPR recorders?

“…As a solution to the problem of reading the mutations, several recent papers have explored the idea of using CRISPR-induced somatic mutations, targeted to artificial sequences inserted as transgenes into the genome (termed ‘CRISPR recorders’) (McKenna et al, 2016; Frieda et al, 2017; Junker et al, 2016; Kalhor et al, 2018; Perli et al, 2016; Alemany et al, 2018; Schmidt et al, 2017; Raj et al, 2018; Attardi et al, 2018; Spanjaard et al, 2018; Junker et al, 2016). The recorders consist of arrays of CRISPR target sites, targeted by their cognate sgRNAs and Cas9 during development.…”

“…Different designs of CRISPR recorders have been implemented, including recorders that register point mutations on arrays of barcoded targets (GESTALT; McKenna et al, 2016; Raj et al, 2018), ones that rely on ‘collapsing’ target arrays through deletions (MEMOIR; Frieda et al, 2017), recorders that target identical target sites located on separate transgenes (ScarTrace and LINNAEUS; Junker et al, 2016; Schmidt et al, 2017; Attardi et al, 2018; Alemany et al, 2018; Spanjaard et al, 2018) and ones that target the CRISPR sgRNA itself (Kalhor et al, 2018; Perli et al, 2016). In this work we have simulated the behaviour of the first two types of recorders, but the insights that we have gained should apply to all types of recorders.…”

Methods for lineage tracing on the organism-wide level

“…The Quake lab recently used a similar strategy in C. elegans [43]. These approaches have so far only been published for bulk material, but several manuscripts that report simultaneous lineage tracing and cell type identification in single cells from dissociated zebrafish have recently become available as preprints [46,47]. The MEMOIR technique [42] is based on a similar construct as in GESTALT, but with multiple integrations per cell.…”

“…Methods such as maximum parsimony [48] and neighbor-joining [49] have been used for GESTALT and MEMOIR, respectively, but it will certainly be possible to use other tree building methods such as maximum likelihood methods [50] or Bayesian inference [51] as well. For proper adaptation of these techniques, it will be important to find ways to implement experiment-specific factors such as differential scar probabilities in CRISPR/Cas9 lineage tracing and loss of information due to dropout events in single-cell sequencing [46].…”

Power in Numbers: Single-Cell RNA-Seq Strategies to Dissect Complex Tissues

“…When the readouts for the edited cassettes are transcribed, the technique is compatible with single-cell RNA-seq. For example, a method called LINNAEUS targeted an endogenous RFP reporter in zebrafish with CRISPR guides, whereupon both lineage and transcriptome information could be read out together in the single cell transcriptome profile (71).…”