Imaging cell lineage with a synthetic digital recording system

Introduction

Cell lineage research is experiencing a renaissance. Already in the late 19th century researchers were interested in describing and comparing cell lineages of different organisms (Wilson, 1898). For many decades, however, the investigation of cell lineages was limited by the lack of techniques that could be used to track the cell division of complex organisms. That has changed now. In recent years, newly developed technologies promise to finally allow the reconstruction of the lineage history of a complex organism.
A popular approach in recent years has been to use CRISPR-based editing systems that generate heritable mutations during the development of an organism (e.g. McKenna et al, 2016). The mutations inside the cells of such organisms are usually recovered by sequencing, which leads to the loss of the spatial context of the cells. A major goal of current research is, therefore, to describe the lineage history of the cells together with their identities (cell types) and spatial distribution.

About the preprint

Chow, Budde and collaborators designed a new lineage recording system (intMEMOIR) that introduces mutations using serine integrases and use Fluorescence in situ Hybridisation (FISH) to read these mutations. Serine Integrases are bacteriophage enzymes that introduce mutations via site- specific recombination after they recognise and bind to attB and attPs sites.
The lineage recorder designed by the authors contains multiple recording units, each of which consists of a barcode sequence flanked by att sites (Fig 1). This configuration leads to either an inversion or a deletion of the barcode sequence after binding of the integrase. Therefore, each recording unit will end up in 1 of 3 possible states: “unmutated”, “inverted” or “deleted”. After a site is “inverted” or “deleted” it cannot further change and it’s inherited in subsequent cell divisions.
Importantly, each recording unit has a unique att site that prevents recombination between units. By linking each of these 3-state sites —or “trits”— to a strong promoter they could express 1) the forward barcode (“unmutated” state), 2) the reverse complement barcode (“inverted”) or 3) no transcript (“deleted”). These “trits” represent a considerable improvement from a previous approach that used 2-state units —or “bits”— (Frieda et al, 2016). This is because recorders made of 2-state units would get to the same end-state if they become completely saturated with mutations, making them uninformative.

Results

After validating the trit design in mouse embryonic stem cells, the authors engineered a cell line (intMEM1) with a recorder array of 10 trits (Fig 2) and where the recording activity can be activated in a controlled manner by the addition of doxycycline and trimethoprim (TMP) to the cell culture. The authors then devised a clever approach to decode the recorders’ mutations by five rounds of hybridisation chain reaction smFISH, using 4 fluorophores and 20 unique probes. Essentially, this approach leads to cells emitting up to 4 colours per round of hybridisation depending on the state of its lineage recorder and on the orientation of the transcribed barcode. The colours/states of all cells in the culture are then used to reconstruct their lineage relationship.

 

In order to asses the accuracy of reconstructed lineages, the authors compared these to ground-truth lineages obtained after tracking cell divisions on time-lapse movies and also made use of computer simulations. The ground truth lineages were based on cultures that underwent recording activity for ~3 generations and ~1-2 generations of clonal expansion. They assessed lineage reconstruction at two different levels: “clonal classification” and “tree reconstruction”.
For “clonal classification” cells that share the same exact barcode were grouped into “clones” and the agreement between these “clones” with the ground truth lineage were assessed. For “tree reconstruction”, the relationship between “clones” was inferred by reconstructing a lineage tree using a maximum likelihood approach incorporating empirical parameters. The accuracy of this tree was determined as the fraction of splits that is shared between the reconstructed and the ground truth trees (Robinson-Foulds metric). Importantly, these two levels of reconstruction (clones vs tree) can be used either together or individually, depending on the biological question. As the authors also show, clonal classification can be useful to analyse the clones’ spatial distribution in a tissue/organ without the necessity of reconstructing a completely accurate lineage tree – a task which remains difficult (Salvador-Martinez, et al 2019).
Their results showed a high accuracy of reconstruction at both levels, with an average of 85% of the cells inside a reconstructed clone being correctly assigned to it (“clone classification”) and 25% of clonal relationships reconstructed perfectly, with the overall reconstruction score significantly higher than the random control (“tree reconstruction”).

Finally, to test their recorder in vivo, they created a Drosophila melanogaster line (termed “Drosophila memoiphila”) with the intMEMOIR construct, where the integrase activity can be regulated via a heat-shock inducible promoter. They used this line to investigate clonal relationships in the fly brain. For this, they induced mutations during early development by heat shock, letting the flies develop until adults. Mutated adult flies were then collected and their brain sectioned for smFISH imaging. As the same imaging approach could be used to read transcripts from endogenous genes, the authors decided to read, in top of the lineage recorder units, 8 genes that could discern between different neuronal cell-types. The combination of gene expression, lineage information, and spatial location within the brain allowed them to identify different known cell types (e.g. GABAergic and dopaminergic neurons) and interrogate how their cell type correlated with their cell lineage. They found that cells of the same clone showed stronger cell type similarity to each other the more physically closer they were. Importantly, this relationship between physical distance and cell type similarity was not observed when comparing cells of different clones.

Why I chose this preprint

With this exciting novel approach, Chow, Budde, and collaborators remind us again about the importance of knowing the lineage history of the cells in order to have a better understanding of their differentiation within their native spatial context, and introduce a new fly line “Drosophila memoiphila” than can be used as a general purpose lineage analysis resource. By doing so, they prove that cell lineage research is alive, rejuvenated and has a promising future.

Question to the Authors:

Q1: One exciting possibility that you suggest is to increase the lineage information that can be read by the incorporation of additional arrays. Is there a limit on the number of FISH readout cycles that can be performed on a slide?
Q2: I find very interesting how the Shannon’s entropy of a mutated array relates to its accuracy of reconstruction. Could the Shannon’s entropy measure serve to determine a reliability threshold for mutated arrays?
Q3: How easy is to differentiate between a “deleted” state from the possibility of having another state but a lack of signal?

 

References

Wilson, E.B. CONSIDERATIONS ON CELL‐LINEAGE AND ANCESTRAL REMINISCENCE. Annals of the New York Academy of Sciences, 11: 1-27 (1898). doi:10.1111/j.1749-6632.1898.tb54960.x

McKenna A, Findlay GM, Gagnon JA, Horwitz MS, Schier AF, Shendure J. Whole-organism lineage tracing by combinatorial and cumulative genome editing. Science (2016). https://doi.org/10.1126/science.aaf7907

Frieda KL, Linton JM, Hormoz S, et al. Synthetic recording and in situ readout of lineage information in single cells. Nature. (2016). https://doi.org/10.1038/nature20777

Salvador-Martínez I, Grillo M, Averof M, Telford MJ. Is it possible to reconstruct an accurate cell lineage using CRISPR recorders? eLife. 2019;8. https://doi.org/10.7554/eLife.40292.001

Tags: cell lineage, lineage recorders

doi: https://doi.org/10.1242/prelights.18724

Read preprint (No Ratings Yet)