Stochastic Labelling and Clonal Analysis
How single neurons and clones are picked out of a driver’s whole expression pattern — FLP-out, MARCM and MCFO — and what that means for the single-neuron images on VFB.
A driver line labels a population. Most of the interesting anatomy is in the individual neurons inside it, and a confocal stack of fifty overlapping cells will not give you the shape of any one of them. Stochastic labelling solves this by making the labelling event itself random and rare, so that a small, resolvable subset of the population lights up in each animal.
This is where a large fraction of the single-neuron light microscopy on VFB comes from, and it is what makes LM morphologies comparable against EM reconstructions by NBLAST.
The mechanism
All of these methods use site-specific recombination. Golic and Lindquist moved the yeast FLP recombinase and its FRT target sites into Drosophila, put FLP under hsp70 control, and flanked a white gene with FRTs: heat shock triggered recombination, read out as white patches in the eye, with the frequency of events varying with the severity of the heat shock and the pattern of mosaicism with the stage at which it was applied (Golic and Lindquist, 1989). That is the whole basis of the methods below — give FLP only a brief window and it acts in some cells and not others. The randomness is the point.
FLP-out. A transcription-terminating cassette flanked by FRT sites sits between the promoter and the reporter. The reporter is silent until FLP removes the cassette. Because excision is stochastic, only some cells in the driver’s pattern ever express it.
MARCM (Mosaic Analysis with a Repressible Cell Marker) puts a dominant repressor of a cell marker — GAL80 — in trans to the mutant gene of interest. Mitotic recombination between homologous chromosomes generates homozygous mutant cells, and those cells are “exclusively labeled due to loss of the repressor” (Lee and Luo, 1999). Because the event happens at a cell division, the labelled unit is a clone: Lee and Luo used it to visualise both “large neuroblast clones and single neuron clones” with membrane-targeted GFP. That is why MARCM is the standard method for lineage work, and why VFB holds clonal as well as single-cell images.
MCFO (MultiColor FlpOut) extends FLP-out with several differently epitope-tagged, membrane-targeted reporters, each behind its own excisable cassette. Different cells end up with different tag combinations, so neighbouring neurons are separable by colour as well as by sparseness (Nern et al., 2015). Using two recombinases lets the number of labelled cells and the number of colour combinations be tuned independently. Most of the Janelia single-neuron imagery on VFB is MCFO.
Reading a stochastic image on VFB
- The cell was chosen by chance, not by identity. A single-neuron image tells you the morphology of one cell that happened to be labelled within a driver’s pattern. Which cell type it is, is a downstream judgement — see cell types.
- Sparse does not mean single. An image may contain two or three faintly overlapping cells. VFB records what was segmented; check the image before treating a morphology as one neuron.
- A clone is a developmental unit, not a cell type. MARCM clones group neurons by shared lineage. Neurons in one clone may belong to several types, and one type may be split across clones.
- Coverage is biased by the driver. You can only label stochastically within a pattern that the driver already produces, so cell types with no good driver are under-represented in the LM data regardless of how sparse the labelling is.
- It is still one animal. As with everything else on VFB, comparison across animals is possible because the images are registered to a common template.
Where it fits
| Question | Method | On VFB |
|---|---|---|
| What does this driver label? | Whole-pattern imaging | Expression patterns |
| Can I get a driver for just this cell type? | Split-GAL4 intersection | Split driver expression |
| What does one of these cells look like? | FLP-out, MCFO | This page |
| Which cells share a lineage? | MARCM clones | This page |
| Is this LM neuron the same as that EM neuron? | Morphological comparison | NBLAST |
Sources
- Golic KG, Lindquist S (1989) The FLP recombinase of yeast catalyzes site-specific recombination in the Drosophila genome. Cell 59:499–509. doi:10.1016/0092-8674(89)90033-0
- Lee T, Luo L (1999) Mosaic analysis with a repressible cell marker for studies of gene function in neuronal morphogenesis. Neuron 22:451–461. doi:10.1016/S0896-6273(00)80701-1
- Nern A, Pfeiffer BD, Rubin GM (2015) Optimized tools for multicolor stochastic labeling reveal diverse stereotyped cell arrangements in the fly visual system. PNAS 112:E2967–E2976. doi:10.1073/pnas.1506763112
- Costa M et al. (2016) NBLAST: rapid, sensitive comparison of neuronal structure and construction of neuron family databases. Neuron 91:293–311. doi:10.1016/j.neuron.2016.06.012