Reconstruction of cell lineage trees in mice

Abstract

The cell lineage tree of a multicellular organism represents its history of cell divisions from the very first cell, the zygote. A new method for high-resolution reconstruction of parts of such cell lineage trees was recently developed based on phylogenetic analysis of somatic mutations accumulated during normal development of an organism. In this study we apply this method in mice to reconstruct the lineage trees of distinct cell types. We address for the first time basic questions in developmental biology of higher organisms, namely what is the correlation between the lineage relation among cells and their (1) function, (2) physical proximity and (3) anatomical proximity. We analyzed B-cells, kidney-, mesenchymal- and hematopoietic-stem cells, as well as satellite cells, which are adult skeletal muscle stem cells isolated from their niche on the muscle fibers (myofibers) from various skeletal muscles. Our results demonstrate that all analyzed cell types are intermingled in the lineage tree, indicating that none of these cell types are single exclusive clones. We also show a significant correlation between the physical proximity of satellite cells within muscles and their lineage. Furthermore, we show that satellite cells obtained from a single myofiber are significantly clustered in the lineage tree, reflecting their common developmental origin. Lineage analysis based on somatic mutations enables performing high resolution reconstruction of lineage trees in mice and humans, which can provide fundamental insights to many aspects of their development and tissue maintenance.

Figure 1. Cell lineage analysis based on somatic MS mutations.

A. The cell lineage tree of an organism is a rooted labeled binary tree representing its development from a single cell until present time. Nodes (circles) represent cells (dead cells are crossed), and edges (lines) connect parent and daughter cells (dashed lines represent several divisions). Uncrossed leaves (nodes with no daughters) represent extant cells. Lineage trees are a snapshot of a specific timepoint, which constantly grow throughout embryonic development and the adult life of the organism. Any cell sample (colored leaves) induces a partial subtree, called the cell sample lineage tree (panel B). B. The genomic composition of cells in the sample tree at three MS loci is shown. Spontaneous somatic mutations (marked by X) in these loci are sufficient to encode the lineage relations between these cells. Cells that are genetically similar tend to share a longer common developmental path, consequently enabling phylogenetic analysis to reconstruct the tree. Although idealized mutations are depicted, even potential hampering factors (such as coincident mutations) are not expected to disrupt analysis if a sufficient number of loci are analyzed . C. Scheme of reconstruction procedure. DNA from isolated cells is amplified either by culture or whole genome amplification, after which each cell is analyzed over a large set of MS loci using PCR and capillary electrophoresis. Capillary histograms are automatically analyzed yielding cell identifiers, which represent the mutations each cell accumulated in the MS set. Phylogenetic analysis of cell identifiers yields a reconstructed tree, which is an estimation of the lineage relations between analyzed cells.

Figure 2. Lineage reconstruction of ML2 and ML4.

Reconstructed trees of 26 cells of ML2 (10 week old mouse) and 25 cells of ML4 (13 week old mouse) are shown. Terminal nodes represent single analyzed cells (colors are used for different cell types), while intermediate nodes (uncolored) represent hypothesized precursor cells. Reconstruction was performed using the Neighbor-Joining algorithm (with the distance function ‘Absolute-distance’, see Methods). The vertical axis represents number of cell divisions, and the estimated depth for each cell is obtained by projecting it to this axis. These estimates were obtained by a method we developed, which is based on the correlation between genetic distance and cell depth . Reconstructed trees show that none of the analyzed cell types is completely and solely clustered on a subtree, demonstrating that none of cell types is a single exclusive clones.

Figure 3. Lineage reconstruction of ML7.

Reconstructed tree of 67 cells of ML7 (5.5 week old mouse) is shown (see Figure 2 for explanation of reconstructed trees). This experiment mainly focused on satellite cells (muscle stem cells), which were isolated from various muscles and myofibers from the mouse body. Although no myofiber is a single exclusive clone, we show that satellite cells from the same myofiber have a significantly larger lineage measure, reflecting that they tend to share a longer common developmental path (see Figure 4F).

Figure 4. Lineage analysis of satellite cells.

A muscle fiber from the Gastrocnemius muscle of ML4 was stained for DAPI (blue, all nuclei within the fiber) and Pax7 (pink, satellite cells). A. DAPI staining. B. Pax7 staining showing two satellite cells. C. Overlay of DAPI and Pax7. D. Single satellite cells were isolated from muscle fibers and cultured ex vivo, creating clones whose DNA approximates the DNA of the founder cell. Part of a Petri dish containing a single cell culture is shown. Myofiber structures are evident also in culture. E. Satellite cells can be categorized based on a hierarchy of body-side, muscle and myofiber. Myofibers are numbered from most superficial (#1) until deepest in the muscle. F. The average MRCA depth between pairs of satellite cells from the same myofiber/muscle/body-side was compared to the average MRCA depth between all pairs of cells. Pairs of satellite cells from the same myofiber had on average a significantly larger MRCA depth than a random pair of cells (p<0.0001) demonstrating that myofiber is a significant lineage unit. G. Correlation between lineage and physical proximity in the two main muscles of ML7. The physical distance between myofibers was defined as the difference between myofiber indices (e.g. the distance between a satellite cell on myofiber #3 and a satellite cell on myofiber #1 is 2). Physical distance and MRCA depth are significantly anti-correlated (p = 0.02 and p = 0.0001 for Gastrocnemius left and right, respectively; values stand for anti-correlation coefficients), demonstrating that satellite cells which are physically close to each other tend to share a common developmental path.