Determining the developmental potential of stem cells is interesting not only in the context of organismal development, but also under steady-state conditions in adult stem cell contexts such as the hematopoietic system. Conceptually, different strategies are available to study this potential. Fluorescent-reporter-based systems (such as Brainbow or Confetti) and molecular barcoding approaches have been harnessed for lineage tracing.

Brainbow and Confetti have limitations, as only few reporters are available, and are mainly useful for tissues that do not exhibit much cell migration. On the other hand, molecular barcoding based on viral insertion sites or virally delivered barcodes provides access to many barcodes but typically requires the cells to be manipulated ex vivo. In addition, viral or transposon insertion sites may lead to adverse effects in cells.

To overcome these limitations, Hans-Reimer Rodewald from the DKFZ in Heidelberg and his collaborators developed the Polylox system. The Polylox locus consists of an array of unique DNA sequences that are interspersed by loxP sites. Barcodes are created in vivo through Cre-dependent recombination.

Polylox is a recombination-based fate-mapping tool. Credit: Figure adapted from Pei, W. et al., Nature Research.

“In the beginning, nobody believed that this would ever work because Cre was thought to completely react on all available loxP sites on such a locus,” says Rodewald. Fortuitously, the team's instinct was correct, and they found that Cre-mediated recombination at the Polylox locus is incomplete when Cre is weakly or briefly expressed and can potentially lead to the generation of 1.8 million unique barcodes. However, the utility of individual barcodes varies according to their frequency. “You get some barcodes more frequently than others,” explains Rodewald. Thomas Höfer's group at the University of Heidelberg, together with Rodewald and colleagues, addressed this issue both computationally and experimentally. They determined that certain barcodes have a low probability of being generated, and these are typically not shared across different experiments and are thus highly informative in fate-mapping experiments.

Rodewald and colleagues applied Polylox fate mapping to the mouse hematopoietic system. “I had tried for many years to get closer to the physiology,” says Rodewald. He thinks that currently used assays involving ex vivo manipulation of hematopoietic stem cells and subsequent transplantation into irradiated mice are quite artificial. With “the transplantation approach, you empty the system...and then you have to start over again,” says Rodewald. Using the Polylox system, the fate of these stem cells can be studied under steady-state conditions, leading to different results from those in transplantation experiments. For instance, under steady-state conditions, the Rodewald and Höfer teams found in an earlier study that many stem cells contribute to the replenishment of the system, but individual contributions are small.

The Polylox system shares some features with the recently described CRISPR–Cas9-based lineage-tracing approaches, which also allow barcode generation in vivo. Yet Rodewald sees advantages in the Polylox system. Each of the individual DNA blocks is about 170 base pairs long, and it is necessary to sequence across the whole locus to obtain the full barcode information. “Even if you have poor sequencing or PCR problems or whatever, you will always recognize these blocks,” Rodewald points out. He thinks that the CRISPR–Cas9-based systems in which barcodes can differ in just single nucleotides may be more prone to PCR and sequencing artifacts than the Polylox system is.

While Rodewald and his team have applied the Polylox technology for fate mapping to the hematopoietic system, it should be possible to use this technology in other tissues or to answer other questions, as long as suitable Cre driver lines are available.