Cre‐driver lines used for genetic fate mapping of neural crest cells in the mouse: An overview

Summary The neural crest is one of the embryonic structures with the broadest developmental potential in vertebrates. Morphologically, neural crest cells emerge during neurulation in the dorsal folds of the neural tube before undergoing an epithelial‐to‐mesenchymal transition (EMT), delaminating from the neural tube, and migrating to multiple sites in the growing embryo. Neural crest cells generate cell types as diverse as peripheral neurons and glia, melanocytes, and so‐called mesectodermal derivatives that include craniofacial bone and cartilage and smooth muscle cells in cardiovascular structures. In mice, the fate of neural crest cells has been determined mainly by means of transgenesis and genome editing technologies. The most frequently used method relies on the Cre‐loxP system, in which expression of Cre‐recombinase in neural crest cells or their derivatives genetically enables the expression of a Cre‐reporter allele, thus permanently marking neural crest‐derived cells. Here, we provide an overview of the Cre‐driver lines used in the field and discuss to what extent these lines allow precise neural crest stage and lineage‐specific fate mapping.


| A B R I E F S Y N O P S I S O F M E T H O D S U S E D F O R NE U R AL CR E S T C E LL F A T E M A P P I N G
The astonishing variety of neural crest derivatives has long been recognized, mainly through the pioneering work by Le Douarin and coworkers who used interspecies transplantation in avian embryos to determine normal fates as well as the developmental potential of neural crest cell populations from different axial levels of the embryo (Bronner & Simões-Costa, 2016;Le Douarin & Dupin, 2003). Together with assays involving dye labeling and retroviral infection of neural crest cells, this approach led to the establishment of comprehensive fate maps and revealed that certain neural crest derivatives (such as peripheral glia, sensory neurons, and melanocytes) are produced from all axial levels, whereas other neural crest cell lineages (such as cartilage, bone, smooth muscle, connective tissue, endocrine cells as well as, for instance, neurons and glia from the enteric and parasympathetic nervous system) originate from discrete levels along the neuraxis.
Given the limited accessibility of mammalian embryos, lineagetracing experiments as performed in avian embryos have not been widely used for fate mapping of mammalian neural crest cells. Rather, various genetic tools have been established that allow noninvasive and long-term tracing of neural crest cells in mice in vivo (Zurkirchen & Sommer, 2017). By far the most frequently used approach in the field is Cre-loxP-based conditional genetic recombination that, when combined with a Cre-reporter line, results in inheritable and irreversible expression of a marker gene in Cre-recombinase-expressing cells and in all of their progeny (Woodworth, Girskis, & Walsh, 2017). Furthermore, inducible forms of Cre-recombinase have been applied to perform stage-dependent fate mapping of neural crest cells and their derivatives or fate mapping at low recombination density for in vivo single cell tracing (Baggiolini et al., 2015;Kaucka et al., 2016). Apart from confirm- establishment of novel lineage trees (revealing, in particular, the broad developmental potential of peripheral glial cells), and to the demonstration of in vivo multipotency of single premigratory and migratory neural crest cells. While these studies have recently been covered elsewhere (Petersen & Adameyko, 2017;Zurkirchen & Sommer, 2017), in the present review we aim to focus on the tool set used in the field, summarizing the findings made with and discussing specific properties of various Cre lines that have been utilized to trace neural crest cells.
A well-established Cre-driver line for neural crest lineage tracing is Wnt1-Cre (Danielian, Muccino, Rowitch, Michael, & McMahon, 1998) ( Table 1). This transgenic mouse line expresses Cre initially in the midbrain and, after closure of the neural tube, in the midlines of the midbrain and the caudal diencephalon, in the midbrain-hindbrain junction, and in the dorsal spinal cord, where it recombines premigratory neural crest cells. By crossing Wnt1-Cre mice with the ROSA26 (R26R) Cre-reporter line (that drives b-galactosidase expression upon Cremediated recombination) (Soriano, 1999), it was shown that Wnt1-Cre is a highly efficient Cre-driver line, resulting in recombination of approximately 96% of all migratory neural crest cells (Hari et al., 2012).
Because Wnt1 is not expressed in migratory neural crest cells and Wnt activity rapidly decreases in neural crest cells after their delamination from the neural tube (Kl eber et al., 2005;Rabad an et al., 2016;Zervas, Millet, Ahn, & Joyner, 2004), it can be assumed that most neural crest cells are very efficiently targeted by Wnt1-Cre before or at the time of their delamination. Intriguingly, however, despite the early activity of Wnt1-Cre in the dorsal neural tube, recombination apparently occurs too late to allow investigation of mechanisms regulating epithelial-tomesenchymal transition (EMT) or delamination of neural crest cells.
Indeed, Wnt1-Cre-mediated ablation of signaling pathways shown in other animal models to be crucial for neural crest EMT, such as canonical Wnt signaling and signaling by TGFb superfamily factors, did not affect early stages of neural crest development (Brault et al., 2001;B€ uchmann-Møller et al., 2009;Hari et al., 2002;Jia et al., 2007). In contrast, the differentiation potential of neural crest cells along all axial levels could readily be monitored using Wnt1-Cre, both during development and at postnatal stages (Chai et al., 2000;Jiang, Rowitch, Soriano, McMahon, & Sucov, 2000;Zurkirchen & Sommer, 2017). A potential caveat of this line is, however, that, at least in the midbrain, Wnt1 is ectopically expressed from the Wnt1-Cre transgene, which could lead to ectopic activation of canonical Wnt signaling (Lewis, Vasudevan, O'neill, Soriano, & Bush, 2013). Although it is not known whether such ectopic Wnt1 expression also affects the neural crest, the use of a new driver line termed Wnt1-Cre2 should be considered (Lewis et al., 2013). In

TEC1
Tg(Tyr-cre)1Gfk Tonks et al. (2003) fact, in studies addressing the role of fibronectin in cardiac neural crest development, considerable phenotypic variances have been reported upon Wnt1-Cre vs. Wnt1-Cre2-mediated recombination, although this could have been due to differences between the two transgenic lines other than aberrant Wnt1 expression (Wang & Astrof, 2016).
Apart from the Cre/loxP system, another site-specific recombination system has also been established to trace the fate of neural crest cells. To this end, two transgenic mouse lines (termed Wnt1-Flpe mice) were independently generated that express Flp recombinase from the Wnt1 promoter (Dymecki & Tomasiewicz, 1998;Hatzistergos et al., 2015). Although the recombination efficiency and the extent of neural crest lineages traceable by these lines have not been described in detail, these lines were instrumental to perform intersectional lineage tracing of cells that concurrently express two distinct promoters. When combined with either the RC::FrePe (Engleka et al., 2012) or RC::Fela (Jensen et al., 2008) dual reporter alleles (which report dual Flp and Cre recombination), a fraction of cKit-CreER T2 -traced cardiac progenitors was shown to derive from the cardiac neural crest (traced by Wnt1-Flpe) (Hatzistergos et al., 2015). Likewise, intersectional fate-mapping with the RC::FrePe allele was used to demonstrate that Isl1 is not an exclusive marker for second heart field cardiac progenitors, as previously suggested, but also marks a subpopulation of cardiac neural crest cells (Engleka et al., 2012).
Another mouse line expressing Cre in the dorsal neural tube and premigratory neural crest is P3Pro-Cre, in which Cre expression is driven from a Pax3 promoter fragment .
Several Cre-driver lines have been generated that, unlike the Although P0 was originally identified as a marker of the peripheral glial lineage, the expression of P0-Cre in transgenic mice was not lineage restricted, but detected already in migratory neural crest cells (Yamauchi et al., 1999). Consequently, P0-Cre/R26R mice displayed b-galactosidase expression in multiple neural and nonneural tissues originating from the neural crest. However, at least in some structures such as the DRG, the recombination efficiency appeared to be consid- Thanks to this high recombination efficiency it was possible to address stage-specific functions of canonical Wnt signaling upon constitutive activation in premigratory neural crest (using Wnt1-Cre) vs. migratory neural crest cells (using Sox10-Cre). Furthermore, the Sox10-Cre line was used together with the Wnt1-Cre line to show an unexpected contribution of neural crest cells in the murine neck region to muscle connective tissue, cartilage and bone, including endochondral bones that were till then believed to exclusively originate from the mesoderm (Matsuoka et al., 2005). However, unlike Wnt1-Cre, Sox10-Cre activity leads to recombination of some adult structures that are not of neural crest origin, such as subpopulations of epithelial cells present in hair follicles ( Figure 1).

| NE URAL CREST LINEAGE TRACING U S I N G I N D U C I B L E C R E -D R I V E R LI N E S
The above-described Cre lines allow in vivo fate mapping or functional analysis of a gene of interest only at the first time the Cre-driving promoter is active. To temporally control Cre activity, inducible forms of Cre have been developed, for instance by fusion with a mutant ligand-binding domain of the human estrogen receptor (CreER) that bind to the synthetic estrogen receptor ligand 4-hydroxytamoxifen (4-OHT) or tamoxifen (TM), but not to endogenous estradiol (Feil et al., 1996). Subsequently, a refined version with a higher TM sensitivity was constructed (termed CreER T2 ) (Feil, Wagner, Metzger, & Chambon, 1997 (Baggiolini et al., 2015). In this study, low dose TM treatment of pregnant mice led to low density recombination of the multicolor Cre-reporter in neural crest cells homozygous for R26R-Confetti, resulting in clones of cells expressing either nuclear green, cytoplasmic yellow, cytoplasmic red, membrane-bound blue, or rare combinations thereof. Intriguingly, around 20% of all the clones derived The assumption that multipotent neural crest cells can self-renew during a given time window was further supported by clonal analysis of migratory neural crest cells, traced shortly after emigration by means of a Sox10-iCreER T2 driver line (Baggiolini et al., 2015;Simon, Lickert, G€ otz, & Dimou, 2012). By using R26R-Confetti as Cre-reporter in Sox10-iCreER T2 mice treated with low TM doses, the vast majority of migratory neural crest cells were shown to maintain multipotency, and the Sox10-iCreER T2 -expressing cells did not display any higher degree of fate restriction as compared to their premigratory counterparts (Baggiolini et al., 2015). The only structure labeled by Wnt1-CreER-but not Sox10-iCreER T2 -lineage tracing was the dorsal neural tube, confirming that Sox10-iCreER T2 is not expressed in the premigratory neural crest, but only once the cells start to migrate. R26R-Confetti-based clonal analysis of neural crest cells has also been performed with another, independently produced Sox10-iCreER T2 line (Kaucka et al., 2016;Laranjeira et al., 2011). In this study, composition and behavior of ectomesenchymal cranial neural crest-derived clones were monitored during early craniofacial development, revealing that cranial neural crest cellderived clones frequently comprise odontogenic, chondrogenic, osteogenic, and adipogenic cells.
Sox10 is not only expressed in migratory neural crest cells, but its expression is maintained throughout development and postnatally in the glial and melanocyte lineage, including in adult tissue (Bremer et al., 2011;Kuhlbrodt, Herbarth, Sock, Hermans-Borgmeyer, & Wegner, 1998;Shakhova et al., 2012). Moreover, it marks cells with neural crest stem cell (NCSC) features that have been isolated from various adult neural crest-derived structures (Shakhova & Sommer, 2010). Importantly, reporter expression in Sox10-iCreER T2 lines nicely matches the known Sox10 expression pattern during development and in the adult (Laranjeira et al., 2011;Simon et al., 2012). Thus, these Sox10-CreER T2 lines (Laranjeira et al., 2011;Simon et al., 2012) are ideally suited to monitor whether neural crest-derived cells give rise to distinct cell populations at different time points or to assess stage-specific roles of genes of interest. In fact, the Sox10-iCreER T2 line by Laranjeira and colleagues has been used to show that glia in the adult enteric nervous system are able to produce neurons upon injury (Laranjeira et al., 2011) or that cells in peripheral nerves can give rise to parasympathetic neurons, dental mesenchymal cells, or neuroendocrine chromaffin cells in the adrenal medulla Furlan et al., 2017;Kaukua et al., 2014). Such studies might also be possible with yet another independently generated Sox10-iCreER T2 line (McKenzie et al., 2014) or with a Sox10ER T2 CreER T2 line termed SECE (He & Soriano, 2015). Of note, while high dose TM treatment in SECE/R26R embryos resulted in labeling of multiple neural and nonneural neural crest derivatives, low dose application of TM was found to affect the reporter gene expression pattern, allowing the tracing specifically of cranial as opposed to trunk neural crest cells (He & Soriano, 2015).
Another genetic tool suitable for neural crest cell fate mapping at different developmental stages is a Plp-CreER T2 mouse line generated by Leone and colleagues (Leone et al., 2003). Plp is a glia-specific FIG URE 2 Sox10-iCreER T2 and Plp-CreER T2 -mediated tracing of adult neural crest-derived cells in the skin. Immunolabeling of adult skin of tamoxifen-free animals carrying inducible Sox10-iCreER T2 (Simon et al., 2012) and Plp-CreER T2 (Leone et al., 2003), respectively, reveals leakiness of these driver lines in combination with the Cre-reporter allele R26R-tdTomato (red) in melanocyte stem cells and glial cells present in or around the hair bulge, melanocytes (DCT (green); star-like cells in the schematics), and Schwann cells in nerve bundles (NB). NF marks axons of nerves. Hoechst nuclei counterstaining, scale bars 50 mm. Note that tamoxifen treatment of adult animals results in significantly enhanced CreER T2 activity in peripheral glia and melanocytes in both Sox10-iCreER T2 and Plp-CreER T2 lines (Parfejevs et al., 2018) DEBBACHE ET AL.  (Leone et al., 2003). Moreover, using this line, an unexpectedly broad developmental potential of peripheral glial cells was revealed, with Plp-CreER T2 -traced cells along peripheral nerves giving rise to cell types as diverse as melanocytes, parasympathetic neurons, mesenchymal cells in teeth, and chromaffin cells of the adrenal medulla (Adameyko et al., 2009;Dyachuk et al., 2014;Furlan et al., 2017;Kaukua et al., 2014;Petersen & Adameyko, 2017 (Furlan et al., 2017). Finally, the Plp-CreER T2 line (Leone et al., 2003) was instrumental to demonstrate by fate mapping that adult peripheral glia become activated upon skin wounding, detach from axons, and colonize the wound bed to support wound healing in a paracrine manner, without notable differentiation into other, nonglial cell types (Parfejevs et al., 2018).
However, inducible Cre activity in the Plp-CreER T2 line turned out not to be specific for the glial lineage (Hari et al., 2012;Leone et al., 2003). Indeed, in PLP-CreER T2 /R26R embryos, TM treatment at early stages of neural crest development (E9.5) resulted in prominent labeling of peripheral neurons, glia, and melanocytes, i.e., a fate map highly reminiscent of multipotent neural crest cells (Hari et al., 2012). The expression pattern of b-galactosidase became gradually restricted upon TM injection at later stages. However, at all stages examined, the melanocytic lineage was marked with considerable efficiency upon PLP-CreER T2 -mediated recombination. Likewise, induction of PLP-CreER T2driven recombination in adult mice marked a substantial fraction of skin melanocytes, independently of the Cre-reporter allele used (Parfejevs et al., 2018). Some of this expression might be due to CreER T2 leakiness during early neural crest development, given that in adult skin of both PLP-CreER T2 /R26R-tdTomato and Sox10-iCreER T2 /R26R-tdTomato mice, about 25% of all hair follicles contain recombined melanocytes even in the absence of any TM treatment (Figure 2) (Parfejevs and Sommer, unpublished). However, TM injection in the adult significantly induced Cre-reporter expression in both peripheral glia and melanocytes in the skin (Parfejevs et al., 2018), demonstrating persistent activity of PLP-CreER T2 and Sox10-iCreER T2 in these adult tissues.
Thus, as with Sox10-CreER T2 -expressing mice, different neural crest derivatives can be traced by means of the PLP-CreER T2 -line, albeit with an apparently lower recombination efficiency. Recently, Kaucka and colleagues made use of this feature to carry out clonal analysis of cranial neural crest cells in PLP-CreER T2 /R26R-Confetti embryos to confirm data obtained with a Sox10-iCreER T2 line (Kaucka et al., 2016).
In conclusion, distinct genetic mouse lines are available for fate mapping premigratory and migratory neural crest cells. Together with Cre-driver lines specific for fate-restricted precursor cells that we have not covered in the present review, there is an increasing tool set available to the community to study the neural crest lineage tree and the molecular mechanisms shaping it. The finding that several Cre and CreER T2 driver lines expected to exhibit lineage-specific expression appear to mark multipotent neural crest cells (although with quite divergent recombination efficiencies) could simply reflect unfaithful transgene expression. Alternatively, however, migratory neural crest cells as well as, for instance, cells in peripheral nerves might comprise distinct subpopulations expressing supposedly lineage-specific markers together with multipotency markers. Conceivably, such cell populations may be more or less ready to respond to the activity of cues controlling fate decisions during development or upon injury. To address such issues, the genetic approaches for prospective lineage tracing of neural crest cells described herein will have to be complemented with other methods, notably including retrospective lineage tracing by single cell transcriptome analysis.