|Home | About | Journals | Submit | Contact Us | Français|
A long-standing problem in development is understanding how progenitor cells transiently expressing genes contribute to complex anatomical and functional structures. In the developing nervous system an additional level of complexity arises when considering how cells of distinct lineages relate to newly established neural circuits. To address these problems, we used both cumulative marking with Cre/loxP and Genetic Inducible Fate Mapping (GIFM), which permanently and heritably marks small populations of progenitors and their descendants with fine temporal control using CreER/loxP. A key component used in both approaches is a conditional phenotyping allele that has the potential to be expressed in all cell types, but is quiescent because of a loxP flanked Stop sequence, which precedes a reporter allele. Upon recombination, the resulting phenotyping allele is ‘turned on’ and then constitutively expressed. Thus, the reporter functions as a high fidelity genetic lineage tracer in vivo. Currently there is an array of reporter alleles that can be used in marking strategies, but their recombination efficiency and applicability to a wide array of tissues has not been thoroughly described. To assess the recombination/marking potential of the reporters, we utilized CreERT under the control of a Wnt1 transgene (Wnt1-CreERT) as well as a cumulative, non-inducible En1Cre knock-in line in combination with three different reporters: R26R (LacZ reporter), Z/EG (EGFP reporter), and Tau-Lox-STOP-Lox-mGFP-IRES-NLS-LacZ (membrane-targeted GFP/nuclear LacZ reporter). We marked the Wnt1 lineage using each of the three reporters at embryonic day (E) 8.5 followed by analysis at E10.0, E12.5, and in the adult. We also compared cumulative marking of cells with a history of En1 expression at the same stages. We evaluated the reporters by whole-mount and section analysis and ascertained the strengths and weaknesses of each of the reporters. Comparative analysis with the reporters elucidated complexities of how the Wnt1 and En1 lineages contributed to developing embryos and axonal projection patterns of neurons derived from these lineages.
Physical fate mapping methods have contributed substantially to understanding developmental mechanisms, but these approaches vary based on methodical procedures, are invasive, and do not provide information on the genetic history of physically marked cells (reviewed in Zervas et al., 2005). By comparison, gene expression analysis alone is not sufficient for following the fate of cells during development because many developmentally regulated genes are rapidly extinguished and gene expression patterns are dynamic in varying populations. For example, the transcription factor Engrailed 1 (En1), which is expressed in the primordia of the Mb (mesencephalon, mes) and Cb (rhombomere 1, r1) is progressively restricted between E8.5 and E12.5 (Fig. 1A–D) (reviewed in Zervas et al., 2005). En1 is also expressed in the dermatome and proximal mesenchyme, costal precursors, limb bud ectoderm, and in the ventral spinal cord (SC) (Davidson et al., 1988). The secreted signaling molecule wingless-related MMTV integration site 1 (Wnt1), which is broadly expressed in the mes and migrating neural crest at E8.5, is dynamically regulated between E8.5 and E9.5 to become more restricted in the mes (reviewed in Zervas et al., 2005) and is also expressed in the posterior Hb and dorsal SC (Fig. 1E–1H) (Wilkinson et al., 1987). While northern blot analysis and in situ hybridization (ISH) provide a snapshot of En1 and Wnt1 gene expression patterns (Davidson et al., 1988; Joyner and Martin, 1987; Wilkinson et al., 1987), they are insufficient to reveal how cells expressing these genes contribute to developing and mature tissues.
It is possible to overcome these limitations using genetic methods (for example Cre/loxP technology), which forge a link between gene expression, cell behaviors, and the ultimate fate of cells (Dymecki and Kim, 2007; Joyner and Zervas, 2006). Cre recombinase (Cre) and conditional (loxP-Stop-loxP) phenotyping alleles (Fig. 1I, 1J) allow for the cumulative marking of cells in a tissue- or cell-type specific manner (Zinyk et al., 1998). The selectivity of marking is achieved by using gene regulatory elements to control the expression of Cre in genetically-defined domains (Supplemental Figure 1). Cre recognizes loxP sequences and mediates recombination in a site-specific manner, which results in the deletion of intervening sequences when the loxP sites are in the same orientation. In this system, Cre-mediated recombination occurs as long as the gene that is controlling Cre is expressed (Fig. 1K–1).
In contrast, GIFM allows progenitors of a defined genetic lineage to undergo recombination with both spatial and temporal control (Joyner and Zervas, 2006) (Fig.1K-2, K-3). Cells, having undergone recombination, are essentially uncoupled from gene expression and are permanently marked. The marked cells and their descendants can then be tracked at varying stages of development and can be detected in their final position in the adult (Dymecki and Kim, 2007; Joyner and Zervas, 2006; Zervas et al., 2005). GIFM is a tripartite system that uses the following components: Cre that is fused with a mutated estrogen receptor ligand binding domain (CreERT), a conditional loxP-Stop-loxP phenotyping allele, and tamoxifen. The CreERT modification ensures that Cre is sequestered in the cytoplasm by heat shock protein 90 (Hsp90) (Feil et al., 1996). Spatial control of CreERT, and therefore recombination, is achieved by placing CreERT under the control of gene regulatory elements. Tamoxifen, which takes approximately 6h to enter the embryo, binds to ER and releases CreERT from Hsp90. Subsequently, CreERT enters the nucleus where recombination occurs (Feil et al., 1996; Hayashi and McMahon, 2002; Robinson et al., 1991; Zervas et al., 2004). Thus, tamoxifen administration confers temporal control of when recombination is initiated while tamoxifen pharmacokinetics ensures that CreERT release, and therefore recombination, happens for 30h (Robinson et al., 1991) (Fig. 1K-2, K-3). CreERT plus tamoxifen concomitant with the presence of a conditional phenotyping allele causes the removal of the loxP flanked Stop sequence and brings the reporter of the phenotyping allele under the control of a constitutive promoter.
The most commonly used conditional reporter line of mice is the R26R line (Soriano, 1999), which consists of loxP-Stop-loxP-LacZ, and when recombined allows LacZ to be expressed and β-galactosidase (β-gal) to be produced (Soriano, 1999). This line has been utilized with an array of CreER driver lines in GIFM experiments to ascertain how the midbrain (Mb), cerebellum (Cb), and limb develop as well as to determine the presence of adult neural stem cells in vivo (Kimmel et al., 2000; Guo et al., 2003; Zervas et al., 2004; Ahn and Joyner, 2004, 2005). Additional reporters are the green fluorescent protein (GFP) phenotyping alleles, Z/EG and mGFP (Fig. 1I, 1J). The Z/EG allele is comprised of a loxP flanked β-geo (LacZ-neomycin) stop cassette upstream to GFP (Fig. 1I). In the absence of Cre, β-gal, but after recombination, LacZ is deleted and GFP comes under the control of the chick beta-actin promoter (Novak et al., 2000). The mGFP allele utilizes the Tau promoter to drive a loxP flanked Stop cassette and a 3’ modified membraneGFP-IRES-nuclearLacZ allele (Fig. 1J). In the absence of Cre, cells are unmarked. Cre-mediated recombination allows a bi-cistronic message of mGFP-IRES-nLacZ to independently produce β-gal, which is localized to the nucleus, and mGFP, which is enriched in axons (Hippenmeyer et al., 2005). A comprehensive comparison delineating the strengths and research value of these reporters has not been undertaken. In particular, the Z/EG (Novak et al., 2000) and mGFP (Hippenmeyer et al., 2005) reporter lines have not been evaluated side by side and in relation to the R26R line in a detailed multi-tissue and lineage-specific manner. In this report, we use three different phenotyping alleles in combination with En1Cre and Wnt1-CreERT mice to assess the recombination patterns in these reporter lines.
We fist compared En1 or Wnt1 expression to Cre expression by ISH to confirm that Cre, and therefore marking, would faithfully occur in appropriate domains (Supplemental Fig. 1). Cumulatively marked domains in sections from En1Cre embryos were compared to adjacent sections labeled with En1 and Cre probes (Supplemental Fig. 1A–J). Cells in the posterior Mb and Cb expressed En1 and Cre at E12.5 in domains that were similar to the regions observed with genetic marking (Supplemental Fig. 1C–E). Notably, neither En1 nor Cre were expressed in the trigeminal ganglia (Supplemental Fig. 1F–J). We also compared domains conditionally marked with Wnt1-CreERT and tamoxifen administration at embryonic day (E)8.5 to adjacent sections labeled with Wnt1 and Cre probes (Supplemental Fig. 1K–T). At E12.5, both Wnt1 and Cre were expressed in similar domains in the Mb (Supplemental Fig. 1K–O) and lateral posterior Hb (Supplemental Fig. 1P–Q), but neither were expressed in the trigeminal ganglia (Supplemental Fig. 1P–T). Because of the overlapping and segregated expression domains (Fig. 1), we directly compared similar regions (e.g. dorsal and ventral Mb) in cumulative and conditional marking conditions. We also took advantage of En1cre and Wnt1-CreERT mice to ascertain developmental aspects of different tissues in which marking has occurred.
We initially assessed mouse embryos at E12.5 because at this stage most tissues are comprised of both proliferating progenitors and newly differentiated cells. In addition, immature neural circuits are forming in the nervous system.
The R26R reporter allele is widely expressed throughout development and in the adult (Soriano, 1999). En1Cre;R26R embryos processed by whole-mount x-gal labeling (4h) revealed that cells with a history of expressing En1 reside in the entire dorsal and ventral Mb, Cb, ventral SC, and intercostal regions as well as in limb buds and craniofacial domains (Fig. 2A–D). The Z/EG reporter is also expressed in numerous domains, with different levels of marking occurring in a tissue-specific manner (Novak et al., 2000; Guo et al., 2002), which may depend on the promoter driving this transgene or the transgene site of integration in the genome. Marked (GFP+) domains in En1Cre;Z/EG embryos at E12.5 were readily detected by whole-mount fluorescent imaging (Fig. 2E–H). The Z/EG reporter revealed marked cells in the dorsal and ventral Mb, Cb, ventral SC, limb buds, and in the body wall lateral to the SC (Fig. 2E–H). Unlike with the R26R reporter, we could not detect marking in intercostal regions in whole-mount preparations with the Z/EG reporter (compare Fig. 2A and E). In contrast to the R26R reporter, faintly labeled axonal projections (GFP+) with the Z/EG reporter could be observed in craniofacial structures (Fig. 2E, inset) and could also be seen emanating from the SC (Fig. 2H, inset). Labeling in En1Cre;mGFP embryos at E12.5 (Fig. 2I–L) appeared substantially different from either En1Cre;R26R or En1Cre;Z/EG embryos. First, labeling was only seen in a subset of the Mb and Cb, as compared to R26R and Z/EG (see Figs. 2A, C, E, G, I, K). In addition, stronger labeling was observed in the SC, but was not detected in the body wall (compare Figs. 2A, D, E, H, I, L). Other differences included an absence of labeling in the limb bud (see Figs. 2I versus E) and very prominent labeling of neuronal projections and vibrissae in the craniofacial area (see Figs. 2J versus F). Collectively, whole-mount analysis revealed numerous similarities in the distribution of cell somata that were marked using the R26R and Z/EG phenotyping alleles while the mGFP reporter appeared to primarily label neuronal populations and produced GFP that was enriched in axonal projections (Fig. 2).
While whole-mount analysis is valuable for the 3-dimensional determination of spatial domains derived from marked cells and for the observation of axonal projections, the relative sensitivity of the detection method for each reporter confounds a direct comparison between the lines. In addition, whole-mount labeling does not allow for detailed cellular or cytoarchitectural analysis. Therefore, we sectioned cryoprotected embryos and processed them for either β-gal or GFP immunolabeling. Recombined (marked) cells in E12.5 En1Cre;R26R embryos were detected by β-gal immunocytochemistry labeling (Fig. 3). The cellular distribution of marked cells recapitulated the findings in whole-mount preparations. There was an abundance of labeled cells in the Mb, Cb, ventral Hb (Fig. 3A–D), SC, and intercostal domains (data not shown). Marked cells were also observed in craniofacial tissue that was adjacent to the more posteriorly positioned and labeled trigeminal ganglia (Fig. 3E). Marked cells overlapped with dopamine neurons in the ventral Mb as previously described (Fig. 3C) (Zervas et al., 2004).
In a complementary manner, E12.5 En1Cre;Z/EG embryos showed marked (GFP+) cell bodies in the dorsal and ventral Mb as well as in the Cb and v. Hb (Fig. 4A–D). In contrast to the R26R reporter, the Z/EG phenotyping allele allowed us to observe axonal projections of marked cells due to GFP distribution throughout the cell. In the ventral Mb, cells were distributed throughout the entire domain spanning the ventricular zone to the mesencephalic flexure and projections appeared to be located proximal to the flexure (Fig. 4C). To enhance our understanding of the projection patterns of cells with a history of expressing En1, we analyzed transverse sections obtained from various dorsal-ventral locations (Fig. 4G). In the dorsal Mb, axons were intermingled and difficult to discriminate amongst marked cells (Fig. 4G-1). This was also true in the ventral Mb (Fig. 4G-2). At the level of the Cb, a bundle of axons was observed to emanate caudally, consistent with sagittal sections (data not shown). We also observed En1Cre;Z/EG marked cells and projections in the trigeminal ganglia (Fig. 4E and F) and distributed throughout craniofacial tissue (Fig. 4F, data not shown). ISH with an anti-sense En1 probe showed that En1 gene expression was not detected in the trigeminal ganglia or in craniofacial tissue (Fig. 1D, Supplemental Fig. 1F–J) suggesting that the En1-derived cells were derived from migrating neural crest at an earlier stage of development.
In sagittal sections from En1Cre;Z/EG embryos, we observed axons emanating from the dorsolateral anterior Hb (Fig. 4E), which coursed toward the trigeminal ganglia (Fig. 4E and F). The projections de-fasciculated at an anterior point with a portion of the projections extending into the trigeminal ganglia and the remaining axons coursing caudally in a conduit that traverses the entire length of the Hb to the SC (Fig. 4E, arrows). In contrast, the ventromedial anterior Hb, which had marked cells confined to this domain by a posterior lineage boundary (Fig. 4D, arrowheads) (Zervas et al., 2004), had a broad swath of axons projecting from the marked domain to more caudal regions (Fig. 4D, bracket). A bilateral cluster of marked neurons was also prominently labeled in the ventral SC (Fig. 4G-3, arrow) with peripherally oriented axons (Fig. 4G-3, brackets). An additional lateral population of En1-Cre marked cells was located in distinct bands in the lateral body wall (Fig. 4G-3).
We next assessed cells marked with En1Cre and the mGFP allele at the cellular level (Fig. 5). An advantage of the mGFP allele is that we were able to detect marked cells with clarity because of strong nuclear β-gal expression, while mGFP expression was faintly, but reliably detected in cell bodies, and robustly present in axonal projections (Fig. 5A). In the dorsal Mb and in the Cb we detected small cohorts of β-gal+ cells organized into a narrow zone distal from the ventricle (Fig. 5B), which were apparently differentiating neurons located in a well-defined differentiated zone (DZ) (Fig. 5B). In contrast, R26R and Z/EG marked both differentiating cells and progenitors in the ventricular zone (VZ) (Fig. 3B, Fig. 4B). This was confirmed by comparing the genetically marked cells with phosphorylated histone H3 (pHH3) immunolabeling, which indicates mitotic cells (Fig. 6). Marked cells in En1Cre;mGFP embryos were not located in the proliferating pHH3+ VZ, but were positioned in the DZ in the Mb and Cb (Fig. 6A, B and H). In contrast, the En1 lineage marked in En1Cre;R26R and En1Cre;Z/EG embryos showed overlap with pHH3 in these regions (Fig. 6I and J). En1Cre;mGFP marked neurons located adjacent to the periphery of the dorsal Mb had short projections that extended toward the surface (Fig. 5B, inset). Transverse sections (planes shown in Fig. 5G) processed for β-gal/GFP double immunocytochemistry revealed that En1-derived cells of the dorsal Mb consisted of a small peripheral cluster of differentiating neurons that largely sent axons in a peripheral and rostral location (Fig. 5G-1). In the ventral Mb, the β-gal+ differentiating neurons (Fig. 5C, DZ) were co-localized with marked (GFP+) axons (Fig. 5C). Transverse sections revealed that the ventral Mb contained two bilateral clusters of marked neurons that had a complex projection pattern with axons that moved radially outward (Fig. 5G-2, arrowheads) although there was a conduit of axons that crossed the midline (Fig. 5G-2, bracket).
By directly comparing the ventromedial anterior Hb using the Z/EG versus mGFP alleles, we observed that marked cells that are confined to r1 by a well-defined posterior lineage boundary (Fig. 5D, arrowheads) (Zervas et al., 2004), have caudally oriented axons (Fig. 5D, brackets vs. 4D, brackets). Similarly, cumulatively marked cells of the lateral Hb were distributed throughout the DZ in the dorsal half of the lateral Hb and did not extend past a sharp border located dorsal to the flexure (Fig. 5F, yellow arrow). Projections from the lateral Hb coursed superficially where they then extended in a caudal direction (Fig. 5E, arrows). Upon closer inspection, this axonal bundle could be partitioned into multiple projection pathways including a deep, broad zone of caudally oriented projections, a dense superficial conduit, and a thick fascicle at the flexure of the Hb that connected with the trigeminal ganglia (Fig. 5F, see section 1.4, Neural Projections for details). En1-derived neurons of the trigeminal ganglia were β-gal+/GFP+ (Fig. 5F) and innervated the rostral craniofacial region (Fig. 5F). The marked cells in the trigeminal ganglia did not express En1 transcripts at E12.5 (Supplemental Fig. 1F–J) and were not mitotic based on the complete absence of pHH3 immunolabeling in all three reporter lines (Fig. 6E–G). This is the first report that we are aware of describing the contribution of the En1 lineage to the trigeminal ganglia. In contrast to En1Cre;Z/EG and En1Cre;R26R lines, which had marked cell bodies in the craniofacial regions, En1Cre;mGFP embryos did not have marked somata in rostral craniofacial tissue suggesting that the En1-derived cells in the craniofacial region were not differentiated neurons. In sagittal sections of SC, a ventral band of cells was observed that extended fine axonal processes ventrally to form a thick bundle of axons (data not shown). Transverse sections through the SC confirmed that the band of marked cells in sagittal sections was indeed two bilateral clusters of marked cells localized to the ventral-lateral SC (Fig. 5G-3). Unlike in En1Cre;Z/EG embryos, marked cells in E12.5 En1Cre;mGFP mice were not observed in the body wall adjacent to the SC (Fig. 5G-3 versus 4G-3).
In summary, both the R26R and Z/EG reporters revealed a similar profile of marked cells that had a history of expressing En1, which consisted of both proliferating and differentiating cells. The mGFP reporter allele yielded extremely bright labeling of neuronal processes and axonal trajectories by whole-mount epi-fluorescence at E12.5 with cumulative marking although cell somata were more difficult to observe at this level because GFP is not heavily distributed in cell bodies. We were unable to detect non-neural tissues with the mGFP reporter, likely because the promoter Tau, under which this GFP allele is controlled, is enriched in the nervous system (Aronov et al., 2001; Hashimoto et al., 2008).
We next tested the three reporter alleles in a GIFM experimental paradigm by breeding Wnt1-CreERT males also carrying our reporter alleles (Wnt1-CreERT;R26R, Wnt1-CreERT;mGFP, or Wnt1-CreERT;Z/EG) to Swiss Webster (wildtype) females and delivered tamoxifen by oral gavage to pregnant females carrying embryos at E8.5. We used our well-characterized Wnt1-CreERT mice because of the ability to elicit recombination in a wide array of tissue/cell types when tamoxifen is administered at E8.5 (Zervas et al., 2004). Tamoxifen pharmacokinetics confer that recombination occurs during a peak period of 12–24h after administration, following a 6h delay while tamoxifen enters the uterus (Robinson et al., 1991). We also took advantage of transient (Wnt1-CreERT) versus cumulative (En1Cre) marking to ascertain how these two lineages contributed to neural crest, Mb, and SC at E12.5. Wnt1-CreERT;R26R embryos marked by tamoxifen administration at E8.5 were analyzed by whole-mount embryo x-gal labeling and by β-gal immunocytochemistry on sections at E12.5 (Fig. 7). Wnt1-derived cells were detected in the Mb, choroid plexus, the dorsal Hb posterior to the Cb and in the SC (Fig. 7A). The Wnt1-derived cells in the SC appeared to be localized dorsally to the cells cumulatively marked in En1Cre;R26R embryos (compare Fig. 7A versus Fig. 2A). Control embryos (R26R plus tamoxifen or Wnt1-CreERT;R26R without tamoxifen) processed in x-gal were devoid of marking (Fig. 7A, inset; data not shown). In sections through the dorsal Mb, cohorts of cells were organized into linear clusters and had the appearance of clonally related cells (Fig. 7B). This organization was unique to medial/near-midline regions because neither dorsal-lateral Mb nor other tissue showed this same pattern (data not shown). While whole-mount x-gal labeling showed only faint labeling of craniofacial structures (Fig. 7A), section analysis revealed labeling through the craniofacial region (data not shown) and the trigeminal ganglia (Fig. 7C). In the SC, β-gal+ cells were localized to the dorsal root ganglia (DRG) (Fig. 7D).
In contrast to Wnt1-CreERT;R26R embryos, we could not detect marked cells by whole-mount GFP fluorescence in Wnt1-CreERT;Z/EG embryos at E12.5 (Fig. 8A). However, in sections from Wnt1-CreERT;Z/EG embryos we could sporadically detect GFP+ cells, and although marked cells were sparse, they were located in the Mb (Fig. 8B and E), posterior Hb (Fig. 8C), and DRG (Fig. 8D) consistent with Wnt1-CreERT;R26R tissue. Thus, the relatively few marked cells with the Z/EG reporter allowed for unambiguous and detailed cellular morphology to be discerned (Fig. 8B–E).
Whole-mount analysis of Wnt1-CreERT;mGFP embryos at E12.5 revealed distinctly marked domains including the Mb, posterior Hb, and SC (Fig. 9A and B). In Wnt1-CreERT;mGFP embryos, the dorsal Mb contained sparse, radially organized projections (Fig. 9D) that were predominantly located at the lateral edges of the Mb (Fig. 9E, arrows). In contrast, a dense plexus was observed in the ventral Mb (Fig. 9D). A tight fascicle of axons was also observed in the caudal Hb (posterior to the Cb), and in the presumptive dorsal SC (Fig. 9B and F). Interestingly, projections from the upper SC coursed laterally and innervated the body wall and proximal forelimb (Fig. 9A, inset; 9C). We validated the marked population by immunocytochemistry on sections with antibodies recognizing β-gal and GFP. Section analysis of the dorsal Mb revealed that differentiating neurons with elaborate morphology formed a superficial zone with their projections radiating outward toward the surface (Fig. 9G). The Wnt1 lineage also gave rise to neurons in the ventral Mb (Fig. 9H). Notably, β-gal+ cells (nuclei) were clustered together distally from the ventricle and had short projections that joined a dense plexus of projections, coursing both rostrally and caudally at the ventral mes flexure (Fig. 9H) suggesting that differentiating neurons were formed into distinct ventral nuclei similar to the organization of the adult ventral Mb (Paxinos, 2004) Wnt1-derived neurons also had axons in the ventral third of the posterior Hb (Fig. 9I). The Wnt1-derived neurons marked at E8.5 in the trigeminal ganglia were largely bipolar (Fig. 9J). This example is the first that we are aware of describing the temporal contribution of the Wnt1 lineage to the trigeminal ganglia. These observations coupled with the similar distribution and pattern of axonal projections as seen in cumulatively marked En1Cre;mGFP embryos (compare Fig. 9J to Fig, 5E and F) strongly suggest that En1 and Wnt1 derived neural crest cells contribute to similar craniofacial structures. Interestingly, cohorts of Wnt1-derived neurons in the SC were distributed in DRG (Fig. 9K). Axons of these individually marked neurons exited the DRG forming thick bundles that passed between the rib cartilage and innervated the body wall (Fig. 9K). In contrast, cumulatively marked En1-expressing neurons (with En1Cre;mGFP) did not contribute to the DRG. This is interesting because the DRG and trigeminal ganglia are both comprised of bipolar neurons connected to the CNS and periphery, but the DRG is derived from En1(−)/ Wnt1(+) precursors while the trigeminal ganglia is derived from En1(+)/ Wnt1(+) progenitors. This report is the first that we are aware of describing the temporal nature of Wnt1-expressing cells contributing to DRG and complements previously published cumulative marking with Wnt1-flipase (Dymecki and Tomasiewicz, 1998).
In summary, β-gal and GFP double immunofluorescent labeling on sections from embryos marked using GIFM and the mGFP phenotyping allele provided an advantage over the R26R or Z/EG reporters because of flexibility and clarity. By comparing the R26R, Z/EG and mGFP reporters at E12.5 with both cumulative marking and GIFM paradigms, we observed that along with distinguishing between neural progenitors and differentiating neurons, we could more clearly follow axonal projections as they fasciculated or bifurcated and innervated distinct regions with the mGFP reporter line.
Determining the developmental profile of axonal projection patterns and elucidating how neural circuits form both anatomically and physiologically is important in the context of brain development, neurological disease, and cell-based therapeutic approaches to ameliorate diseases of the nervous system. Numerous technical advances and genetic approaches have allowed for the examination of neural circuits. However, most of the methods in mouse models rely on recombinant viral technology (Morrow et al., 2008) or by genetically modifying an allele by inserting Tau-LacZ or GFP under the control of a specific gene such that manipulated cells and their circuits will be discernible (reviewed in Luo et al., 2008). It was previously suggested that Cre-mediated recombination concomitant with second-generation phenotyping alleles could uncover the logic of genetic neuroanatomy (Joyner and Zervas, 2006). Although GIFM has led to a deeper understanding of how cells expressing genes at distinct time points in development contribute to developing and mature structures (Kimmel et al., 2000, Zervas et al., 2004, (Ahn et al., 2005, experimental evidence relating specific lineages to developing neural circuits has not been ascertained.
Cumulative marking with En1Cre and comparing Z/EG and mGFP phenotyping alleles demonstrated that the trigeminal ganglia contained En1-derived differentiated neurons (Fig. 5F) and was devoid of pHH3+ progenitor cells (Fig. 6E–G). Notably, GFP+ projections emanated from the differentiated neurons in the lateral anterior Hb and joined a complex plexus of axons. One portion of the plexus traversed along the anterior-posterior axis as a broad fascicle located interiorly within the ventral brain stem (Fig. 5E, arrows; Fig. 5F, bracket/arrow). The more peripheral axons fasciculated as a tight band that was parallel to the internal bundle (Fig. 5F, yellow arrowhead/small bracket). From the internal projections, axons bifurcated into columns that passed through the posterior half of the trigeminal ganglia (Fig. 5F, white arrowheads) and were directly connected to trigeminal neurons. In contrast, the anterior portion of the trigeminal ganglia had a thick conduit of axons that innervated the lateral craniofacial tissue including the vibrissae and three invariant points around the eye (Fig. 5F, Fig. 2J). Thus, En1-derived neurons formed a continuous network (lateral-anterior-dorsal Hb to trigeminal ganglia to craniofacial structures) that shared a common genetic background, being derived from En1 expressing cells. Because the trigeminal ganglia does not have En1 expressing cells at E12.5 as determined by En1 in situ hybridization (Supplemental Fig. 1F–J), the En1-derived cells were likely derived from earlier migrating neural crest cells.
The Wnt1 lineage, marked by GIFM and tamoxifen administration at E8.5, showed that the trigeminal ganglia also consisted of differentiating neurons derived from Wnt1 expressing progenitors (Fig. 10). The mosaic nature of GIFM was advantageous because we could clearly delineate neurons in the trigeminal ganglia, which revealed their bipolar morphology and axons that coalesced to form localized fascicles (Fig. 10A and B). Notably, the Wnt1-lineage contributed to the full extent of the trigeminal ganglia when temporally marked at E8.5. The Wnt1-derived projections on the posterior side of the trigeminal ganglia connected directly to the lateral Hb flexure (Fig. 10C and D) and were clustered into tight columns (Fig. 10C) that were then loosely connected with the Hb (Fig. 10D). Wnt1-derived neurons sparsely populated the inner portion of the Hb flexure (β-gal+) (Fig. 10E) and were positioned on the opposite side of the flexure (Fig. 10 reference, arrowheads) to where the En1 (r1)-derived neurons were observed. These Wnt1-derived neurons formed long sparse tangential projections of the inner plexus (Fig. 10E). The more peripheral, dense projections at the flexure appeared to be projections that connected with the trigeminal ganglia. Wnt1-derived trigeminal neurons innervated the craniofacial region and occasionally branched from the primary fascicle and terminated in 3-dimensional clusters (Fig. 10F) and innervated positions that were similar to the En1-derived connections around the eye, although there were fewer projections that joined the whisker pad vibrissae. High magnification of 1 um thick optical sections and z-series acquisition demonstrated that the mGFP allele allowed for detailed analysis of single axons as they came in close opposition to each other where they formed micro-columns in the trigeminal (Fig. 10G, ovals). Similarly, detailed analysis of the axons in the Hb flexure revealed that axons from the trigeminal tended to course over the longitudinally projecting axons and elaborated branches in the interior of the Hb (Fig. 10H).
In contrast, analysis of the SC and body wall revealed that En1 and Wnt1 derived neurons did not populate the same region of the cord nor did they form connections with common targets. En1 is expressed in SC (Joyner and Martin, 1987; Davidson et al., 1988) and with En1Cre we confirmed that these cells contributed to the ventral SC as previously described (Saueressig et al., 1999). An advantage of the mGFP reporter was that we were able to observe marked cells by nuclear β-gal labeling and follow their axons with GFP, which showed that En1-derived cells are located in two bilateral columns in the SC with projections that radiate outward toward the periphery of the cord where they form a fascicle that exits the plane of the cord (Fig. 4G-3, Fig. 5G-3), which is in agreement with Saueressig et al. (1999). Adjacent to the cord a non-neural En1-derived domain was detected in the intercostal region of the body wall. In contrast, the Wnt1 lineage marked at E8.5 contributed to DRG neurons. Thus, unlike the brainstem-trigeminal ganglia-craniofacial complex, En1 and Wnt1 lineages give rise to non-overlapping structures in the spinal cord. A novel finding ascertained by comparing En1Cre;Z/EG or En1Cre;R26R embryos in which the body wall is marked (Fig. 4G-3) with Wnt1-CreERT;mGFP embryos (Fig. 9K) is that Wnt1-derived DRG neurons (marked at E8.5) passed between En1-derived intercostal tissue and penetrated lateral-ventral tissue including the limbs (Fig. 9A; inset, 9C). In summary, using the mGFP phenotyping allele in cumulative marking versus GIFM we coupled genetic lineage to neural circuit formation during embryogenesis and suggest that this is a viable approach for reconstructing selectively marked circuits in vivo.
We assessed embryos at E10.0 that had been marked either cumulatively or with GIFM to compare how the reporters behaved at two developmental stages (E10.0 and E12.5). For cumulative marking we compared En1Cre;R26R, En1Cre;Z/EG, and En1Cre;mGFP embryos (Supplemental Fig. 2A–C). For GIFM, we used Wnt1-Cre;R26R, Wnt1-Cre;Z/EG, and Wnt1-Cre,mGFP embryos that had been subjected to tamoxifen at E8.5 and analysis 36h later (Supplemental Fig. 2D–F). The R26R phenotyping allele produced marked cells, as detected with x-gal labeling, in both cumulatively and conditionally marked embryos (Supplemental Fig. 2A and D). The distribution of marked cells reliably labeled domains that are apparent precursors to the domains observed at E12.5. Comparatively, when using the Z/EG allele in conjunction with En1Cre cumulative marking, cell somata and in the case of the nervous system, neuronal processes, could be readily detected in whole-mount embryos without tissue processing using a standard epi-fluorescent dissecting microscope (Supplemental Fig. 2B). The Z/EG phenotyping allele revealed cumulatively marked regions that appeared similar to domains detected with the R26R reporter including the mes and r1 as well as neural crest derivatives migrating into craniofacial regions and into the first branchial arch (Supplemental Fig. 2A and B). In sections of En1Cre;Z/EG embryos, cumulatively marked (GFP+) cells were easily detected in neural and non-neural tissue with neurons displaying clear morphological features including axonal projections (data not shown). Surprisingly, we did not detect any marked cells in En1Cre;mGFP embryos by whole-mount analysis at E10.0 (Supplemental Fig. 2C), although a relatively small number of differentiating neurons were observed in sections labeled with β-gal/GFP antibodies (data not shown). Of the three phenotyping alleles used in GIFM experiments in which tamoxifen was administered at E8.5, only the R26R reporter showed marking by whole-mount analysis at E10.0 (Supplemental Fig. 2D–F). Thus, cumulative marking or temporal GIFM of genetic lineages with the R26R allele allows for the detection of cell somata with minimal tissue processing in whole-mount embryos. In sections, β-gal+ cells were readily detected in E10.0 Wnt1-CreERT;R26R embryos (Zervas et al., 2004). In contrast, a small amount of differentiating GFP+ cells could be detected in sections from E10.0 Wnt1-CreERT;mGFP embryos, but no marked cells were observed in E10.0 Wnt1-CreERT;Z/EG embryos (data not shown). However, when these alleles were used with Wnt1-CreERT and marked at the same time point (E8.5), temporally marked cells could be detected at E12.5 in whole-mount embryos and in sections. These findings suggest that at E10.0 there are very few differentiated neurons, which are only detectable by antibody-enhanced labeling, but not by whole-mount fluorescent detection. The reason for the lack of marking with the Z/EG reporter in GIFM experiments at E10.0 may be that the EGFP from the Z/EG allele requires a sufficient amount of time to be produced and accumulate in cells. This is an important consideration in GIFM experiments where the ascertainment of morphogenetic movements is desired. In particular, use of the Z/EG allele in conditional marking experiments may preclude the ability to determine of the distribution of the initially marked population prior to subsequent movements (reviewed in Joyner and Zervas, 2006).
The use of appropriate phenotyping alleles provided excellent details of developing cell populations. We also determined their value in understanding how progenitor cells transiently expressing genes contribute to complex anatomical and functional structures in the adult brain (Supplemental Fig. 3). Tamoxifen administration at E8.5 resulted in marking of cells distributed in the dorsal Mb of the adult consistent with previous findings using the R26R reporter (Zervas et al., 2004). We focused on the GFP phenotyping alleles because of their putative ability to delineate cell morphology and axonal projections. The Z/EG and mGFP phenotyping alleles yielded similarly marked spatial domains (for example the dorsal Mb), but marked cells in the Wnt1-CreERT;Z/EG background were far more sparse than observed with the Wnt1-CreERT;mGFP line (Supplemental Fig. 3A and B). The Z/EG line provided clear morphological details and cellular resolution with GFP immunolabeling, for example revealing processes emanating from neuronal somata (Supplemental Fig. 3A, inset). In contrast, the mGFP line revealed a rich plexus of projections (GFP+) and β-gal+ nuclei (Supplemental Fig. 3B, inset), which may be useful in combination with marker analysis. Close inspection of double immunolabeled neurons in Wnt1-CreERT;mGFP mice revealed that GFP was faintly localized to neuronal somata as opposed to intensely labeled projections (Supplemental Fig. 3B, inset). We also administered tamoxifen to Wnt1-CreERT;Z/EG and Wnt1-CreERT;mGFP embryos at E11.5, when Wnt1 expression has become more restricted and substantially marks the posterior Hb (Zervas et al., 2004). Neurons had well-discernible morphological features when GFP antibody labeling was used with the Z/EG phenotyping allele (Supplemental Fig. 3C). The mGFP reporter yielded more marked cell nuclei and discrete projections, as observed with β-gal/GFP double immunolabeling (Supplemental Fig. 3D) when compared to the same domain in the Z/EG line. Finally, we examined cumulative marking in the posterior Hb of adult En1Cre;Z/EG mice (Supplemental Fig. 3E) to maximize the number of cells expressing the Z/EG reporter. We observed numerous morphologically distinct neurons including calretinin-positive neurons (Supplemental Fig. 3E, arrows) and could also detect fine axonal projections oriented longitudinally in the posterior Hb. With Wnt1-CreERT;mGFP marking, we also observed substantial marking in the posterior Hb, but could not clearly discriminate neuronal morphology (Supplemental Fig. 3F).
In summary, we used cumulative marking and GIFM with three different reporter alleles and ascertained the strengths of each phenotyping allele in a comparative analysis (Table 1). Each of these reporters provides a robust view of developmental biology and uncovered unique features related to lineage and development. We demonstrate that when used comparatively, the three reporters provide a very detailed analysis of marked cells and axonal projections. This was exemplified in both cumulative marking and GIFM approaches. Notably, we also have gained a further understanding of the overlapping and complementary contribution of En1 and Wnt1 expressing cells to both neural and non-neural tissue over time. The information provided here is likely to be valuable to the developmental biology and neurobiology community to help elucidate which allele is most applicable to specific biologically relevant questions.
The generation of Wnt1-CreERT and En1Cre mice were previously described (Zervas et al., 2004; Kimmel et al., 2000). The R26R reporter mice were generously provided by P. Soriano (Soriano, 1999), the Z/EG mice (Novak et al., 2000) were obtained from Jackson Laboratories (stock # 003920), and TaumGFP (lox-STOP-lox-mGFP-IRES-NLS-LacZ-pA) mice (Hippenmeyer et al., 2005) were generously provided by S. Arber. Mice were housed and handled in accordance with Brown University Institutional Animal Care and Use Committee guidelines.
Embryonic tail was digested in 100µL of tail lysis buffer (containing Proteinase K) for 12h at 60 °C followed by heat activation at 90 °C for 10 min. A 600bp fragment of Cre or CreER was amplified in a 20uL reaction (16.54 µL of ddH2O, 2 µL of 15mM MgCl2 10x buffer, 0.16 µL of 100mM dNTPs (25mM each nucleotide), 0.10 µL each of 150pmol/µL NLS-Cre forward primer (5’-TAA AGA TAT CTC ACG TAC TGA CGG TG-3’) and 150pmol/µL NLS-Cre reverse primer (5’-TCT CTG ACC AGA GTC ATC CTT AGC-3’), 0.10 µL of Taq enzyme (Invitrogen; Carlsbad, CA) and 1.0 µL of tail lysate DNA template] using a Eppendorf PCR cycler (94 °C for 2 min, 30 cycles of 94 °C for 30 s, 58 °C for 1 min and 72 °C for 30 s, and 72 °C for 10 min). A 250 bp fragment of the R26R allele was amplified in a 20uL reaction [12.44 µL of ddH2O, 2 µL of 15mM MgCl2 10x buffer, 0.16 µL of 100mM dNTPs, 0.10 µL of 150pmol/µL R26R primer 1 (5’-GCG AAG AGT TTG TCC TCA ACC-3’), 0.10 µL of 150pmol/µL R26R primer 2 (5’-GCG AAG AGT TTG TCC TCA ACC-3’), 0.10 µL of 150pmol/µL R26R primer 3 (5’-GGA GCG GGA GAA ATG GAT ATG-3’), 0.10 µL of Taq enzyme and 5.0uL of tail lysate DNA template] with the following PCR program (94 °C for 2 min, 30 cycles of 94 °C for 1 min, 64 °C for 1 min and 72 °C for 1 min, and 72 °C for 10 min). A 600bp amplicon indicating the mGFP or Z/EG allele was amplified in a 20 µL reaction [15.54 µL of ddH2O, 1 µL of DMSO, 2 µL of 15mM MgCl2 10x buffer, 0.16 µL of 100mM dNTPs, 0.10 µL of 150pmol/µL EGFP sense primer (5’-CTG GTC GAG CTG GAC GGC GAC G-3’), 0.10uL of 150pmol/µL EGFP anti-sense primer (5’-CAC GAA CTC CAG CAG GAC CAT G-3’), 0.1 µL of Taq enzyme and 1.0 µL of tail lysate DNA template] using the following PCR program (94 °C for 3 min, 30 cycles of 94 °C for 30 s, 60 °C for 30 s and 72 °C for 1 min, and 72 °C for 3 min). 4uL of 6x loading dye was added to the PCR products and heated to 65 °C for 5 min. Samples were then run on a 2% agarose gel containing SYBRsafe (Invitrogen, S33102 at a concentration of 1µl/100mL in TBE) at 140V for 1 hour. Gels were visualized using a blue light box.
A 20-mg/ml stock solution of tamoxifen (T-5648, Sigma) was prepared by dissolving tamoxifen in pre-warmed corn oil (C-8267 Sigma) followed by intermittent vortexing during a two-hour incubation on a nutator at 37°C. The tamoxifen stock was protected from light, stored at 4°C, and used for a one month duration. Fate mapping experiments were conducted by crossing Wnt1-CreERT;R26R, Wnt1-CreERT;Z/EG, Wnt1-CreERT;mGFP male mice with Swiss Webster (SW, wildtype; purchased from Taconic) female mice. The morning (0900) of the day a vaginal plug was detected was designated as 0.5 days post-coitus. Tamoxifen was administered at a dose of 4mg (200µl) to time-pregnant SW females by oral gavage at 0900 hours. Mice were sacrificed according to Brown University Institutional Animal Care and Use Committee guidelines at described experimental time points. Mouse embryos (minimally, n≥3 embryos across two or more litters) were dissected free, genotyped, and processed for whole-mount or X-gal labeling, ISH, or were cryoprotected for tissue and cellular analysis.
Embryos were fixed with 4% paraformaldehyde (PFA) for 20 minutes at room temperature (RT), rinsed in PBS, and incubated twice (10 min/each) in X-gal wash buffer (2mM MgCl2, 0.1% Igepal Ca-30, 0.05% deoxycholate in PBS). Subsequently, Wnt1-CreER;R26R embryos were incubated overnight at 37°C in X-gal reaction buffer [0.106g potassium ferrocyanide, 0.082g potassium ferricyanide, 48 ml of X-gal wash buffer and 2 ml of X-gal substrate stock solution (25 mg x-gal/ml in DMSO)]. For En1Cre ;R26R embryos, a shorter processing time (4h) was also used because we observed that the marked expression domains became so thoroughly labeled that the blue reaction product obscured detailed tissue analysis. For whole-mount RNA ISH, embryos were fixed in PFA overnight at 4°C, and subsequently washed in PBS, SSC, and hybridization buffer at 60°C for 2–3h, and hybridized with digoxigenin (dig)-labeled RNA probes (1–2µl) in 500µl hybridization buffer. Embryos were washed with high stringency buffers, blocked in embryo powder/NGS/PBS, and processed for alkaline phosphatase (AP)-conjugated anti-dig antibody (Roche) over night at 4°C. The next day embryos were washed with NTMT and incubated in AP substrate (BM purple) in NTMT-levamisole until the desired color reaction was achieved.
Embryos identified for tissue/cellular analysis were dissected in PBS over ice and a small (1–2mm) tail sample was obtained for genotyping. Embryos were fixed in 4% PFA for 4h to overnight at 4°C. Subsequently, embryos were rinsed in PBS and immersed in 15% sucrose and 30% sucrose until submerged. We removed the sucrose and equilibrated the embryos briefly in Optimal Cutting Temperature (OCT) followed by orienting and embedding embryos in OCT in cryomolds. Subsequently embryos were immersed in 2-methyl-butane as described (http://www.adam.com.au/royellis/fr.htm). Briefly, 2-methyl butane is placed in a polypropylene beaker, which is carefully placed in an appropriate vessel to hold liquid nitrogen. Once the temperature reaches −150°C, the specimen is placed into the cooled 2-methyl-butane until frozen. Samples were stored in sealed bags at −20°C. Sections were obtained on a Leica cryostat and mounted on Fisher Biotech Probe On Plus slides (No. 15-188-52).
Cryosections (12µm) were obtained and stored at −20°C until thawed for use. Sections were fixed in 4% paraformaldehyde (PFA) in PBS for 10 min, then washed with PBS two times for 5 min each. Slides were incubated in 2µg/mL proteinase K in PBS for 4 min, washed with PBS for 5 min, re-fixed in 4% PFA for 5 min and washed in PBS four times for 5 min each. Slides were dehydrated in 70% EtOH for 5 min and in 95% EtOH for approximately 2 min and allowed to air dry slightly. The appropriate in situ probe and hybridization solution were combined at a concentration of 2µL/mL and heated at 80°C for 2 min. We generated a Cre probe by sub-cloning a 1kb fragment of Cre into pKS plasmid. The anti-sense in situ probe was generated by NotI digest and digoxigenin-nucleotide labeling with T3 polymerase; 4uL/mL of probe was added to hybridization buffer. For En1, we digested a plasmid with ClaI and used T7 polymerase and for Wnt1, we digested a plasmid with ClaI and used SP6 polymerase. Slides were placed in a hyboven cassette containing autoclaved ddH2O. Probe-containing hybridization solution (300µL) was added to each slide and coverslipped with RNAse-free Hybridslips; samples were hybridized overnight at 55°C. Coverslips were floated off using pre-warmed 5x SSC and slides were washed as follows: pre-warmed high stringency wash 1 time for 30 min at 65°, RNase buffer warmed to 37°C 3 times for 10 min each, pre-warmed RNase buffer treated with 20µg/mL of RNase A at 37°C for 30 min, pre-warmed RNase buffer 1 time for 15 min, pre-warmed high stringency wash 2 times at 65°C for 20 min each, pre-warmed 2x SSC at 37°C for 15 min, 0.1x SSC at 37°C for 15 min, and 0.1% Tween-20 in PBS (PBTween) for 15 min at room temperature. Slides were then placed in a humid box and blocked with 300µL of 10% heat inactivated goat serum in PBTween at room temperature for 1 hour and coverslipped with parafilm. Blocking solution was removed and slides were incubated overnight at 4°C in alkaline phosphatase-coupled anti-digoxigenin antibody diluted 1:5000 in PBT with 1% goat serum and coverslipped with parafilm. The next day, slides were washed 4 times in PBTween at room temperature for 15 min each, then in NTMT buffer containing 0.5mg/mL levamisole (Sigma; St. Louis, MO) 2 times for 10min each. Gene expression domains were detected by staining slides with 0.5mg/mL levamisole in BM Purple AP substrate (Roche; Indianapolis, IN) at room temperature overnight. Once development was complete (determined empirically) slides were washed in PBS for 5 min, post-fixed with 4% PFA for 2 min and washed in PBS for another 5 min. Sections were then counter-stained by incubations in water (1 min), Fast Red (Poly Scientific; Bay Shore, NY) for 3 min, water (1 min), 70% EtOH (1 min), 95% EtOH 2 times (1 min each), 100% EtOH 2 times (1 min each), and xylene 3 times (1 min each). Slides were allowed to air dry slightly and were then coverslipped using Permount (Fisher Scientific).
Cryosections (12um) were rinsed in PBS for 5 min and fixed in 4% PFA in PBS for 5 min. Slides were then rinsed 3 times in 0.2% TritonX-100 (Fisher Scientific; Waltham, MA) in PBS (PBT) for 5 min each and blocked in 10% donkey serum in PBT for 2h at room temperature in a humid box. Anti-β-gal primary antibody (Biogenesis; Sultan, WA; Cat# 4600-1409; donkey anti-goat IgG) was prepared at 1:500 in 10% donkey serum in PBT and was used on all slides. Anti-TH primary antibody (Chemicon; Billerica, MA; Cat# AB152; Lot # 0603025098 donkey anti-rabbit IgG) or Anti-CALRET primary antibody (Chemicon; Billerica, MA; Cat# AB1550; donkey anti-rabbit IgG) was prepared at 1:500 in 10% donkey serum in PBT. Anti-EGFP primary antibody (Molecular Probes; Carlsbad, CA; Cat # A-11122; donkey anti-rabbit IgG) was prepared at 1:600 in 10% donkey serum in PBT and was used on all sections positive for the mGFP and Z/EG alleles. Anti-phospho-histone H3 (Ser10) antibody (Millipore; Temecula, CA; Cat #06–570; donkey anti-rabbit IgG) was prepared at 1:200 in 10% donkey serum in PBT and was used to distinguish mitotic cells. Slides were incubated in 300µL of primary antibody solution at 4 °C in a humid box with a parafilm coverslip overnight. Slides were allowed to come to room temperature, coverslips were removed, and slides were washed 5 times with PBT for 10 min each. Alexa 555 secondary antibody (Molecular Probes; Cat # A-21432; donkey anti-goat IgG) and Alexa 488 secondary antibody (Molecular Probes; Cat # A-21206; donkey anti-rabbit IgG) were prepared at a concentration of 1:500 in 1% donkey serum in PBT. Sections were incubated in 300µL of secondary antibody solution for 2 hours at room temperature in a humid box with parafilm coverslips. Slides were then washed with PBT 5 times for 10 min each and counterstained with .01% Hoechst 33342 (Molecular Probes; Cat # H-3570) in PBS for 5 min in the dark. Slides were washed 2 times with PBS for 2 min each, dried and coverslipped.
Whole-mount fluorescence and x-gal processed embryos were obtained on a Leica MZ16F stereo fluorescent dissecting microscope using PictureFrame software. Images of sections were obtained on a Leica DM600B epifluorescent microscope with a 2.5x objective (low magnification images). High magnification images were obtained using a motorized stage with a 20x or 40x objective to collect 1 µm optical sections (in the z-axis), which were then analyzed with the Volocity 5.1 visualization module. All images were pseudo colored live as part of the acquisition palettes. 3-dimensional renderings of neurons shown in figure 8D–E, figure 10C–F were generated in the 3-dimension module of Volocity 5.1 using the entire z-stack of 1 µm optical sections followed by adjusting the density, brightness and contrast. It should be noted that the gamma values for 3D rendering were adjusted from the standard value of 1.0. However, the gamma value was unadjusted for all other data processing. Imaging data sets were exported to Adobe Photoshop CS2 or Adobe Illustrator CS2 where montages of representative data were generated.
Cre expression in E12.5 En1Cre and Wnt1-CreER embryos mimics endogenous En1 and Wn1 gene expression. We used in situ hybridization (ISH) with probes that recognize Cre and En1, or Cre and Wnt1 on adjacent sections to confirm that the expression domain of Cre or CreER in En1Cre and Wnt1-CreER E12.5 embryos, respectively, mimicked the endogenous expression of the genes used to drive Cre. In En1Cre;Z/EG embryos, Cre (A, C) and En1 expression, which was slightly weaker than Cre (B, D), were identical in medial sections. In addition, GFP+ marked cells in the Mb and Cb were in the same domains as Cre and En1 RNA (C-E). In contrast, neither Cre (F, H) nor En1 (G, I) were detected in the trigeminal ganglia (Tg) or lateral Hb in lateral sections despite numerous GFP+ marked cells (H-J). Thus, En1-derived cells in the trigeminal were not due to persistent expression of En1 or Cre in this domain. In Wnt1-CreER;mGFP embryos, Cre (K, M) and Wnt1 expression (L, N) were identical in medial sections and correlated with Wnt1-derived neurons(β-gal+, red) marked at E8.5 (M-O). In the ventral mes, β-gal+ neurons of the Wnt1 lineage marked by tamoxifen administration at E8.5 were distributed in a broader domain than Cre or Wnt1 at the stage of analysis (E12.5) and contributed to TH+ dopamine neurons (O). In contrast, Cre (P, R) and Wnt1 (Q, S) were not expressed in the trigeminal ganglia on lateral sections from E12.5 Wnt1-CreER;mGFP embryos. Note that expression of both transcripts was observed in a small domain of the lateral-posterior Hb (P, Q). However, numerous β-gal+ fate mapped cells of the Wnt1 lineage marked at E8.5 were observed in the trigeminal ganglia in adjacent sections (R-T). This illustrates that the Wnt1-CreER transgene mimics Wnt1 endogenous expression and that cells transiently expressing Wnt1 contribute to the trigeminal ganglia. Note that medial and lateral sections were processed together in the same run, which was repeated and that the Cre probe was consistently stronger than En1 or Wnt1.
Comparison of the phenotyping alleles at E10.0. Cumulative marking in En1Cre;R26R or En1Cre;Z/EG embryos at E10.0 revealed a similar pattern when analyzed by whole-mount x-gal (A) or GFP fluorescence (B), respectively. In both cases, the entire mes and r1 was marked as were craniofacial (cf) domains and cells within the first branchial arch (ba). En1Cre;mGFP mice were devoid of fluorescent labeling by whole-mount (C). Tamoxifen was administered to Wnt1-CreERT;R26R, Wnt1-CreERT;Z/EG and Wnt1-CreERT;mGFP embryos at E8.5 (D-F). The R26R reporter was the only phenotyping allele that showed marking by whole-mount (D). GFP fluorescence was undetectable in Wnt1-CreERT;Z/EG and Wnt1-CreERT;mGFP embryos (E, F).
Comparison of the Z/EG and mGFP phenotyping alleles in adult CNS. The GFP phenotyping alleles revealed striking differences and unique features in the adult. GIFM was done by administering tamoxifen to pregnant females harboring Wnt1-CreERT;Z/EG or Wnt1-CreERT;mGFP embryos at E8.5 (A, B) or E11.5 (C, D, F) followed by analyzing adults. Cumulative marking was assessed in adult En1Cre;Z/EG mice (E). Markers were detected by immunolabeling with antibodies recognizing the indicated antigens. Both GFP reporters revealed cells in the posterior Mb (inferior colliculus) marked by tamoxifen at E8.5. Marked cells with the Z/EG reporter were consistently sparser than those marked with the mGFP allele. Fine projections and cell soma of marked cells with the Z/EG reporter were GFP+ while unmarked cells were β-gal+ (A) consistent with the reporter design (Fig. 1I). Marked cells with mGFP were detected by their nβ-gal+ nuclei and GFP+ network of projections and faintly labeled somata (B). Insets (A, B) show 3-D renderings of marked cells, Hoechst 33342 counterstaining indicates nuclei (blue). The Wnt1 lineage marked at E11.5 using the Z/EG reporter revealed distinctly labeled (GFP+) neuronal somata and dendrites as well as very fine processes in the adult posterior Hb (C). Marked cells in the posterior Hb could also be detected using the mGFP phenotyping allele, which showed neurons with β-gal+ nuclei and GFP+ projections and faintly labeled cell bodies (marquees indicate two examples with well defined somata); note the lack of distinct cell morphology (D). Cumulative marking with En1Cre;Z/EG revealed marked neurons (GFP+) with clear morphological detail in the posterior Hb (E). Some of the cells in the post. Hb with a history of expressing En1 are calretinin (CALRET) positive (E, arrows). GIFM with Wnt1-CreERT;mGFP embryos marked by tamoxifen at E11.5 demonstrates that Wnt1-derived neurons substantially contribute to the posterior Hb in the adult (F); note that neurons are readily detected by nuclear β-gal labeling and GFP processes.
We are grateful to S. Arber for the mGFP mice and to members of the Zervas lab who critically read the paper and provided technical assistance. N. Hagan was supported by an NINDS training grant (1T32NS062443, Brown University Department of Neuroscience) and by a Brain Science Program Graduate Research Award, through the Brain Science Program Reisman Fund. This research was also funded by startup research funds (MZ).