Progenitors within the neocortical ventricular zone (VZ) first generate pyramidal neurons and then astrocytes. We applied novel piggyBac transposase lineage tracking methods to fate-map progenitor populations positive for Nestin or glutamate and aspartate transpoter (GLAST) promoter activities in the rat neocortex. GLAST+ and Nestin+ progenitors at embryonic day 13 (E13) produce lineages containing similar rations of neurons and astrocytes. By E15, the GLAST+ progenitor population diverges significantly to produce lineages with 5–10-fold more astrocytes relative to neurons than generated by the Nestin+ population. To determine when birth-dated progeny within GLAST+ and Nestin+ populations diverge, we used a Cre/loxP fate-mapping system in which plasmids are lost after a cell division. By E18, birth-dated progeny of GLAST+ progenitors give rise to 2–3-fold more neocortical astrocytes than do Nestin+ progenitors. Finally, we used a multicolor clonal labeling method to show that the GLAST+ population labeled at E15 generates astrocyte progenitors that produce larger, spatially restricted, clonal clusters than the Nestin+ population. This study provides in vivo evidence that by mid-corticogenesis (E15), VZ progenitor populations have significantly diversified in terms of their potential to generate astrocytes and neurons.
Radial glia provide structural scaffolds for the migration of immature neurons (Rakic 1972; Sidman and Rakic 1973) and are the major progenitor cell type producing neurons and astrocytes in the developing neocortex (Noctor et al. 2001, 2002; Malatesta et al. 2003). Analysis of progenitors in the neocortical VZ by immunocytochemistry, retroviral lineage tracing, and live-cell imaging in rodents indicates that most, if not all, dividing cells in the VZ arise from, or are themselves, radial glial cells (Noctor et al. 2001, 2002; Weissman et al. 2003). Birthdating and lineage analysis indicate that progenitors in the neocortex undergo a marked change over time in potential to generate neurons and then astrocytes (Götz et al. 2002; Kriegstein and Götz 2003; Costa et al. 2009). Similarly, in vivo fate-mapping methods with Cre-recombinase (CRE) transgenes show that Nestin+ and glutamate and aspartate transpoter (GLAST+) radial glia populations in the forebrain generate significantly more neurons or more astrocytes, respectively, in mice (Anthony et al. 2004; Anthony and Heintz 2008).
The generation of the astrocyte in the neocortex occurs both by a direct transitioning of radial glia to astrocytes (Levitt et al. 1983) and by the proliferation of astrocyte-generating progenitors and astrocytes away from the VZ surface (Levison et al. 1993; Levison and Goldman 1993). The proliferation of astrocyte progenitors occurs both within the SVZ in prenatal and postnatal cortex (Levison et al. 1993), and recently, it has been shown by 2-photon live-cell imaging that local proliferation of astrocytes within neocortical lamina contributes significantly to the generation of astrocytes (Ge et al. 2012). It is not currently known whether the Nestin+ and GLAST+ radial progenitor populations originating from the embryonic ventricular zone (VZ) in the neocortex differ in terms of the types of the astrocyte-generating mechanisms they display. Similarly, it remains unclear when progenitors at the surface of the neocortical VZ transition from neuronal to astrocyte-generating progenitors, and whether subpopulations cohabit the same regions of the neocortical VZ.
In this study, we used in utero electroporation (IUE) to fate-map GLAST+ and Nestin+ neocortical progenitor populations at the VZ surface. Advantages of IUE fate mapping not achieved in previous fate-mapping studies include spatially targeting and temporally targeting radial progenitors that contact the VZ surface. We were able to fate-map the progenitors restricted to the patches of the dorsal–lateral neocortical VZ across 3 time points through the neurogenic period (E13, E15, and E18) in the embryonic rat neocortex. In addition, we used a combination of 2 complementary plasmid-based fate-mapping approaches to label either the immediate “birth-dated” progeny or the complete lineage. Episomal inactivation of plasmids electroporated into VZ progenitors allowed for birthdating and fate mapping of immediately generated progeny. In addition, a binary piggyBac transposon system was used to stably incorporate a green florescent protein (GFP) transgene, avoiding inactivation, and thus allowed for complete lineage labeling (Chen and Loturco 2012). In the current study, our results using the Nestin and GLAST promoters driving piggyBac transposase (PBase) or CRE indicated a progressive diversification in the progenitor population within the dorsal neocortex such that GLAST+/Nestin− progenitors emerged at E15 from GLAST+/Nestin+ progenitors at E13. This emergent progenitor gave rise to lineages with significantly more astrocytes. In addition, we used a new PBase-based multicolor clonal-labeling method to reveal that the GLAST+ progenitors generate clones with larger astrocyte clusters. Together, our results support the model that by mid-neurogenesis radial progenitor populations diversify in their capacity to generate astrocytes.
Materials and Methods
Pregnant Wistar rats were obtained from Charles River Laboratories, Inc. (Wilmington, MA, United States of America) and maintained at the University of Connecticut vivarium. Animal gestational ages were determined and confirmed during surgery. Male and female embryos were used. All procedures and experimental approaches were approved by the University of Connecticut IACUC.
In Utero Electroporation Surgery
IUE was done as previously described (Chen and Loturco 2012). Briefly, rats were anesthetized with a mixture of ketamine/xylazine (100 and 10 mg/kg, intraperitoneally). Metacam analgesic was administered daily at dosage of 0.01 mg/kg subcutaneously for 2 days following surgery. During surgery, the uterine horns were exposed and one lateral ventricle of each embryo was pressure injected with 2–3 μL of plasmid DNA (0.5–1.5 μg/μL concentration of each plasmid). Binary Cre/loxP and piggyBac transposon vectors were electroporated at 1:1 DNA ratios. Injections were made through the uterine wall and embryonic membranes by inserting a pulled glass microelectrodes into the lateral ventricle and by injecting by pressure pulses delivered with a Picospritzer II (General Valve). Electroporation was accomplished with a BTX 8300 pulse generator (BTX Harvard Apparatus) and BTX tweezertrodes.
pCAG-Cre and pCALNL-GFP were obtained from Constance Cepko (Matsuda and Cepko 2007), and pCAG-PBase and pZG-s were provided by Mario Capecchi (Wu et al. 2007). PNestin-Cre and pNestin-PBase were made by inserting CRE from pCAG-Cre or PBase from pCAG-PBase downstream of the Nestin second-intron enhancer in the plasmid Nestin/hsp68-eGFP provided by Steven Goldman (Roy et al. 2000). This 637-bp enhancer of the second intron of rat Nestin gene (GenBank: AF004334.1) was located between bases 1162 and 1798 and is sufficient to control gene expression in the central nervous system neuroepithelial progenitor cells (Lothian and Lendahl 1997). pGLAST-Cre and pGLAST-PBase were made by inserting CRE from pCAG-Cre or PBase from pCAG-PBase downstream of the GLAST promoter obtained from Dr D.J. Volsky (Kim et al. 2003). This 1973-bp GLAST promoter was from human excitatory amino acid transporter 1 (GenBank: AF448436.1). PCAG-monomeric red florescent protein (mRFP) is the same as previously described (Loturco et al. 2009; Manent et al. 2009). Multicolor piggyBac constructs pPBCAG-GFP, pPBCAG-mRFP, and pPBCAG-cyan fluorescent protein (CFP) are constructed as previously described (Chen and Loturco 2012).
Microdissection and Acute Culture
Three separate litters were transfected at E13 with pGLAST-PBase and pZG-s vectors and harvested separately at E13.5 (12 h), E16 (72 h), or E18 (120 h) after IUE. Timed-pregnant rat dams were sacrificed and embryos were quickly harvested in the presence of ice-cold 1× Hanks' buffered solution. Transfected brains were isolated with the aid of a fluorescence dissecting microscope to help localize the transfected region. Following a medial incision with scalpel blade, both hemispheres were exposed and transfected VZ cells were peeled away from superficial cortical plate cells. GFP-labeled VZ epithelial tissue was pooled from entire transfected litter and dissociated in the presence of trypsin ethylenediaminetetraacetic acid. Following trypsin inactivation, single-cell suspensions were produced and plated onto poly-d-lysine and laminin-coated cover slips. Dissociated progenitor cells were maintained in the presence of serum-free medium for approximately 6–8 h to allow cells to adhere to cover slips and to allow the expression of surface proteins for subsequent immunocytochemistry.
EdU Experiment and Analysis
IUE was conducted with pGLAST-PBase and pZG-s vectors at E13, and litters were harvested separately at E14 (24 h), E16 (72 h), and E18 (120) post-electroporation. Two hours prior to harvesting time point, pregnant rat dams were given 50 mg/kg 5-ethynyl-2′-deoxyuridine (EdU) by intraperitoneal injection. Harvested embryos were first processed for EdU labeling (Click-iT EdU Alexa Fluor 647, Invitrogen, Carlsbad, CA, United States of America), and then stained with mouse anti-Nestin (Chemicon) with goat anti-mouse 568 (Invitrogen) and rabbit anti-GFP with goat anti-rabbit 488 (Invitrogen) using the standard immunohistochemistry protocol. High magnification (100×) confocal single z-sections were acquired in 488 (green channel), 568 (red channel), and 647 (far-red channel) of the VZ of embryonic rat sections from 3–4 brains per time point. Total numbers of GLAST piggyBac GFP+ and Nestin+ cells that colabeled with GFP+ cells were quantified. Nestin primarily stained the processes and cytoskeleton of VZ cells, while GFP filled the entire cell. This allowed for the determination of GLAST+/Nestin+ cells versus GLAST+/Nestin− cells. EdU+ nuclei were quantified and percentages were determined for the GLAST+/Nestin+ versus GLAST+/Nestin− population of all labeled cells.
Immunocytochemistry and Immunohistochemistry
Cover slips were fixed for 20 min in the presence 4% paraformaldehyde (PFA) solution and washed multiple times with 1× phosphate buffered saline (PBS). Postnatal brains were removed and fixed in 4% PFA solution by transcardial perfusion. Postnatal brains were postfixed overnight at 4°C in 4% PFA and then stored in 1× PBS prior to sectioning. Sections were acquired at 50–60 μm on a vibratome (Leica VT1000S) and processed as free-floating sections. Antigen retrieval was performed by steaming free-floating sections submerged in citrate buffer at pH 6.0 for 20 min to break protein crosslinks and unmask nuclear antigens, sections were then processed for immunohistochemistry. Sections or cover slips were blocked, permeabilized, and immunostained by standard methods. The following primary antibodies were used: mouse anti-GFP (Invitrogen), rabbit anti-GFP (Invitrogen), rabbit anti-dsRed (Clontech), mouse anti-glial fibrillary acidic protein (GFAP) (Chemicon), rabbit anti-S100β (ImmunoStar), rabbit anti-CDP/cut-like homeobox 1 (CUX1) (Santa Cruz Biotechnology), rat anti-CTIP2 (Abcam), rabbit anti-GLAST (Abcam), mouse anti-Nestin (Chemicon), mouse anti-CC1 (1:200, Santa Cruz), and TOPRO-3 nucleic acid counterstain (Invitrogen). The following secondary fluorescence antibodies were used: goat anti-rabbit IgG-conjugated Alexa 488 (Invitrogen), goat anti-mouse IgG-conjugated Alexa 488 (Invitrogen), goat anti-rabbit IgG-conjugated Alexa 568 (Invitrogen), goat anti-mouse IgG-conjugated 568 (Invitrogen), goat anti-rat IgG-conjugated 568 (Invitrogen), goat anti-rat IgG-conjugated Alexa 647 (Invitrogen). Immunoperoxidase-based detection was used to stain GFP for light microscopy (Vector Labs). Photomicrographs were acquired using a Nikon Eclipse E400 microscope (Tokyo, Japan) equipped with a digital spot camera (Diagnostic Instruments, United States of America) for light and fluorescence microscopy. In addition, confocal fluorescence microscopy was performed using the Leica TCS SP2 laser scanning confocal microscope (Nussloch, Germany). Maximum projection images are presented unless otherwise specified. Photomontages were generated using Adobe CS3 (San Jose, CA, United States of America) with a manual photomerge tool and Leica TCS SP2 biomapping widget.
Image Analysis and Reconstruction
Fluorescence microscopy images of dissociated progenitor cells or brain slices used to calculate normalized neuronal migration distance, and long-term fate-labeling studies, were processed and analyzed using Adobe Photoshop CS3 and ImageJ software package with cell counter plug-in (NIH, Bethesda, MD, United States of America). Comparable images according to brain landmarks from anterior to posterior regions were used for the quantification of transfected cells and analyzed in a randomized method to control for a bias. All labeled cells within an image were quantified. Neuronal migration analysis was conducted on matched slices cropped from the border of layer VI and white matter to the pial surface (top of layer I). The position of labeled neurons in the cortex (center of the soma relative to the pial surface) was determined for each cell and, for statistical comparisons, cell distances were binned into 5 distance intervals of approximately 200 μm each (bins 1–5 in Fig. 4A).
The quantification of neurons and astrocytes in the gray matter was performed with Neurolucida workstation software (MicroBrightField, Inc., VT, United States of America), ImageJ software with cell counter plug-in or manual cell counting using a Nikon Eclipse E400 upright microscope. In analyzed sections, all GFP-labeled neurons and astrocytes were quantified within transfected slices. Neurons and astrocytes were easily distinguishable based on morphological characteristics, and partially reconstructed cells within the Z-stack or flat-field image were omitted from the analysis. All pyramidal neurons possessed long apical dendrites, basolateral processes with spines, and large cell bodies indicative of glutamatergic pyramidal neurons. Astrocytes were identified as neocortical protoplasmic astrocytes, with dense, bushy short processes and relatively small nuclei. Astrocytic morphological identification was validated with immunohistochemistry for GFAP and S100β.
HEK 293 Transfection
Transfection of human embryonic kidney 293 cells (HEK293) was performed with Lipofectamine 2000 (11668-019, Invitrogen) in OPTI–MEM medium (Invitrogen) according to the manufacturer's protocol. pPBCAG-GFP/mRFP/CFP and pGLAST-PBase were used to transfect HEK293 cells. HEK293 cells were maintained in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% fetal bovine serum and 1% penicillin and streptomycin. Two days post-transfection, HEK293 cells were passaged and plated at the density of 10 cells/μL. Five days after passage cells were briefly fixed in 4% PFA and then mounted with ProLong Gold Antifade medium (Invitrogen). Images were acquired using Stereo Investigator (MicrobrightField).
Multicolor Image Acquisition and Cluster Size Analysis
Fixed brain tissues and cells were imaged with ×10 (0.3 numerical aperture [NA]), ×20 (0.5 NA), ×40 (0.75 NA) air objectives or ×100 (1.4 NA) oil objectives. A Zeiss Axio imager M2 microscope with Apotome, using 488/546/350 nm filter cubes, with the X-Cite series 120Q light source was also used. BrightLine® (Semrock, Rochester, NY, United States of America) single-band 482 nm excitation, 536 nm emission, 506 nm dichroic bandpass filter cube was used to image GFP. Single-band 543 nm excitation, 593 nm emission, 592 nm dichroic bandpass filter cube (Semrock) was used to image mRFP. Single-band 640 nm excitation, 676 nm emission, 654 nm dichroic bandpass filter cube (Semrock) was used to image Cy5. Chroma 31000v2 filter cube (Chroma, Belows Falls, VT, United States of America) with 350 nm excitation, 460 nm emission, and 400 DCLP was used to image CFP. Images were acquired using Stereo Investigator (MicrobrightField) with the HAMAMATSU ORCA-R2 digital camera C10600. Montage images were taken using the virtual slice function of Stereo Investigator with GFP/RFP/CFP channels. To determine the exposure times, clip detect function in Stereo Investigator was used to prevent saturated pixels. Overlayed pictures were automatically generated by Stereo Investigator after image acquisition. Image stacks were also acquired using Stereo Investigator, and maximum intensity projections were made. All the images were further processed and analyzed in Adobe Photoshop CS3 software. Astrocyte clusters were defined as cells sharing the same color that were <300 μm apart from another cell of the same color without an intervening cell of a different color.
Either 1-way or 2-way analysis of variance (ANOVA) was performed by KaleidaGraph version 4.0 (Synergy Software 2006), or Prism version 5.0 (GraphPad Software Inc.) software. Post hoc analyses were conducted for within dataset comparison by either Student–Neuman–Keuls or by Tukey's honestly significance test. A confidence interval of 95% (P < 0.05) was required for values to be considered statistically significant. All data are presented as standard error of the mean.
Episomal CRE and PiggyBac Tranposase Plasmid Methods for Mapping “Birth-Dated” Cells and Lineage
To determine whether neocortical progenitor cells located in the surface of the dorsal VZ at defined times during neurogenesis are capable of producing progeny with divergent fates, we used 2 types of plasmid-based fate-mapping tools (Fig. 1A). One system uses a conventional CRE expression system to gate the expression of GFP in progenitor cells following a recombination event. A second system uses PBase to insert a GFP transgene into the genome of progenitor cells (Fig. 1A). In initial tests of these 2 systems, we used the ubiquitous and very strong cytomegalovirus early enhancer and chicken beta-actin promoter (CAG) promoter to drive expression of either enzyme. To test the CRE system, we coelectroporated CAG-Cre + CAG promoter with floxed “stop cassette” (CALNL)-GFP at E13, E14, E15, E17, and E18 and determined the placement and cell types that were GFP labeled in the adult brain at postnatal day 21 (P21). As shown in Figure 1B, and consistent with birth-dating studies in the rat, a clear relationship existed between the GFP-labeled cohort and the time of transfection. E13 IUE of the CRE system resulted in GFP labeling of only pyramidal neurons that occupied deeper cortical layers (from layer 6 to layer 4). Transfections at E14, E15, E17, and E18 resulted in GFP+ neocortical pyramidal neurons that successively occupied more superficial laminar positions. There was a very little overlap in the laminar positions of labeled neuron from day to day, indicating that in the rat neocortex, the CRE plasmid system reliably fate-labels birth-dated cells. At E18, IUE resulted in labeling of astrocytes and pyramidal neurons (Fig. 1B). To further confirm that the CAG-Cre + CALNL-GFP system labeled birth-dated cells, we pulsed with BrdU 2, 24, and 48 h after transfection of CAG-GFP at E14 and found that approximately 25%, 5%, and then 0% of GFP-labeled neurons at P21 were positive for both GFP and BrdU. Thus, the Cre-expressing plasmid system fate-maps the immediately generated birth-dated progeny, but does not fate-map all cells expected in the lineage (i.e. astrocytes) of radial glia.
The failure of the Cre–episomal plasmid system to label the complete lineage suggested a robust loss or inactivation of plasmid within 2 days following transfection. To get around this loss, we reasoned that if a GFP transgene could be integrated genomically it would escape loss and inactivation and then label the entire lineage. PiggyBac transposon has been optimized and adapted for use in binary nonviral systems to stably integrate a transgene from a “donor” plasmid (Ding et al. 2005; Lu et al. 2009; Chen and Loturco 2012). To test this system in IUE fate mapping, we transfected a “helper” plasmid containing PBase under the control of the CAG promoter, Nestin promoter, or GLAST promoter (see Materials and methods for details) with a donor plasmid that contains a transgene flanked by inverted terminal repeats. Upon expression of PBase, the transgene in the donor plasmid is integrated into the genome of the cell by transposition. As shown in Figure 1C, transfection of the helper and donor plasmid system along with an mRFP expression plasmid at E15 resulted in mRFP expression restricted to layer 2/3 pyramidal neurons, as expected for an episomal plasmid. In contrast, GFP+ cells included pyramidal neurons, astrocytes spanning all layers of the neocortex, oligodendrocytes in white matter (Fig. 1D), and olfactory bulb interneurons (Fig. 1E) all of which were not labeled by expression from the episomal mRFP plasmid. This indicates that the “PBase” system is suitable for fate labeling the entire lineage of neocortical progenitors by IUE while that the episomal plasmid system is suitable for “birth-dating” VZ surface progenitors.
The patterns of results further suggest that the PBase system acts efficiently and rapidly, within a cell division, to label progenitors. Moreover, it would appear to act within 1 cell division, because the Cre-labeling plasmid system effectively birth-dates cells. We further determined the efficiency of PBase lineage labeling by decreasing the PBases plasmid to 0.5 μg and by comparing the ratios of neurons to astocytes labeled by this low concentration to that labeled by 2 μg. Results showed that 0.5 μg/μL of CAG-PBase labeled 64.7 ± 4% of astrocytes relative to neurons at E14, whereas 2.0 μg/μL of CAG-PBase labeled 67.9 ± 6% (P > 0.05) of astrocytes.
Lineages of Nestin+ and GLAST+-Labeled Progenitors Diverge From Early to Mid-Neurogenesis
Nestin and GLAST are expressed in rodent VZ radial progenitors and radial glia. Previous expression and cell culture studies in the neocortex have suggested that the overlapping GLAST and Nestin progenitor populations may have different fate potentials, with the GLAST population competent to produce neurons and astrocytes. However, to date, there have been no direct in vivo fate-mapping tests to compare the lineages of neocortical VZ progenitors positive for Nestin or GLAST. To test this we compared the cells labeled by 2 different PBase systems, one in which the Nestin promoter was used to drive PBase expression, and another in which the GLAST promoter was used to drive the expression of PBase. A CAG-mRFP expression plasmid was also added to label birth-dated progeny, but not the lineage. Electroporations were performed at E13 and E15. As shown in Figure 2A–C, when the binary piggyBac system was electroporated at E13, the labeled lineages for the Nestin and GLAST systems contained GFP+ pyramidal neurons from layers 6 through 2 and mRFP + neurons restricted to deeper layers. GFP+ astrocytes were also present in for both promoters, although cohort labeled by the Nestin system contained approximately 1/2 as many astrocytes (6 ± 2%) as did the GLAST cohort (15 ± 4%). This difference in astrocytes labeled, however, was only apparent when normalized to total neurons (mRFP and GFP positive). If the analysis and normalization is restricted to GFP-labeled only neruons and astrocytes (e.g., cells that are specifically labeled by the PBases lineage-labeling system and then divided after E13 (losing the mRFP label), then the Nestin+ and GLAST+ populations labeled at E13 are not significantly different in terms of the ratios of astrocyte to neuron (Fig. 2G).
At E15, the differences between GLAST+ and Nestin+ lineages became significantly more pronounced (Fig. 2D–G) with a large increase in potential to generate astrocytes in the GLAST+ population. The Nestin piggyBac system delivered at E15 labeled primarily neurons (86 ± 2%) and a smaller proportion of astrocytes (14 ± 2%), while the GLAST system labeled a smaller proportion of neurons (37 ± 3%) and a 4-fold greater proportion of astrocytes (61 ± 3%) relative to Nestin (ANOVA, P < 0.05; Fig. 2F). The astrocytes to birth-dated neurons ratio (GFP+ astrocytes: mRFP+ neurons) for the GLAST system at E15 was 1.7 ± 0.9, while the astrocyte to birth-dated neuron ratio resulting from labeling by the Nestin plasmid system was 0.16 ± 0.4: A more than 10-fold increase in the ratio of astrocytes generated relative to birth-dated neurons generated in the cohort. Thus, from E13 to E15, there is a significant difference in the lineage generated from Nestin+ and GLAST+ progenitors such that the GLAST+ progenitors diverge to produce more astrocytes than the Nestin+ population (Fig. 2G; significant interaction in 2-way ANOVA, P < 0.045).
Emergence of a GLAST+/Nestin− Progenitor Population in the VZ
The fate mapping above suggested that, around mid-neurogenesis (E15), there may appear a GLAST+/Nestin− progenitor population within the VZ that produces lineages that are highly astrocytogenic. We thus sought to determine whether a GLAST+/Nestin− progenitor population emerged from an earlier GLAST+/Nestin+ population. To determine this, we electroporated E13 embryos with the GLAST piggyBac transposon system to label a starting VZ progenitor pool, then harvested, microdissected, and acutely dissociated VZ cells at E13.5, E16, and E18 (Fig. 3A). The dissociated cells were immunolabeled for GLAST or Nestin (Fig. 3B), and the proportion of GFP+ cells that were immunopositive for GLAST or Nestin was determined for each harvest time (Fig. 3C). As shown in Figure 3B and quantified in 3C, nearly all GFP+ cells were GLAST+ at each harvest time indicating that the VZ progeny of GLAST+ progenitors labeled in the VZ at E13 maintain their GLAST positivity at least up to E18. In contrast, there was a progressive decrease in the proportion of GFP+ cells in the VZ, which were positive for Nestin. This amounted to 20% of cells being GFP+/Nestin− at E15 and 30% GFP+/Nestin− at E18. In addition, in sections double immunostained for GLAST and Nestin at E16 (Fig. 3D), we found that while nearly every cell in the VZ labeled positive for GLAST (99%), 18% of GLAST+ cells in the VZ were not positive for Nestin (Fig. 3E), which is consistent with the piggyBac transfection experiment described above. Together, these results are consistent with the GLAST+/Nestin− population of progenitors in the VZ emerging from an earlier GLAST+/Nestin+ progenitor population.
GLAST+ and Nestin+ Progenitors Generate Similarly Fated Neuronal Progeny at E15
Given that at E15, there is a difference in the lineage potential of GLAST+ and Nestin+ progenitors as demonstrated by increased astrocytes, we asked whether the pyramidal neurons generated at E15 by these progenitor populations might differ in either their laminar or their molecular identities. To test for this possibility, we used both the Cre-episomal plasmid system to label E15 birth-dated neurons from Nestin and GLAST progenitors and used the piggyBac system at E13 to assess the GFP+/RFP− population of neurons generated from Nestin and GLAST progenitor populations. Neocortical progenitors were electroporated at E15 or E13 and brains harvested at P21. We found that E15 transfections of either the Nestin or the GLAST CRE systems labeled pyramidal neurons, but never astrocytes (Fig. 4A). We further compared the laminar positions and relative positivity for CUX1 and CTIP2 between progeny labeled by the CAG, Nestin, or GLAST promoters driving CRE. We found that CAG, GLAST, and Nestin promoters generated similar neurons that migrated to layer 3 and deep layer 2 with no apparent significant differences in their laminar positions (Fig. 4A, Nestin = 710 ± 9 μm and GLAST = 674 ± 26 μm from the pial surface; ANOVA, P > 0.1). We also confirmed the common identify of Nestin and GLAST progenitor-generated neurons by CUX1 and CTIP2 immunopositivity. Anti-CUX1 antibodies label primarily neurons in layer 2/3, and anti-CTIP2 antibodies label pyramidal neurons in deeper layers and scattered layer 2/3 pyramidal neurons. We determined the percentage of GFP+ neurons that were either CUX1+ or CTIP2+ in CAG, GLAST, and Nestin promoter conditions (Fig. 4B) and, similar to laminar position, found no significant differences between the promoter conditions. In CAG, GLAST, and Nestin, we found a small percentage of E15 fate-mapped cells that were CTIP2+ (3–4%), while the majority of the cells were CUX1+ (79–91%). These differences were not significant across the 3 promoter conditions for either CUX1 or CTIP2 (ANOVA, P > 0.05; Fig. 4B). In addition, when we quantified the upper layer, GFP “only” positive neurons that were labeled by the piggyBac system at E13, we similarly found no significant differences between the percentage of labeled neurons that were positive for CUX1 between the Nestin and GLAST promoter conditions (Fig. 4C). These results together suggest that although the lineages of GLAST+/Nestin− and GLAST+/Nestin+ progenitors are significantly different in their subsequent astrocyte-generating capacity, they do not differ in terms of the laminar position or positivity for CUX1.
GLAST+ Progenitors Generate Significantly More Birth-dated Astrocytes Than do Nestin+ Progenitors in Late Embryogenesis
We next applied the Nestin and GLAST CRE plasmid system to progenitors at E18, a time when CAG-cre (Fig. 1B) first begins to label the significant numbers of astrocytes, to determine whether birth-dated progeny of GLAST and Nestin positive progenitors at the VZ surface diverge before birth and without additional proliferation. In contrast to E15 progenitors, we found marked differences in the number of astrocytes between GLAST and Nestin promoters (Fig. 5A–D). GFP+ cells mapped with the Nestin promoter were primarily neurons (79 ± 4%) with some astrocytes (21 ± 4%). In contrast, GFP+ cells fate-mapped with the GLAST promoter generated approximately equal proportions of neurons (55 ± 3%) and astrocytes (45 ± 3%; Fig. 5C). The strong ubiquitous CAG promoter fate-mapped an intermediate number of astrocytes and neurons with 72 ± 2% and 28 ± 2%, respectively (ANOVA, P < 0.0001, Fig. 5C). A comparison of astrocyte-to-neuron ratios generated from the Cre-mapping system at E15 and E18 for all 3 promoters illustrates the shift in cell type production for GLAST and Nestin progenitors (Fig. 5D). Unlike E15, the immediate or birth-dated progeny of Nestin+ progenitor populations generate significantly more neurons than astrocytes, while the GLAST+ progenitors produce approximately equal numbers of pyramidal neurons and astrocytes. Thus, the difference in lineage that is apparent at E15 does not show itself in the generation of birth-dated progeny until near E18, when astrocytes are first beginning to be generated. Furthermore, as electroporation targets cells at the VZ surface, GLAST+ radial progenitors have significantly diversified to astrocyte biased fates in the rat neocortex at E18.
The GLAST+ Population Generates Larger Clonally Related Astrocyte Clusters
As described above, the largest difference in fates between Nestin+ and GLAST+ progenitors was observed when the lineage of progenitors was labeled at E15 with the piggyBac tansposon system. Relative to birth-dated neurons (mRFP+), approximately10-fold more astrocytes were generated from the GLAST+ VZ progenitors than from the Nestin+ progenitors. Astrocytes are generated in the neocortex directly from radial glia transitioning to astrocytes, from SVZ proliferation of astrocyte progenitors (Levison and Goldman 1993), and from local proliferation of astrocytes within neocortical lamina (Ge et al. 2012). To address whether a component of the increased astrocyte-generating capacity of GLAST+ progenitors was due to the clonal expansion of astrocyte progenitors, we applied a multicolor clonal-labeling approach. This approach, as represented in Figure 6A, takes advantage of the stochastic integration of piggyBac donor transgenes and the resulting variety of colors that can be achieved with varying expression levels of RFP, GFP, and CFP. Because the integrated transgenes are stably inherited by cells in a lineage, clonally related cells will maintain a particular color code. To demonstrate this color consistency in clonal groups, we transfected HEK293 cells with the 4 plasmid system and split the cells and let them proliferate at both clonal densities and above clonal density (Fig. 6B). We found that, after expansion, multicellular clusters of the same color were apparent in clonal cultures, and also in more dense cultures, indicating that clonal identity can be designated by unique colors resulting from the combination of fluorescent proteins expressed.
We next combined this multicolor approach with the Nestin and GLAST promoters driving PBase expression to determine whether there was a change in the size of astrocyte clonal clusters in GLAST+ and Nestin+ progenitors from E15. We reasoned that if there were no differences in local proliferation, but solely a change in radial progenitor fates, there would be no difference between the two in the size of the same-colored clonally related clusters. In contrast, if there was a shift in astrocyte clonal size in the GLAST+ population then this would suggest an increased proliferation of clonally expanded astrocytes. We found a clear increase in the number of astrocytes in same-colored astrocyte clusters in the GLAST relative to the Nestin lineages (Fig. 6C–E). For example, the Nestin astrocyte groups (N = 138 in 3 brains) were most frequently observed in clonal groups of 1 or 2 cells (61 ± 8%), while 35 ± 2% of the GLAST lineage clusters (316 clusters in 3 brains) contained 1–2 cells (Fig. 6D,E). Moreover, the GLAST+ lineage population contained large astrocyte clusters of up to 20 cells (e.g., Fig. 6C). In fact, while 14 ± 3% of astrocyte clusters contained 10 or more astrocytes of the same color in the GLAST lineage, only 4 ± 1% of the Nestin lineage contained clusters of 10 or more astrocytes. A 2-way ANOVA comparing the percentage of clonal groups containing astrocytes of different numbers between GLAST and Nestin lineages indicated a highly significant interaction between the promoter used and the percentage of astrocytes in a cluster of a given size (2-way ANOVA; P < 0.0001). This interaction, as seen in Figure 6D, indicates that the GLAST lineage contains significantly fewer 1-cell astrocyte components and significantly more clones larger than 5 astrocytes. This increase in astrocyte clonal cluster size is consistent with the GLAST lineage differing from the Nestin lineage in increased local, spatially restricted, proliferation of astrocyte progenitors or astrocytes.
In this study, we have used 2 complementary in vivo fate-mapping methodologies, one using CRE and the other PBase, to map radial glia/radial progenitors in the rat neocortical VZ. Our results show evidence for GLAST+ and Nestin+ progenitors that progressively diversify into distinct progenitor pools between early (E13) and mid- (E15) neocortical neurogenesis. The results are consistent with and extend previous fate mapping in mice that used CRE and transgenic methods (Götz et al. 2002; Anthony et al. 2004; Anthony and Heintz 2008; Pinto et al. 2008). We find that early progenitor pools in the rat are almost exclusively GLAST+/Nestin+, while mid- and late neurogenesis pools show evidence for the emergence of a GLAST+/Nestin− population. Fate mapping of the lineages of GLAST+ and Nestin+ populations using the piggyBac transposon system from early neurogenesis (E13) to mid-neurogenesis (E15) revealed that the Nestin+ population was consistently biased toward producing fewer astrocytes than the GLAST+ pool.
The increase in the number of astrocytes generated by the GLAST+ progenitors relative to Nestin+ progenitors could be due to a fate bias, or a bias in the proliferative potential of astrocytes generated by the GLAST+ population, or a combination of the 2. Our data suggest that a combination of both fate bias and proliferation potential bias contribute to the increases in astrocyte-generating capacity of the GLAST+ population. The E18 episomal plasmid experiment (shown in Fig. 5) tracks the birth-dated population, independent of proliferation, because of inactivation or loss of plasmid, and there is a difference of approximately 2–3-fold more astrocytes generated by the GLAST radial progenitors at E18 shown by this experiment. This increase is far below the 10-fold increase in the ratio of astrocytes to neurons generated from the GLAST population tracked with the piggyBac lineage method. This suggests that an additional reason underlying the difference in the generation of astrocyte in the lineage may be in the proliferation of astrocytes or astrocyte progenitors. The multicolor lineage and clonal-labeling method suggest that the increased astrocyte number in the GLAST lineage is associated with an increase in the size of the astrocyte clonal clusters. Thus, a combination of fate bias to astrocyte-generating progenitors and increased proliferative capacity (larger astrocytic clonal components) are likely to account for the increase in the generation of astrocyte in GLAST+ radial progenitors (Fig. 7).
One set of potential concerns with respect to promoter-based fate and lineage mapping used here is that differences in promoter strength could result in the perseverance of recombinase or transposase, and thus, the fate label may not reflect progenitor differences but rather differences in the amount of CRE or transposase that is expressed. This differential perseverance may then differentially extend labeling into latter born cells in the lineage. There are several pieces of evidence that this is not likely to be an issue for the differences between the Nestin and GLAST conditions in this study. First, as shown in Figure 1B and above in Figure 5C,D, the strongest and most ubiquitous promoter (CAG) does not fate-map more astrocytes than the GLAST promoter. Secondly, when we compare the number of neurons/section generated by the Nestin-Cre fate-labeling system to the number generated by the GLAST-Cre system, there are actually more neurons labeled in the Nestin condition (336 ± 27 for Nestin vs. 238 ± 12 for GLAST, N = 6), which argues against lesser functional CRE expression in Nestin promoter conditions. Finally, to directly test for potential differences in CRE expression, we transfected GLAST-Cre or Nestin-Cre with CALNL-GFP into neocortical progenitor cells at E15 and harvested either 1, 2, or 3 days later. We found for all promoters a gradual reduction in CRE expression over time as determined by CRE immunohistochemistry. In 3 days, GFP+/Cre+ cells were <5% of GFP-labeled cells for Nestin and GLAST promoters. This uniform decrease in protein expression, coupled with the demonstration of rapid plasmid inactivation or loss in the fate-labeling system, (Fig. 1B) indicates that the fate-labeling systems delivered by IUE in this study result from a relatively discrete 1- and 2-day window of progenitor labeling.
Our results are consistent with earlier cell culture (Price et al. 1991; Davis and Temple 1994; Mayer-Proschel et al. 1997) and retroviral lineage tracing (Luskin et al. 1988; Price and Thurlow 1988; McCarthy et al. 2001) experiments that indicated an early divergence of neurogenic and astrocytogenic progenitors in the neocortex. Our studies extend these findings by indicating that populations coexist within spatially defined patches of neocortex in vivo, which progressively diverge. Based on our findings using 2 complementary fate-mapping techniques, we put forward a tentative model for transitions in radial glia in the VZ through neocortical neurogenesis (Fig. 7). The majority of VZ radial glia at E13 are both GLAST+/Nestin+. At E15, radial glia populations have diversified with the lineages generated by the GLAST+ population generating far more astrocytes than does the Nestin+ population. The immediate progeny of both GLAST+ and Nestin+ progenitors at E15 are uniformly neuronal. This suggests the presence of a GLAST+/Nestin− population that has a high potential to generate astrocytes, and a GLAST+/Nestin+ population that produces primarily neurons. We also show that that GLAST+/Nestin− population is from the direct lineage of early GLAST+/Nestin+ cells and is responsible for astrocytes generated in the neocortex from approximately E18 onwards. In addition, a Nestin+/GLAST− population of progenitors may exist, which produces primarily neurons. Finally, the GLAST+ population has a greater capacity to generate lineages with larger clonal astrocyte clusters, suggesting an increase in local astrocyte proliferation in the GLAST lineage.
Lineage tracing experiments using transgenic mice expressing CRE under the control of different promoters have revealed the fate maps of radial glia positive for Nestin, GFAP, or GLAST promoter activity. Using the radial glia-specific “human” glial fibrillary acidic protein promoter to drive the expression of CRE, Malatesta et al. (2003) showed that fate-mapped hGFAP+ radial glia of ventral telencephalon give rise to astrocytes, while those of dorsal cortex give rise to astrocytes and projection neurons. Furthermore, Anthony and Heintz (2008) revealed that the GLAST promoter driving CRE fate-maps the majority of neurons and astrocytes in the adult brain, including neocortex. By comparing the differential expression of hGFAP and GLAST reporters, Anthony and Heintz (2008) further provided evidence that the observed difference in fate maps for hGFAP and GLAST in the ventral forebrain may result from delayed activation of the hGFAP promoter. Our results are in complete agreement with a progressive differentiation of GLAST+ cells from neuronogenic to astrocytogenic progenitors and further provide evidence that a GLAST+ progenitor population or subpopulation becomes highly astrocytogenic at E15.
Our results with IUE fate mapping indicate local diversity of progenitors within the VZ. Previous evidence for local diversity using IUE has come from a study in which plasmids-expressing fluorescent proteins under the control of brain lipid binding protein, Nestin, Tα1, and GLAST promoters were electroporated into the mouse neocortical VZ (Gal et al. 2006). Cells labeled in this manner were shown to have different morphologies and cell cycle kinetics, and this led to the identification of an short neural precursors (SNPs) cell in the VZ of the mouse (Gal et al. 2006; Stancik et al. 2010). It has also been reported that, at E14.5, SNPs fate-mapped in the mouse neocortex by the Tα1 promoter driving CRE label a different neuronal population than those fate mapped by either Nestin or GLAST promoters at the same time point (Stancik et al. 2010). We also found that Nestin and GLAST promoters fate-maps identical pyramidal neuron populations at E15 in the rat. These neurons were of the same laminar position and molecular identity based on CUX1 and CTIP2 immunopositivity.
The present study demonstrates novel applications of the binary piggyBac transposon plasmid system for fate mapping in the mammalian brain. This technology has recently shown by our group to be amenable to lineage tracing radial glia in the developing neocortex, and we have developed a toolbox allowing for a combination of gain and loss of function approaches (Chen and Loturco 2012). We have extended the use of piggyBac transposon-based transgenesis for fate mapping in this study by showing that different promoters driving transposase will label different lineages. There are several advantages of the binary piggyBac transposon system for fate mapping. First, since it is a plasmid-based system it can be spatially directed by IUE. Secondly, it can be used in essentially any species. The system can be easily adapted for genetic manipulation experiments requiring stable transgenesis of transgenes in subpopulations of progenitors by simply subcloning sequences of interest into the donor plasmids. Finally, the binary transposon system overcomes the limitations of episomally based plasmid systems used in IUE fat-mapping previously, and combined with such systems can lead to differential labeling and manipulation of immediate progeny and the complete lineage of a defined population of progenitors. We took advantage of this property in the present study to fate-map both the lineage and immediate progeny of Nestin+ and GLAST+ progenitors in the rat, a species in which lineage-based mapping with Cre- transgenic and floxed reporter lines is not available. Future studies using this system can now interrogate whether there are differences in the molecular mechanisms that govern the differing cellular outputs of the GLAST+ and Nestin+ progenitor types.
All authors have read and approved the submitted manuscript. Specific contributions are as follows: F.S. preformed primary experiments, conducted analysis, generated figures, and assisted in writing. F.C. helped with primary experiments and major intellectual contribution. A.A. helped with quantification and analysis. C.F. assisted with animal surgeries. J.L. conceived the study and was involved with data analysis, preparation of figures, and paper writing.
This study has been supported by funding from NIH grant # R01MH056524.
The authors thank the labs of Constance Cepko, Mario Capecchi, Steven Goldman, and Volsky for the reagents utilized in this study. Conflict of Interest: None declared.