Unraveling the adult cell progeny of early postnatal progenitor cells

NG2-glia, also referred to as oligodendrocyte precursor cells or polydendrocytes, represent a large pool of proliferative neural cells in the adult brain that lie outside of the two major adult neurogenic niches. Although their roles are not fully understood, we previously reported significant clonal expansion of adult NG2-cells from embryonic pallial progenitors using the StarTrack lineage-tracing tool. To define the contribution of early postnatal progenitors to the specific NG2-glia lineage, we used NG2-StarTrack. A temporal clonal analysis of single postnatal progenitor cells revealed the production of different glial cell types in distinct areas of the dorsal cortex but not neurons. Moreover, the dispersion and size of the different NG2 derived clonal cell clusters increased with age. Indeed, clonally-related NG2-glia were located throughout the corpus callosum and the deeper layers of the cortex. In summary, our data reveal that postnatally derived NG2-glia are proliferative cells that give rise to NG2-cells and astrocytes but not neurons. These progenitors undergo clonal cell expansion and dispersion throughout the adult dorsal cortex in a manner that was related to aging and cell identity, adding new information about the ontogeny of these cells. Thus, identification of clonally-related cells from specific progenitors is important to reveal the NG2-glia heterogeneity.


Results
StarTrack clonal analysis to decipher the progeny of postnatal NG2-progenitors. The Star-Track approach 22 is based on 12 piggyBac plasmids that encode up to 6 different fluorophores (with either a nuclear or cytoplasmic localization), each driven by the GFAP-promoter, which are transfected along with a piggyBac transposase plasmid (hyPBase). This method allows single progenitor cells and their cell progeny to be targeted by piggyBac-driven stochastic integration into the genome. This strategy was modified to target the NG2 lineage in vivo by using novel plasmids carrying the mouse NG2-promoter (mNG2), referred to as NG2-StarTrack ( Fig. 1A 22 ). Here, the mixture of NG2-StarTrack plasmids and the hyPBase transposase was injected into the lateral ventricles (LVs) of mice, which were electroporated at P0-P1 (Figs. 1B, S1). At P90, P240 and P365 we then performed a clonal analysis to assess the age-related changes in the postnatal NG2 derived cell progeny and in the progenitor cell potential. NG2-StarTrack labelled cells in the dorsal cortex formed either big clusters, small groups or remained as individual cells (Fig. 1C). Different morphologies could be distinguished in WM (Fig. 1D) and GM (Fig. 1E,F), although the labeled immature cells located close to the ventricle were not considered in these clonal analyses. The cell progeny of targeted NG2-progenitors displayed an inheritable and stable color code at the single-cell level (Figs. 1G and S1). The different fluorescent reporter proteins were detected in separate channels to define the presence/absence of each fluorophore: 1, YFP; 2, mKO; 3, mCerulean; 4, mCherry; 5, mTSapphire; 6, EGFP (Fig. S1B). Each channel was assigned an emission color, except for mT-Sapphire that was represented as dark blue and far red in grey color. Accordingly, the cellular barcode allows the clonally-related cells to be rapidly recognized based on the presence (represented as 1-6) or absence (0) of the fluorescent proteins and their location, whereby the first number corresponds to cytoplasmic labeling and the second number to its nuclear location (Figs. 1G and S1C). Hence, the theoretical color-code created for each clonal cluster can produce more than 14,000 combinations 23 . Sequential sections along the rostro-caudal axis were used to analyze both the location and spatial dispersion of the sibling cells. The frequency of the different color-code combinations was also estimated to rule out the clones that appeared more frequently (data not shown).
Characterization of the adult NG2 derived progeny of early postnatal progenitor cells. NG2-StarTrack can target single progenitor cells with an active NG2 promoter and their derived cell progeny, identified using different neural markers at distinct adult ages (Fig. 2). The cell's morphology and immunolabeling indicated the identity of the sibling cells, and among the 9 animals used in the clonal analysis (3 animals at each age) the proportion of NG2-cells was around 97% of the total cells analyzed (4496 of total cell analyzed) (Fig. 2B). This proportion was higher than that of the sibling astrocytes analyzed at all ages, which represented just 3% of the total cells (Fig. 2B). Similarly, the dispersion of these clonal astrocytes along the rostro-caudal axis was more restricted than that of the NG2 sibling clusters (Fig. 2B).
Although, the morphology of the labelled cells might be sufficient to identify the different types of glial cells, except for those with only nuclear labeling, we performed immunohistochemistry to assess the distribution of different glial markers. Labelled astrocytes were identified through the expression of both glial fibrillary acidic protein (GFAP: Fig. 2C) and S100 calcium binding protein beta (S100β: Fig. 2D). Cells in the NG2 lineage colocalized with labelling for Olig2 (Fig. 2E,F) and PDGFRα (Fig. 2G,H). The expression of NG2 is downregulated in oligodendrocytes, and some immature oligodendrocytes were labelled by these antibodies and identified by their expression of adenomatous polyposis Coli (APC/CC1: Fig. 2I-J). These cells did not express neuronal markers. After their characterization, the different groups of cells were classified in function of their cortical location, either in the GM, WM or both, revealing the identity of the progeny of dorsal postnatal progenitors after NG2-StarTrack targeting. The clonally-related cells in the GM and WM were mainly identified as NG2-cells, although some labelled astrocytes were detected in the GM. There were differences among the clusters of glial cells in terms of the number of cells and their dispersion along rostro-caudal axis.
Clonal analysis of adult derived NG2-cell progeny after NG2-StarTrack targeting of individual postnatal progenitors. We then analyzed the clonal cell pattern of the pallial glial cells derived from post- www.nature.com/scientificreports/ natal (P0) progenitors with an active NG2 promoter (Fig. 3A) at different adult ages (P90, P240 and P365). We randomly selected 66 NG2-cell clones from three different animals of each age (P90:18, P240:30 and P365:18), attending to their color code and frequency. Their rostro-caudal location was calculated from the rostral end of the LV as the initial point of electroporation, and their dispersion was defined by the position of all the sibling cells at the different levels along the rostro-caudal axis. There was an increase in the number of cells per NG2 clone with age ( Fig. 3C) and therefore, there was greater cell dispersion of most sibling NG2-cells (Fig. 3B), forming larger clones ( Fig. 3C) with age ( Fig. 3D). While the clones at P90 dispersed across 50-200 microns, in older animals the clones were larger and had a more heterogeneous dispersion. Hence, the biggest NG2 clone contained 607 cells at P240 (8 months) and the smallest just 2 cells (Fig. 3D). In addition, sibling NG2-cells were located in the GM, WM or both as mentioned above ( Fig. 3E-I). At P240, the clones located in the WM and GM had more cells and a greater dispersion than at other ages ( Fig. 3H,I). Nonetheless, the sibling cells located in the GM alone had fewer cells (Fig. 3E,G,I) and less cell dispersion than those in the WM. In addition, the presence of regionally mixed clones formed by NG2-cells located in both the GM and WM was notable, occupying the deeper cortical layers and the corpus callosum (Fig. 3F). Those clones located in both cortical areas constituted 62% of total labelled NG2-cells (Fig. 3G) displaying a huge cell dispersion, between 200 and 650 microns, at P240 and P365 (Fig. 3H). In this regard, the biggest clonal clusters were located in the WM (P240) or in both the WM and GM at the different ages analyzed (Fig. 3I). Together, the clonally-related NG2-cells formed larger clones in older animals, with an increase in cell dispersion. Moreover, some clonal cells were located in both the corpus callosum and the deeper layers of the dorsal cortex. These mixed clones were identified at all ages and they had more cells per clone. The number of fibrous astrocytes per clone was very low, usually just per one cell and thus, they were not considered in the clonal analysis. The cells were characterized based on both their morphology and marker expression (Figs. 4B and S1D,E). The astroglial clones displayed significant fewer cells per clone than the NG2 clones ( Fig. 4B,C) and at P90, 8% of the labelled cells were astrocytes and 92% corresponded to NG2-glia. By contrast, astrocytes made up 2% of the tagged cells at P240 and P365. Comparing between ages, the number of protoplasmic astrocytes per clone at P90 was significantly different in older animals (Fig. 4D), with a maximum of 15 cells per clone at P240. Some labelled astrocytes with no sibling cells were observed at different ages (P90:3; P240:1; P365:3) and although yet they were not considered in the final clonal analysis, they could provide information about the behavior of early postnatal progenitors. In addition, the dispersion of astrocytes was apparently similar at all the ages selected (Fig. 4E,F). The majority of clonally-related astrocytes dispersed from 50 to 200 microns along the rostro-caudal axis (Fig. 4F), while the clonal size varied from 2 to 15 cells (Fig. 4D,F). Indeed, in relationship to the size of the clones, astrocytes exhibited homogeneous proliferation with age, in contrast to the NG2-glia clones in which the number of sibling cells increased in older animals (Fig. 5B). Comparing the cell distribution over the electroporated area (Fig. 5C), labelled astrocytes were preferentially located caudally to the NG2 labelled cells, which was most evident at P240 and P365 (Fig. 5C). Nevertheless, the distribution of these glial cells along the rostro-caudal axis was not significantly different (Fig. 5D). In summary, astroglial NG2-derived cell clones had a smaller dispersion and there were fewer cells per clone at the different ages analyzed, reflecting a homogeneous behavior of the tagged postnatal progenitor cells that contrasted with that of the progenitors that gave rise NG2-cells.

Clonal analysis of adult derived-astroglial progeny following NG2-
To conclude, we reveal that clones of NG2-cells in the cortical GM and WM show the largest cell dispersion at the different ages. In addition, we showed a lack of regional mixing of the clones of NG2-cells. These results witness the heterogeneity in both the postnatal NG2-progenitors and their cell progeny.

Discussion
This study is a NG2-StarTrack clonal analysis of the adult cell progeny derived from individual NG2 pallial progenitors at early postnatal ages. To tag NG2-progenitor cells, we used the genomic multicolor genetic tracing tool, NG2-StarTrack, designed to track the cell progeny of individual postnatal progenitors from the SVZ in vivo 8 . Our data revealed the presence of clones of either NG2-glia or astrocytes with different spatio-temporal extensions. However, in contrast to embryonic NG2-progenitors 7,8,24 , no neuronal cells were derived from postnatal progenitors in cortical areas. Previous lineage analysis of neural cells reported the capability of progenitor cells to generate both astrocyte and oligodendrocyte cells in vitro [25][26][27][28] and in vivo 29 .
NG2-cell clones were widespread throughout the cortical rostro-caudal axis, either in the WM and GM. In addition, a subpopulation of astroglial cells were tagged with the NG2-StarTrack mixture. These clones had a constant number of cells over time, unlike the NG2-cell clones that increased in number and dispersion. We also reveal the existence of regional mixed clones formed by clonally-related NG2-cells in both the cortical GM and WM. These mixed clones had more cells and a greater dispersion in the temporal analysis, arguing the NG2cell heterogeneity may not only be related to their fate but also, to their location and to ontogenic processes 6,11 . Regional heterogeneity of NG2-cells has been reported, based on different properties like cell cycle length 30,31 , proliferative response 18 and differentiation rates 32 . Even, NG2-cells in the same area present differences in terms of their transcription factor expression or that of the GPR17, G-protein couple receptor (GPCR 33 ). In the adult brain, our lineage-tracing analyses of NG2-postnatal precursors revealed they proliferated distinctly and displayed different cell fate patterns across adulthood. As reported for the adult progeny of embryonic progenitors,  Otherwise, other studies using Cre-lox mice reported that the proliferative rate of NG2-cells decreased with age 3-5 . As previously suggested using a different StarTrack approach 21 , it is possible that those pallial-derived NG2 clones increase their size at the expense of the direct differentiation in oligodendrocytes of ventral derived NG2 clones. Thus, those differences can be explained as a heterogeneous pool of progenitor cells with different proliferative properties throughout age and domain. Further, previous data reported that the mitotic status of adult NG2-cells is unrelated to their developmental origin 3 . However, there is a coexistence of both slow-cycling stem cell-like NG2-cells with more rapidly cycling amplifying cells 3,20,30 , promoting that the cell cycle of NG2-cells varies in relation to the brain areas and age. Moreover, a smart approach on lineage-targeted transcriptomics reveals the role of transcription factors in lineage transition of glial cells in both physiologic and pathological conditions 36 . At this respect, complementary genetic approaches might assess new insights related to their NG2-cells heterogeneity and fate potential. Our data show that postnatal progenitor cells gave rise to only glial lineages in cortical areas, while embryonic NG2-progenitors produce both neuronal and glial cell lineages 8 . In this regard, time-lapse imaging 37 and in vivo StarTrack tracing 38 revealed the impact of progenitor location on fate potential. Furthermore, it has been proposed that NG2-glia can differentiate into neurons under specific conditions and locations 39 . In this respect, several in vitro and in vivo approaches have been used to decipher the multipotent potential of progenitor cells and their capacity to generate different neural cell types 40,41 . Cre-inducible mouse lines produced few astrocytes after induction 19 , but in other transgenic mice, no astrocytes or neurons were found at adult stages 42 . All these   39,46,47 . The small subpopulation of astroglial cells in relation the number of NG2-cells, derived from NG2 postnatal progenitors, revealed that these cells have different proliferation rates related not only to age but also, to their origin. Clonal astrocytes were less abundant than NG2-cells at all stages and their cell dispersion in the rostro-caudal axis was homogeneous. Thus, as well NG2-glia, astrocytes are characterized by their heterogeneity at different levels, including ontogeny 22,48 , morphology 49,50 , cortical GM and WM location 22,51 , transcriptomic signatures 52,53 or response to brain lesions 54,55 . In addition, astrocytes could influence NG2-glia behavior during development, an interaction that may have an important influence on gliogenesis, ageing and even injury 56 . Recently, sibling astrocytes were shown to preferentially establish GAP-junction coupling relative to the unrelated astrocytes 57 . This coupled response between sibling cells may be implicated in physiological and pathological events both, astrocytes and NG2-glia 55,58 . All these data reinforce the strong heterogeneity of these neural cells at a morphological, functional and genetic level, opening the window to develop new lineage tracing approaches not only based on multicolor labeling but also, on stable genome editing 59 . New transcriptome tools could shed light on the pathologies mechanisms and further therapies 60 .
Ageing is related to the loss of myelin, which is in turn correlated with the loss of cognitive and motor skills, which could be a consequence of the generation of fewer oligodendrocytes 61 . However, NG2-cells preserve the ability to divide in adulthood 62,63 and moreover, NG2-glia respond early to brain insults (like astrocytes), migrating towards to the injury site and increasing their rate of proliferation 19,25 . Several reports indicate that www.nature.com/scientificreports/ most NG2-cells in the cortical GM could be part of a different population of NG2-cells under physiological conditions 24 , although their functions are not fully understood. Thus, NG2-cells increase in number with age, which may be useful to design new therapies to combat cell aging related phenomena. Their capability to generate neurons is not yet fully understood, yet their stem cell like characteristics or their possible reprogramming could be relevant to neurodegenerative diseases and ageing 12,13,64 .
In conclusion, beyond their role in myelination, NG2-glia represent a significant pool of glial cells with diverse functionalities, as well as unique properties that contribute to CNS homeostasis and development. The importance of their origin, cell fate and heterogeneity is still unclear. Thus, further clonal analysis complemented with genetic cell identity strategies might help gain new insights into their behavior, heterogeneity and fate potential.

Materials and methods
Animals. All  Vectors. NG2-StarTrack constructs were designed as described previously 8 . Briefly, the GFAP-StarTrack constructs were used to generate the different NG2 piggyBac vectors with the six different reporter proteins and H2B histone sequence to drive the tag into the nucleus. The human GFAP promoter was removed and replaced with a murine NG2 promoter 65 . The hyperactive transposase of the PiggyBac system (CMV-hyPBase) was kindly provided by Dr Bradley and all the plasmids used were sequenced to confirm successful cloning (Sigma-Aldrich; Merck KGaA, Darmstadt, Germany). For all injections, the plasmid mixtures contained the twelve NG2-StarTrack constructs and a hyperactive transposase of the PiggyBac system under the CMV promoter to integrate copies of the NG2-StarTrack plasmids randomly into postnatal progenitor cells.
Postnatal electroporation. Postnatal electroporation was performed as described previously 38 . Briefly, a plasmid solution containing the plasmid mixture (1-2 µg/µl) and 0.1% Fast Green was injected into the LVs of perinatal animals on day P0-1 using a glass micropipette. After plasmid injection, all pups were electroporated with electrode paddles, placing electroconductive LEM Gel (DRV1800, MORETTI S.P.A.) on both paddles to avoid damage to the pups and to achieve successful current flow. We administered five pulses of 100 V, each pulse lasting 50 ms and separated by 950 ms intervals. The positive electrode was positioned on top of the dorsal cortex to direct the negatively charged DNA. After the five pulses, the electroporated animals were reanimated for several minutes on a 37 °C heating plate before returning them to the mother's nest. The mice were analyzed from P30 onwards (at least three animals per experimental group).
Histological and immunohistological procedures. The adult animals were anesthetized with pentohbarbital (Dolethal, 40-50 mg/Kg) and when fully anesthetized, they were perfused with 4% paraformaldehyde (PFA), and their brain was removed and placed overnight in small tubes with 4% PFA in 0.1 M phosphate buffer (PB). Serial vibratome brain section (50 µm thick) were mounted onto glass slides with Mowiol and stained for different neural markers. Slices were permeabilized with phosphate buffer saline containing Triton X-100 (PBS-T) and then incubated for at least 30 min in blocking solution (5% normal goat serum-NGS-in PBS-T 0.1%). The sections were then incubated O/N at 4 °C with the following antibodies markers: rabbit polyclonal anti-Olig2 (Millipore-AB9610); rabbit polyclonal anti-GFAP (Dako-31745); mouse monoclonal anti-APC (Calbiochem (OP80); rabbit polyclonal anti-PDGFRα (Cell Signalling-3169); a mouse monoclonal anti-S100β (Abcam-Ab66028). After at least three washes with buffer, the sections were incubated for up to 2 h with Alexa far red goat anti-rabbit or goat anti-mouse IgG (1:1.000, Alexa Fluor 633 or 647, Molecular Probes). Finally, the sections were washed several times with buffer, mounted on slides, coverslipped and observed in an epifluorescence microscope (Eclipse E600; Nikon, USA). Imaging acquisition. The sections were examined under an epifluorescence microscope equipped with GFP (FF01-473/10), mCherry (FF01-590/20) and Cy5 (FF01-628/40-25) filters. Images were then acquired on a TCS-SP5 confocal microscope (Leica, TCS-SP5). The confocal laser lines were maximal around 40% in all samples and the conditions for each laser was constant for each animal. The different reporter proteins were taken in separate channels controlling the overlapping between them. The wavelength of excitation (Ex) and emission (Em) was (in nanometers): mT-Sapphire (Ex: 405; Em: 520-535), mCerulean (Ex: 458; Em:468-480), EGFP (Ex:488; Em: 498-510), YFP (Ex:514; Em: 525-535), mKO (Ex: 514; Em: 560-580), mCherry (Ex: 561; Em: 601-620), and Alexa Fluor 633/647 (Ex: 633; Em: 650-760). Maximum projection images were analyzed using LASX software (Leica) and Fiji software ImageJ. All stitching and contrast adjustments were performed with the LasX software (LasX Industries) and Photoshop CS5 software (Adobe). The rostro-caudal axis was reconstructed using the Traken2 plug-in for ImageJ and was estimated as the distance between the beginning of the LVs and the last slice containing cells of that clone. Moreover, the clonal dispersion was calculated considering the distance between the first and last slice containing clonally related cells.

Scientific Reports
| (2020) 10:19058 | https://doi.org/10.1038/s41598-020-75973-y www.nature.com/scientificreports/ Data analysis. For each experiment, the sections were analyzed serially and the cells were counted using the manual cell counter plug-in of ImageJ software. Afterwards, the proportion of those cells in the pallial areas was calculated. For statistics, GraphPad Prism 6.0 (GraphPad, USA) was used and the statistical significance between two groups was assessed with two-tailed unpaired Student's t-tests, using ANOVA for multiple comparisons between the groups. The values were represented as mean ± SEM along the experimental data. A confidence interval of 95% (p < 0.05) was determined for the statistically significant values. Critical values of *p < 0.05, **p < 0.01, and ***p < 0.001 were adopted to determine statistical differences. Graphs were obtained using Excel Office, GraphPad Prism 6.0 (San Diego, USA) and CorelDRAW Graphic Suite 2018 (Corel Corporation, Ottawa, Canada). Clonal analysis was performed on numbered cells and examined through the presence/absence of fluorophore and their location. A barcode was created as a binary signature (0 = absence, 1 = presence of cytoplasmic and nuclear marker of YFP, mKO, mCerulean mCherry, mTSapphire and EGFP) in all the animals analyzed. Since cell labeling and intensity in the maximal projection depend on the z-position of the cells, that could not be constant in relation to the intensity of the different reporter proteins, we followed specific criteria to avoid misleading. When the labelled cell does not exhibit a clear label of the cell nucleus, we just considered that this corresponds to a cytoplasmic labeling. On the other hand, when there is a clear limit discriminating the nucleus and the cytoplasm, within the same fluorescent channel, we consider the labeling to be both nuclear and cytoplasmic. Finally, the cells sharing the same combination/location of fluorophores/signature were catalogued as clones after classifying all the labeled cells.