Lrig1 mRNA identifies stem cells in the adult V-SVZ neurogenic lineage
In addition to our line of investigation (Additional file 1), others have also previously observed Lrig1 expression in the V-SVZ quiescent neural stem cells (qNSC’s) [12]. To corroborate these observations and to determine whether Lrig1 would be a useful marker for further investigations, we analyzed the single cell RNA sequencing datasets in the public domain [13, 22, 23]. Bioinformatic analysis of the three datasets with established software programs [38, 39] yielded similar results. Here, we present our analysis of the largest dataset from [23].
Using the Seurat program, we first clustered the cells in an unbiased way. This step yielded distinct clusters (Fig. 1a) as well as subclusters within some clusters (Fig. 1b). The identity of the cells within each cluster was revealed by plotting the expression levels of known marker genes from [23] (Fig. 1c). Then, we confirmed the neurogenic lineage progression within two clusters by plotting additional known marker genes (Fig. 1d). For example, Agt expression was highest in the astrocyte cluster then high in the qNSC cluster. Lrig1 was expressed in a pattern very similar to Agt (Fig. 1e).
To determine in what subclusters Lrig1 expression is the highest, we calculated the mean expression level within each subcluster. To verify the accuracy of the result, several well-known marker genes were also analyzed: Nr2e1 [40, 41], Ascl1 [42], and Dcx. As expected, this analysis revealed that Nr2e1 and Ascl1 expression levels peaked in activated stem cells, and Dcx expression level peaked in neuroblasts (Fig. 1f-g). In contrast, Lrig1 expression level was highest in the astrocyte subcluster, high in the qNSC then steeply decreased along the neurogenic lineage.
To extend this analysis, we constructed a “pseudotime” lineage of the neurogenic lineage cells using the Monocle program (Fig. 1h). The accuracy of the ordering was confirmed by comparison to the ordering in [25] (Fig. 1i). The gene expression trends in the two pseudotime lineages were identical. With the ordered cells, we generated a pseudotime gene expression heatmap. In this analysis, Lrig1 transcript levels were the highest in the cells ordered first in the pseudotime (i.e., in the astrocytes) and remained high (i.e., in the qNSC’s) until the Thbs4+ activated stem cells (Fig. 1j). This was consistent with the trend obtained from the manual calculation of the expression levels in the neurogenic lineage subclusters. Taken together, these analyses suggested that the stem cells at the earliest steps of neurogenic differentiation express high levels of Lrig1.
Lrig1T2A-iCreERT2 knock-in allele
The analysis above suggested Lrig1 could be a useful marker gene with which to analyze the V-SVZ neurogenic stem cells in vivo. Mouse lines were generated to determine the effectiveness of Lrig1 as a genetic marker gene. Previously, two “knock-in knock-out” transcriptional reporter lines of Lrig1 were generated [18, 19]. There is no haploinsufficiency phenotype because of Lrig1 heterozygosity (also see below). Nevertheless, we generated a non-disruptive co-translational reporter allele using the 2A ribosome skip sequence [43] because this design allows multiple reporter alleles [44], even of the same gene. A T2A-sfGFP-iCre-ERT2-FNF cassette was knocked in between the end of the coding sequence and the 3′ untranslated region while removing the endogenous stop codon (Fig. 2a). Utilizing a similar design, an Lrig1T2A-tdTomato allele and a Cdk6T2A-td-sfGFP allele were also generated.
To characterize the sfGFP-iCre-ERT2 recombinase protein we utilized, cell-based assays were performed with the NIH/3T3 transformed mouse fibroblasts. First, as expected, the linker between sfGFP and iCre-ERT2 had no ribosome skip activity. TdTomato fused to H2B-sfGFP with that linker localized to the nucleus [45] (data not shown). Second, the fusion of the iCre-ERT2 fragment to sfGFP with that linker almost completely abolished the sfGFP fluorescence in the cells (data not shown). Replacing the linker with a synthetic (Glycine-Serine-Alanine) × 9 linker did not restore the sfGFP fluorescence (data not shown). This suggested the iCre-ERT2 rather than the linker was destabilizing. Third, comparing the 4-hydroxytamoxifen induced cre activity revealed no difference in activities between the three variants (Fig. 2b). Thus, the expression of the sfGFP-iCre-ERT2 fusion protein from the CAG promoter resulted in faint sfGFP fluorescence in the cells, possibly due to short half-life of ERT2 [46, 47] and/or iCre in the cell, but its tamoxifen-inducible recombinase activity was identical to the activity of iCre-ERT2.
The targeting vector (Fig. 2c) was constructed. Next generation sequencing of the entire targeting vector confirmed the vector sequence (Fig. 2d). To transmit the Lrig1T2A-iCreERT2 allele through the germline, we utilized the G4 129 × B6 F1 hybrid embryonic stem cells [26] in which one of the two sets of autosomes are of the B6 background. Southern blot hybridizations of the ES cells’ genomic DNA (Fig. 2f) revealed many positive clones with Lrig1 3′ region targeted in one of the two chromosomes (Fig. 2g). Importantly, a SNP assay [48] revealed that the B6 Lrig1T2A-iCreERT2 targeting vector recombined into Lrig1 on a B6 chromosome (Fig. 2h). The mice resulting from these ES cells were backcrossed to the C57BL/6J background for six generations. When heterozygotes of this C57BL/6J congenic mouse line were interbred, normal homozygotes could be weaned at Mendelian ratios (Fig. 2i-j). Furthermore, these homozygotes could breed normally with each other and generate normal size litters (data not shown), suggesting that the homozygosity of this allele does not result in adverse developmental effects even in the inbred background. This result also confirmed that Lrig1 must have been knocked in initially and carried through the generations.
Lrig1T2A-iCreERT2 knock-in allele is non-disruptive
We sought to establish that our Lrig1 knock-in allele is not a hypomorph. The Lrig1 protein is processed into multiple fragments making it difficult to quantify by Western blots. Thus, the dosage analysis paradigm from classical genetics was utilized. Because Lrig1 heterozygous mice are normal (see below), we could generate trans-heterozygous mice, i.e., mice with the Lrig1T2A-iCreERT2 allele over a null allele (the Coffey Lrig1creERT2 allele [19]). If the Lrig1T2A-iCreERT2 allele is normal, the reduction in the Lrig1 dosage from the null allele is “complemented” by this allele, and no phenotypes should be evident. As for the phenotypes to quantify, a behavioral and a cellular phenotype were sought in the Lrig1 knock-out mice.
Mouse behaviors were measured in an unbiased way with the commercial Laboras platform (https://www.metris.nl/en/products/laboras/). Measurement over 3 days suggested that the platform could measure changes in mouse home cage behaviors (data not shown). Essentially, after the mice are put into the Laboras platform cage, the mice explore the new cage environment over the first 12 h, then are acclimated in the next day and a half. At that point, the mice can be stimulated by caffeine administration in drinking water (data not shown). In this paradigm, the Lrig1 knock-out mice showed statistically significant differences during the first 12 h in some of the measures related to exploratory behavior. Increases in the distance traveled and rearing (i.e., standing up on hindlimbs and looking around at the edge of the cage) were observed (Fig. 2k’’-k’’’). Next, we measured cellular phenotypes in the lateral wall using protein markers of stem cell proliferation and neurogenesis. Preliminary experiments suggested that our immunostaining protocol revealed all Ki-67+, all Mcm2+, all Ascl1+, and all Dcx+ cells in the lateral wall (Additional file 2). Lrig1 knock-out mice showed increased proliferation in the V-SVZ (data not shown, a manuscript in preparation) consistent with the previous reports on the intestinal crypt stem cell niche [19, 49].
Importantly, the increases in the distance traveled and rearing were absent in the Lrig1 heterozygous mice and the trans-heterozygous mice. Furthermore, the counts of Ascl1+ nuclei and Ki-67+ nuclei as well as Dcx+ pixel areas in the trans-heterozygous mice were comparable to wildtype mice when young and old (Fig. 2l-q”). These indicated that the Lrig1T2A-iCreERT2 allele must lead to production of functional Lrig1 protein because the trans-heterozygous mice don’t show the behavioral and cellular phenotypes observed in the knock-out mice. Thus, we infer that our knock-in Lrig1T2A-iCreERT2 allele is not a hypomorph.
We also compared the expression pattern of Dcx through aging in wildtype and trans-heterozygous mice uninjected with tamoxifen and Lrig1T2A-iCreERT2/+; ROSA26Ai14/+ mice injected tamoxifen once (Fig. 2q” and Fig. 3g). There were no measurable differences. Taken together, these analyses established that the mouse line that carries the non-disruptive Lrig1T2A-iCreERT2 allele is phenotypically indistinguishable from C57BL/6J mice in the measurements we performed when young and old.
Characterization of the Lrig1T2A-iCreERT2 allele
Lrig1 is expressed in the skin and the intestinal crypt stem cell niche, as well as in other organs [18, 19, 50]. Tamoxifen induction in developing embryos and in adults showed that skin and intestinal crypt could indeed be labeled (Fig. 2r-s’) consistent with the known Lrig1 expression domain.
Then, we characterized the allele expression in the brain. Lrig1 is expressed in many different cell types in the brain in addition to astrocytes [50,51,52]. The pattern of Lrig1T2A-iCreERT2 activity as visualized by the ROSA26Ai14 reporter matched the known expression pattern (Fig. 2t-t”). Of note, no neurons were labeled in the olfactory bulb at 3 days after tamoxifen injection (Fig. 2t”).
Next, we attempted to determine whether the sfGFP-iCre-ERT2 is co-expressed with Lrig1 in the V-SVZ cells. Interestingly, although we could readily detect sfGFP by indirect immunofluorescence with the anti-GFP antibody and confocal microscopy on brains from the Cdk6T2A-td-sfGFP mouse line, the sfGFP-iCre-ERT2 from the Lrig1T2A-iCreERT2 allele was not detectable (Fig. 2u-u’). To establish that Lrig1 is expressed in the V-SVZ cells, we turned to a very sensitive Lrig1T2A-tdTomato reporter allele utilizing the extremely bright red fluorescent protein tdTomato (similar to [53]). By flow cytometry, this mouse line revealed several discrete cell populations with tdTomato fluorescence from the Lrig1 locus, demonstrating that Lrig1 is indeed expressed in the V-SVZ cells (Fig. 2u”). However, although we could again detect the sfGFP signal from the Cdk6T2A-td-sfGFP allele by flow cytometry, we still could not detect any sfGFP signal from the Lrig1T2A-iCreERT2 allele (Fig. 2u”). These suggested that (1) Lrig1 protein expression level is low in the V-SVZ cells and (2) the sfGFP-iCre-ERT2 reporter fluorescence is not very sensitive either (as described above). Consistent with low Lrig1 protein level, an indirect immunofluorescence with an anti-Lrig1 antibody and confocal microcopy also did not reveal any signal in the V-SVZ (Fig. 2v).
Thus, we next utilized flow cytometry to detect and measure co-labeling of the RFP lineage label and the endogenous Lrig1. First, using the Lrig1T2A-tdTomato allele we determined that the commercial anti-Lrig1 antibody indeed labels Lrig1-T2A-tdTomato+ cells (Fig. 2w). Conversely, all of the Lrig1-T2A-tdTomato+ cells in the Lrig1T2A-tdTomato/+ mice were labeled by the anti-Lrig1 antibody (Fig. 2w). Next, we detected the RFP+ cells in the tamoxifen-induced Lrig1T2A-iCreERT2/+; ROSA26Ai14/+ mice and observed that almost all of the RFP+ cells were labeled by the anti-Lrig1 antibody (Fig. 2x). This indicated that the RFP lineage labeling occurs only in Lrig1+ cells in the Lrig1T2A-iCreERT2/+; ROSA26Ai14/+ mice.
To characterize the kinetics of labeling, we injected tamoxifen into a cohort of 3 month-old Lrig1T2A-iCreERT2/+; ROSA26Ai14/+ mice once on the same day. Then, we analyzed at progressively longer time points the lateral walls from these mice by whole mount immunofluorescence and confocal imaging. When quantitated, the number of singlet cells plateaued at 7 days (Fig. 2y), indicating that the tamoxifen-induced in vivo labeling of cells with RFP was complete by day 7. That is, no additional cells were labeled after 7 days, but new RFP+ cells were born from the stem cells already labeled with RFP. Thus, although the cell number is under-sampled, day 3 is a good time point to assess the initially labeled cell population because the new cells are not born yet.
We note that the same complement of cell types (see Fig. 4a-g’) was consistently labeled by the Lrig1T2A-iCreERT2 allele with the low (80 mg/kg) or high (230 mg/kg) tamoxifen doses. Using the ROSA26Ai14 reporter line [30], labeling of certain cell types could not be intentionally excluded by lowering the tamoxifen dose. Thus, despite the low Lrig1 level, all of the different Lrig1+ cell types in the V-SVZ expressed levels of the recombinase sufficient for ROSA26Ai14 reporter recombination at the tamoxifen doses utilized (80–230 mg/kg). There could nevertheless be a bias against labeling of cells with lower recombinase levels. For example, with very low tamoxifen doses (20–40 mg/kg), it was possible to obtain RFP labeling that was stochastically devoid of the rare Ki-67+ stem cells (data not shown), suggesting that the proliferating stem cells express lower level of Lrig1 and recombinase relative to the quiescent stem cells.
Finally, we compared our Lrig1T2A-iCreERT2 allele to the Coffey Lrig1creERT2 allele. The V-SVZ cell types labeled with our allele (see Fig. 4a-g’) were also labeled with the Coffey allele (data not shown). Taken together, these characterizations indicate that (1) the mice that carry the non-disruptive Lrig1T2A-iCreERT2 allele are indistinguishable from wildtype mice, (2) the Lrig1T2A-iCreERT2 allele faithfully reports Lrig1+ cells, and (3) the labeling by the Lrig1T2A-iCreERT2 allele is also indistinguishable from an independent mouse line from another laboratory utilizing the same Lrig1 locus to drive the reporter gene.
Consistent labeling of non-proliferative cells with the Lrig1T2A-iCreERT2 allele
To trace and quantitate the neurogenic output from the Lrig1+ cell lineages, we determined whether our knock-in allele allows consistent labeling. Three days after tamoxifen induction, whole mount immunofluorescence, confocal microscopy, and quantitation revealed consistent numbers of RFP+ cells labeled within each induction cohort and between different induction cohorts (Fig. 3a-b).
Next, because Lrig1 is known to regulate quiescence in other stem cell systems, we determined whether non-proliferative cells are preferentially labeled with the Lrig1 reporter mice. Ki-67, Mcm2, and Ascl1 are markers of proliferating cells in the V-SVZ. Whole mount immunofluorescence analysis on the lateral walls from the tamoxifen-induced Lrig1 reporter mice revealed that among the RFP+ cells, Ki-67+ or Mcm2+ or Ascl1+ RFP+ cells were extremely rare (Fig. 3b), meaning most of the initially labeled RFP+ cells were non-proliferative.
The neurogenic Lrig1+ cell lineage is active throughout adult life
Whether the non-proliferative Lrig1+ cells in the V-SVZ labeled by the Lrig1T2A-iCreERT2 allele possessed neurogenic activity throughout adult life was examined. Analysis of the olfactory bulbs 6 months (Fig. 3c-f) and 1 year (data not shown) after induction with a single 120 mg/kg tamoxifen dose revealed RFP+ granule cells (Fig. 3c) and RFP+ periglomerular cells (Fig. 3d). Counts of the RFP+ NeuN+ cells in the granular cell layer indicated more than a hundred RFP+ NeuN+ granule cells per 50 μm section (Fig. 3e-f). These suggested that proliferative activities arising from the non-proliferative RFP-labeled stem cells in the lateral wall gave rise to the RFP+ olfactory bulb interneurons.
Before further analysis, as a control, we determined whether adult neurogenesis was at a steady-state at the time of tamoxifen induction. We measured the Dcx+ pixel area in the lateral wall at progressively longer time points from the time of tamoxifen induction (n = 4–5 lateral walls from 4 to 5 mice per time point). These measurements revealed the Dcx+ pixel area did not increase or decrease from 3 months of age to 4 months of age (3 days to 1 month after injection, Fig. 3g), suggesting adult neurogenesis was at a steady-state at the time of tamoxifen induction. Generally consistent with a previous report [54], we observed a gradual decrease in the Dcx+ pixel area during aging, perhaps reflective of decrease in adult neurogenesis. This decrease in the Dcx+ pixel area first became statistically significant at 1 year of age.
To analyze the dynamics of newborn RFP+ Dcx+ cells in the lateral wall, the Lrig1 reporter mice were induced once with a high dose of tamoxifen (230 mg/kg) at 3 months of age to label as many Lrig1+ cells as possible and perfused at progressively longer time points. The lateral walls were analyzed by whole mount immunofluorescence and confocal imaging (n = 4–5 lateral walls from 4 to 5 mice per time point). At day 3 after induction, very few RFP+ Dcx+ neuroblasts were detected (Fig. 3h). Significant numbers of RFP+ Dcx+ neuroblasts were first detected at 1 month post induction (mpi). At 3 mpi, the numbers of RFP+ Dcx+ neuroblasts had increased. By 6 mpi, the numbers of RFP+ Dcx+ neuroblasts had decreased. However, at 9 mpi, the numbers of RFP+ Dcx+ neuroblasts had stabilized. Notably, the numbers of RFP+ Dcx+ neuroblasts remained significant at even 1 year after induction, i.e., at 1 year and 3 months of age.
To analyze the dynamics of proliferating clusters, the Lrig1 reporter mice were induced once at 3 months with a low dose of tamoxifen (80 mg/kg) to label the Lrig1+ stem cells at low density. The mice were perfused at progressively longer time points to analyze the lateral walls by whole mount immunofluorescence and confocal microscopy as above (n = 3–10 lateral walls from 3 to 5 mice per time point). The numbers of RFP+ cell clusters and the numbers of cells in each cluster of RFP+ cells (cluster size) were scored (Fig. 3i). First, the numbers of RFP+ cell clusters appeared cyclic: high at 1 mpi, low at 3 mpi, then high again at 9 and 12 mpi. Second, the mean cluster size remained similar between 1 mpi and 3 mpi although there were outliers of larger clusters at 1 mpi. From 3 mpi on, in atypical distributions, small (< 20 cells) clusters were present at all time points while the outlier cluster sizes increased from 6 mpi to 9 mpi, then decreased at 12 mpi.
Whether the neurogenic Lrig1+ cell lineage continues to contribute significantly to neurogenesis in the old mice was also determined. Analysis of the percentage of RFP+ Dcx+ cells pixel area over Dcx+ cells pixel area (Fig. 3j-j”) suggested that at 1 year after induction, 37.8 ± 10.8% of the Dcx+ cells arose from the Lrig1+ cell lineages (mean ± S.D., n = 4 lateral walls from 4 mice). In comparison, at 1 mpi, 32.3 ± 5.91% of the Dcx+ cells arose from the Lrig1+ cell lineages (mean ± S.D., n = 4 lateral walls from 4 mice).
Heterogeneity among the Lrig1+ cells
The study of the Lrig1+ cell lineages in the lateral wall V-SVZ revealed robust neurogenic activity throughout adult life from this pool. As aforementioned, almost all of the RFP+ cells labeled by the Lrig1T2A-iCreERT2 allele were non-proliferative, suggesting that if any stem cells are labeled by this allele, they are likely to be in a quiescent state. Thus, we examined the non-proliferative RFP+ cells in greater detail using thymidine analog EdU to identify the quiescent neurogenic stem cells among the Lrig1+ cells. Lack of EdU incorporation (the EdU- cells) means that the cells had not gone through an S-phase and can be inferred to have remained quiescent, at least for the duration of the EdU administration. Because the RFP-labeled stem cells were activated in significant numbers starting at 1 mpi (Fig. 3h), the mice were induced once with tamoxifen at 3 months of age, then thymidine analog EdU was administered for 7 days at 3 months after induction to label all proliferating cells. The lateral wall whole mounts were analyzed for RFP and Ki-67 immunoreactivities and EdU incorporation. This revealed the entire repertoire of the RFP+ cells: EdU- Ki-67- quiescent cell types (Fig. 4a-g’) as well as EdU+ Ki-67+, Ascl1+, or EdU+ Ki-67+ Dcx+ proliferating cells (data not shown).
In sum, we identified two morphological subtypes of potential quiescent neurogenic stem cells among the RFP+ EdU- Ki-67- cells, categorized by morphometry of cell depth (superficial vs. deep), number of branches (more or less than 4), and the length of the basal process (short or long). We named these morphotypes of Lrig1+ cell lineages α (Fig. 4a-a’) and β (Fig. 4b-b’). The α and β morphotypes are similar in that they both have branches, but different in that the β morphotype has a shorter basal process. Other cell types that were also labeled (Fig. 4c-g’) were excluded as candidates for stem cells because (1) these cells are known not to be of neurogenic stem cell lineages, and (2) we did not observe proliferating clusters of these cells during our experiments We counted from 4 lateral walls from 4 mice: 297 α morphotype cells; 159 β morphotype cells; 15 tanycytes; 12 striatal astrocytes; 219 ependymal cells; 3 neurons; 162 mural cells.
Cell cycle entry of the Lrig1+ neurogenic stem cells
Having identified the potential quiescent Lrig1+ stem cells in the V-SVZ, we again used proliferation markers, this time to identify the Lrig1+ cells that proliferate and give rise to neurogenic progeny – the minimum criteria for a neurogenic stem cell identity. The tamoxifen induced Lrig1 reporter mice were administered EdU for 7 days at 3 mpi. Then, the mice were perfused after a 14 days chase. After the chase, the EdU-labeled neuroblasts had migrated out of the V-SVZ to the olfactory bulbs, and sparse EdU-labeled stem cells (cells that had previously entered S-phase) remained in the lateral wall (Fig. 4h-i, 5161 cells counted, 5 lateral walls from 4 mice). Among these EdU+ cells, we observed very rare EdU+ RFP+ cells with the α/β morphologies (Fig. 4h-i, 21 cells counted, 5 lateral walls from 4 mice), suggesting that the RFP+ cells with these morphologies are the only Lrig1+ cells that can enter the cell cycle. We did not observe any label retention among any other RFP+ non-stem cell types (i.e., the cells in Fig. 4c-g’, 5 lateral walls from 4 mice), suggesting that these cells do not proliferate often at steady-state at this age. We also infer that these cells cannot be the neurogenic stem cells.
The retention of EdU label only in the α/β morphotype cells meant that only these cells had entered S-phase during the EdU administration, fulfilling a criterion for a stem cell identity. Thus, we searched for the Lrig1+ stem cells at an earlier phase of neurogenesis. Specifically, we looked for EdU+ Ki-67+ Ascl1+ RFP+ activated stem cells before the 2 weeks of chase. The tamoxifen induced Lrig1 reporter mice were administered EdU for 7 days, then analyzed immediately after. Again, the non-stem cell types were not labeled with EdU, and we reiterate that most EdU-labeled cells were EdU+ Ascl1+ RFP+ transit amplifying cells and EdU+ Dcx+ RFP+ neuroblasts. However, we observed very rare singlet RFP+ cells with α/β morphologies that were EdU+ Ascl1+ and Ki-67+ (Fig. 4j, 38 cells counted from 7 lateral walls from 7 mice), suggesting that these cells are the quiescent RFP+ cells that entered the cell cycle to proliferate. Expression of Ascl1 demonstrated that these activated stem cells were in the neurogenic lineage. The singlet EdU+ Ascl1+ Ki-67+ RFP+ cells showed large nuclei, suggesting that the increased nuclear size is a characteristic of the activated neurogenic stem cells. The percentage of the RFP-labeled activated stem cells among all RFP-labeled stem cells was ~ 2.3% (~ 4.2 EdU+ RFP+ label-retaining cells per lateral wall, ~ 5.4 EdU+ Ascl1+ Ki-67+ RFP+ cells with large nucleus per lateral wall, ~ 211 RFP+ α/β morphotype cells per lateral wall).
Spatial distribution of the Lrig1+ neurogenic stem cells in the lateral wall
The analysis above suggested that the RFP+ α/β morphotype cells could be the Lrig1+ quiescent neurogenic stem cells. To determine whether the Lrig1 expression identifies all spatial subtypes of the V-SVZ neurogenic stem cells [55], we analyzed the spatial distribution of the α/β morphotype stem cells in the lateral wall. Mice were induced sparsely such that the full morphology of each cell could be visualized without compromise. After whole mount immunofluorescence, we performed high magnification imaging of the entire lateral walls (Fig. 4k, Additional file 3). The α/β morphotype stem cells (Additional file 4) were observed throughout the lateral wall (Fig. 4l-m, n = 5 mice), suggesting that all spatial subtypes of neurogenic stem cells are labeled by this driver.
Ara-C infusion induced activation of the Lrig1+ neurogenic stem cells
The RFP+ cells of the α/β morphotypes were the only cells observed to enter the cell cycle, suggesting that these cells are the Lrig1+ quiescent neurogenic stem cells. Furthermore, these cells were observed throughout the lateral wall. We further tested the quiescence and subsequent activation of the α/β morphotype stem cells via infusion of the nucleoside anti-mitotic Ara-C. As a classical chemotherapeutic, Ara-C infusion kills the dividing neurogenic cells in the V-SVZ, and concomitantly induces regeneration of the neurogenic lineage from the remaining qNSC’s [56]. In our implementation of this paradigm, we infused Ara-C into the cerebrospinal fluid of the lateral ventricle [15] rather than into the cortical parenchyma [56] for more rapid kinetics. In our implementation, temporary up-regulation of Gfap in the ependymal cells was observed, but olfactory bulb interneurons were nevertheless generated from the V-SVZ after the infusion, as determined by thymidine analog pulse-chase.
We labeled the Lrig1+ cells with a single injection of tamoxifen. Fourteen days after the induction, mice were infused with Ara-C for 6 days. After the infusion, whole mount immunofluorescence and confocal imaging of the lateral wall revealed RFP+ cells that survived the infusion and persisted (Fig. 5a, 1458 α/β morphotype cells counted from 6 lateral walls from 6 mice), indicating that these RFP+ cells were quiescent. Among these quiescent cells would be the quiescent stem cells.
To identify the quiescent neurogenic stem cells, in another cohort of mice, we ceased the infusion after 6 days then analyzed the lateral wall after waiting a day when the RFP+ quiescent stem cells were activated to participate in the subsequent regeneration (Fig. 5b). Interestingly, we observed evenly spaced Ki-67+ nuclei at this time point. Among the RFP+ cells, some but not all were also Ki-67+ (Fig. 5b-f). Singlet Ki-67+ RFP+ cells showed large nuclei. The numbers of Ki-67+ RFP+ cells were increased over the uninfused mice at steady-state (Ara-C-induced, 219 cells counted from 3 lateral walls from 3 mice; steady-state, 38 cells counted from 7 lateral walls from 7 mice). The Ki-67+ RFP+ cells were exclusively of the α/β morphotype (Fig. 5g-h), indicating that only these subtypes could enter the cell cycle in response to the Ara-C infusion. The Ki-67+ RFP+ cells were observed throughout the anterior-posterior and dorsal-ventral axes (data not shown). We did not observe any other RFP+ cell types with Ki-67 expression.
At later time points, clusters of RFP+ cells that co-labeled with the neurogenic lineage markers Ascl1 (Fig. 5i) and Dcx (Fig. 5j) could be observed, indicating that the Ara-C infusion-associated activation of the RFP+ quiescent stem cells results in neurogenic progeny and not simply glial scarring. Thus, the Lrig1+ α/β morphotype cells could enter the cell cycle from quiescence and generate neurogenic progeny, i.e., they fulfilled the minimum criteria for a neurogenic stem cell identity.
Location of the Lrig1+ neurogenic stem cells throughout the depth of the lateral wall
We determined whether the Lrig1+ neurogenic stem cells show phenotypes consistent with known V-SVZ stem cells. Lrig1 reporter mice were induced as above. Whole mount immunofluorescence, confocal imaging analyses (Fig. 6a-c), and measurements of z-stacks (in orthogonal view, the distance between the center of the RFP+ cell and the ventricular surface revealed by β-catenin immunoreactivity, Fig. 6c) revealed that, on average, the α/β morphotype stem cells were located deep in the lateral wall, but they were found throughout the depth of the lateral wall (Fig. 6d-e). As a comparison, the ventricle-contacting tanycytes were analyzed. As expected, these cells always contacted the ventricular surface. The differences in the cells’ locations between the α/β morphotype cells and tanycytes were statistically significant (Fig. 6e, α vs. tanycytes, p < 0.01, β vs. tanycytes, p < 0.05, t test).
Marker profile of the Lrig1+ neurogenic stem cells
Finally, the expression of known markers in the Lrig1+ neurogenic stem cells were characterized. Since the Lrig1+ α/β morphotype stem cells resemble oligodendrocyte precursor cells (OPC’s) morphologically, we examined a canonical marker of the OPC’s, Pdgfra [57], in lateral walls from tamoxifen-induced Lrig1 reporter mice. In the V-SVZ, the Pdgfra immunostaining revealed (1) signal near the ventricular surface (Fig. 6f') and (2) signal deep in the lateral wall (Fig. 6g'). At the surface, the RFP+ Dcx+ neuroblasts co-labeled with Pdgfra (Fig. 6f”). However, deeper in the wall, RFP did not co-label at all with Pdgfra (i.e., the OPC’s, Fig. 6g”), indicating that the Lrig1+ α/β morphotype stem cells are not OPC’s.
Next, analyses of the Gfap protein in the lateral walls from the tamoxifen-induced Lrig1 reporter mice (Fig. 6h-k”) surprisingly revealed only infrequent expression in the Lrig1+ stem cells (Fig. 6l). More than half of the α/β morphotype stem cells labeled by the Lrig1 reporter did not express the Gfap protein. In contrast, nearly all cells in the control group of tanycytes expressed the Gfap protein (Fig. 6l). Interestingly, this heterogeneity of Gfap expression was also suggested in the single cell RNA sequencing analysis. Gfap expression was scattered and non-uniform in the astrocyte and the qNSC clusters (Fig. 6m). Furthermore, the calculation of the mean Gfap expression level showed that the Gfap expression level was highest in the activated stem cells (Fig. 6n-o).