In the lateral wall of adult mouse brain, Lrig1 expression identifies quiescent stem cells that largely do not contact the ventricle
To study the adult neural stem cells and their immediate progeny in the adult mouse brain, we analyzed the Tg(Nr2e1-EGFP) BAC transgenic mouse line from the GENSAT Project [32]. To our knowledge, this mouse line was not previously characterized. In this mouse line, the regulatory sequences of a stem cell marker gene Nr2e1 also known as Tlx [33] drive expression of EGFP marker protein in the neural stem cells and their immediate progeny. The hemizygous transgenic mice were analyzed at 13 weeks of age using an optimized whole mount immunofluorescence protocol and confocal imaging [21, 28]. To reveal the layer of ependymal cells and the gaps between them, we utilized an antibody against the S100 protein [20]. Consistent with previous work [34], this analysis revealed a subventricular or subependymal layer of EGFP+ cells, presumably neural stem cells and more differentiated precursor cells (Fig. 1A).
Immunoreactivities against the EGFP marker protein and proliferation markers such as KI-67 or ASCL1 were not mutually exclusive, with significant overlap (Fig. 1A). Most of the EGFP+ cells were KI-67+ and ASCL1+ (86 ± 2.5%, n = 4 ventricular walls from 4 mice), but the rest were KI-67- and ASCL1-. The cell densities were, all EGFP+ cells, 3905 ± 517 cells per mm2, n = 4 ventricular walls from 4 mice; KI-67+ ASCL1+ EGFP+ cells, 3373 ± 532 cells per mm2, n = 4 ventricular walls from 4 mice. The EGFP+ KI-67+ ASCL1+ triple-positive cells showed less intense EGFP immunoreactivity than the EGFP+ KI-67- ASCL1- cells. The observation of the EGFP+ KI-67- ASCL1- cells suggested the expression of Nr2e1 can indeed be used to identify quiescent stem cells in the lateral ventricular wall, consistent with single cell RNA sequencing analyses [13, 16, 17].
As aforementioned, the S100 immunoreactivity revealed the gaps between the ependymal cells where the apical extensions of the B1 type stem cells are located. Consistent with the notion that the EGFP+ KI-67- ASCL1- cells are quiescent B1 type stem cells, we could easily identify the EGFP+ KI-67- ASCL1- cells filling the gaps between the S100+ ependymal cells (Fig. 1B). The cell bodies of the EGFP+ KI-67- ASCL1- B1 type stem cells were in the subventricular zone under the layer of the S100+ ependymal cells. Thus, we could identify the previously known population of the quiescent B1 type stem cells using the whole mount method [21, 28].
Next, we analyzed the brains from our Lrig1T2A-iCreERT2 mouse line [21] using the same method. The Lrig1T2A-iCreERT2/+; Rosa26Ai14/+ mice were induced once with tamoxifen at 12 weeks of age then perfused 1 week after. During that 1 week, the mice were also administered a thymidine analog, ethynyl deoxyuridine (EdU), in the drinking water at 0.15 mg/ml. This labeling paradigm labels the Lrig1-lineage cells in the young adult brain, and also reveals how many of them are in S-phase at the time of their labeling. First, confirming our previous observations in Nam and Capecchi, 2020, analysis of the lateral walls revealed only a few RFP-labeled B1 type stem cells (5.0 ± 1.7 cells per mm2, n = 3 ventricular walls from 3 mice). Almost all of the RFP-labeled presumptive stem cells showed the α and β morphologies we described in the previous work (61.5 ± 11.6 cells per mm2, n = 3 ventricular walls). The RFP+ cells with the α and β morphologies were highly branched at the cell body in contrast to the B1 type stem cells and are referred to in this work as the Lrig1-lineage stem cells. Almost all of the Lrig1-lineage stem cells were EdU- (Fig. 1C, 94.2 ± 3.6%, n = 3 ventricular walls) indicating that these cells had not entered S-phase during the time of tamoxifen-induced RFP labeling.
Then, we analyzed the S100 and RFP immunoreactivities in the Lrig1 reporter mouse line as we analyzed the Tg(Nr2e1-EGFP) mouse line. In contrast to the Nr2e1 reporter mouse line, we did not observe many ventricle-contacting apical extensions between the S100+ cells from the Lrig1-lineage stem cells (Fig. 1C). Nevertheless, consistent with the locations of the EGFP+ KI-67- ASCL1- cells in the Tg(Nr2e1-EGFP) mice, the cell bodies of the Lrig1-lineage stem cells were also in the subventricular zone under the S100+ ependymal cells (Fig. 1C).
We then analyzed additional neural stem cell reporter mouse lines to determine whether the observation above holds true. The Coffey Lrig1creERT2 knock-in knock-out mouse line [35] and PdgfrbP2A-creERT2 knock-in mouse line [36, 37] were analyzed using the same method (Fig. 1D). When the locations of the RFP+ cells with the stem cell morphologies, i.e., branches or no branches, round cell body, and a long basal process, were quantitated, similar distributions were obtained from different reporter mice, without statistically significant differences (Fig. 1E, Student’s t test; Fig. 1F, Kolmogorov-Smirnov test). Therefore, with or without the apical extension that contacts the ventricle, the adult stem cells in the adult mouse brain ventricular-subventricular zone have their cell bodies located in the subventricular zone under the S100+ ependymal cells. Finally, the Pdgfrb reporter labeled more ventricle-contacting cells than the Lrig1 reporters (Fig. 1G, Chi-square test, p < 0.01).
The Lrig1-lineage stem cells in the adult brain largely do not proliferate during juvenile development
Above, we molecularly defined a population of quiescent neurogenic stem cells in the subventricular zone of the lateral wall. We then examined whether these Lrig1-lineage stem cells had proliferated earlier during juvenile development using thymidine analog labeling. Control experiments were performed first.
Previously, an EdU dose titration was performed with 12 week old mice [26], and we tested the same low dose of EdU on juvenile mice (postnatal day 21 to postnatal day 28). The administration of the EdU at 0.15 mg/ml in drinking water did not disrupt expression patterns of proliferation marker KI-67 and neuroblast marker DCX in the lateral wall (Fig. 2A-B, n = 3 ventricular walls from 3 mice). Next, we quantitated how much of the EdU labeling is diluted out by the continuous proliferation in the lateral wall. Mice were administered EdU for 1 week, then chased up to 14 weeks of age. Although the density of the EdU+ cells (the number of EdU+ nuclei / mm2) decreased by approximately half by 14 weeks, a significant number of EdU+ cells remained (Fig. 2C, n = 3–4 ventricular walls from 3 to 4 mice per time point). Because EdU toxicity can reduce proliferation, we then examined whether the EdU+ cells are in fact continuing to divide using a second pulse of BrdU. After sequential 1 week pulses of EdU then BrdU, mice were examined at 8 and 14 weeks of age. Decreasing percentage of EdU+ BrdU+ / EdU+ nuclei from 8 to 14 weeks indicated that the EdU+ cells were in fact continuing to divide, suggesting minimized toxicity from the low dose EdU pulse (Fig. 2D, n = 4–6 ventricular walls from 4 to 6 mice).
The Lrig1T2A-iCreERT2/+; Rosa26Ai14/+ mice were pulsed with EdU for 1 week starting at 3 and 4 weeks of age, then injected with tamoxifen at 12 weeks of age to label the Lrig1-lineage stem cells. The lateral walls from these mice were analyzed as above. This analysis revealed very low percentage of RFP+ EdU+ cells / EdU+ cells, indicating that more than 99% of the previously proliferated EdU+ cells were not labeled by the Lrig1 reporter allele (Fig. 2E-F, n = 5–6 ventricular walls from 5 to 6 mice).
To determine whether an earlier tamoxifen induction reveals greater percentage of RFP+ EdU+ double-positive cells, the Lrig1T2A-iCreERT2/+; Rosa26Ai14/+ mice were administered EdU from 3 to 4 weeks, then tamoxifen was injected at 4 and 5 weeks of age to label the Lrig1-expressing cells. Analysis of the lateral walls from these mice after 1 week indicated that even with an earlier tamoxifen induction, almost all EdU+ cells were still RFP- as before (Fig. 2G, 0 RFP+ EdU+ cells per mm2 / 613 ± 264 EdU+ cells per mm2, n = 3 mice per time point).
Finally, we repeated and extended the analysis above with additional time points of EdU administration. EdU was administered starting from 3 weeks of age to 11 weeks of age. Then, tamoxifen was injected at 13 weeks of age and mice perfused after 1 week. The percentage of RFP+ EdU+ cells / EdU+ cells was again less than 1% across the entire time points (Fig. 2H, n = 5–7 ventricular walls from 5 to 7 mice).
Although only less than 1% of the EdU+ cells were RFP+ EdU+ double-positive cells, we identified the rare singlet RFP+ EdU+ double-positive cells to document their cellular phenotype. The RFP+ EdU+ cells were KI-67- (Fig. 3A-B), and therefore they were label-retaining cells, i.e., stem cells that had proliferated in the past but were not proliferating at the time of perfusion. These cells showed a long basal process without many branches at the cell body. Thus, the morphology of these RFP+ EdU+ cells clearly differed from the morphology of the Lrig1-lineage stem cells. Rare doublet cells were also identified (Fig. 3C).
Morphologically distinct populations of stem cells can be labeled with the Lrig1
T2A-iCreERT2 allele at juvenile age
Because the EdU pulse-chase during juvenile development and tamoxifen induction in adults did not identify many RFP+ EdU+ double-positive cells, we analyzed the Lrig1-expressing cells from an earlier age to identify proliferating cells. Thus, the Lrig1T2A-iCreERT2/+; Rosa26Ai14/+ mice were induced with tamoxifen at postnatal day 21 (3 weeks of age). The analysis of the Lrig1T2A-iCreERT2/+; Rosa26Ai14/+ mice during juvenile development revealed at least 2 morphologically distinct populations. When tamoxifen induced at 3 weeks and analyzed at 4 weeks, we could easily identify many RFP-labeled cells that fit the description of the canonical B1 type stem cells, i.e., the cells with a round cell body, few branches, a long basal process, and an apical extension that contact the ventricle (Fig. 4A, n = 10 ventricular walls from 10 mice). In addition to the B1 type stem cells, we also identified RFP-labeled cells that were similar to the Lrig1-lineage stem cells that we observed in the adult brain, i.e., the cells with many branches at the cell body, no ventricular contact, and a long basal process (Fig. 4A).
We then analyzed the mice induced during postnatal day 21 at 13 weeks of age. At this age, we could still observe the two morphotypes of the RFP+ stem cells, i.e., the first one that contacts the ventricle as well as the second one that does not and is branched (Fig. 4B, n = 10 ventricular walls from 10 mice).
The locations of the two morphotypes of the RFP+ stem cells – i.e., the B1 type stem cells and the Lrig1-lineage stem cells – were quantitated from 3 weeks of age to 4 and 13 weeks of age (Fig. 4C, Student’s t test, p < 0.01). The change in the distribution of cells’ locations over time suggested divergent fates for the cells (Fig. 4D, Kolmogorov-Smirnov test, p < 0.01, Fig. 4E, Chi-square test, p > 0.05). Next, the numbers of branches in the RFP+ stem cells were quantitated (Fig. 4F). For the Lrig1-lineage stem cells, the number of branches did not increase significantly during juvenile development (Student’s t test, p > 0.05), suggesting that these cells were already in place by 4 weeks of age. However, for the ventricle-contacting B1 type stem cells, (1) they showed fewer branches than the Lrig1-lineage stem cells at 4 weeks of age (Student’s t test, p < 0.01) and (2) the number of the branches increased from 4 to 13 weeks (Student’s t test, p < 0.01). In other words, at 13 weeks of age, even the ventricle-contacting B1 type stem cells showed more branches than at an earlier time point.
Selective proliferation of a stem cell subset during juvenile development
The analyses above suggested different behaviors of the two stem cell subsets in the ventricular-subventricular zone leading to divergent fates. We determined whether cellular proliferation is different among these subsets. Thus, after tamoxifen induction of the Lrig1 reporter mice at 3 weeks of age, EdU was administered for 1 week until 13 weeks of age. The mice were perfused at the end of the EdU administration, such that all S-phase cells could be detected without the confounding EdU label dilution over time. Again, the B1 type stem cells as well as the Lrig1-lineage stem cells were revealed (Fig. 5A, n = 3–5 ventricular walls from 3 to 5 mice per time point). Surprisingly, most of the RFP+ stem cells of both subsets still had not gone through S-phase from 3 to 12 weeks of age (week 3 time point quantitation, 42.2 ± 7.6 RFP+ cells per mm2, 1.0 ± 0.91 RFP+ EdU+ cells per mm2, 2.4 ± 2.3% RFP+ EdU+ cells / RFP+ cells).
Thus, we searched for the rare singlet RFP+ stem cells that were also EdU+. The very rare singlet RFP+ EdU+ double-positive stem cells that were found and imaged at higher magnification were all B1 type stem cells (6 cells from the 10 mice of the first two time points). The cells that initially contacted the ventricle with an apical extension were EdU- (Fig. 5B). They entered the cell cycle to incorporate the EdU into their genomic DNA (Fig. 5C). Subsequently, these cells completed mitosis to form doublets (Fig. 5D). Intriguingly, we did not find any singlet RFP+ EdU+ double-positive stem cells that did not contact the ventricle and were branched at the cell body.
Origin of the distinct stem cell subsets in the postnatal brain
Because the distinct morphotypes of the largely quiescent stem cells were already in place at 4 weeks of age, we determined whether they originate from the same pool of progenitor cells earlier on, or whether there exist entirely different progenitor pools for the different morphotypes of adult stem cells. We injected tamoxifen to newborn Lrig1T2A-iCreERT2/+; Rosa26Ai14/+ pups at postnatal day 0 then analyzed the mice at 1 week, 4 weeks, or 13 weeks of age. When analyzed at 1 week, we observed many RFP+ postnatal radial glial cells [24] that showed a round cell body and a very long basal process (Fig. 6A, n = 5 ventricular walls from 5 mice). The RFP+ postnatal radial glial cells were all VCAM1+ [38] and mostly KI-67- (RFP+ KI-67- cells, 92.3 ± 3.0% of all RFP+ postnatal radial glial cells, 120.1 ± 44.2 cells per mm2), suggesting they were quiescent at the time of perfusion (Fig. 6B). Importantly, we did not observe cytoarchitecturally distinct subsets of the RFP+ postnatal radial glial cells – all of the RFP+ postnatal radial glial cells had cell body under the ventricular wall and contacted the ventricle with an apical extension (Fig. 6C).
The analyses of the Lrig1 reporter mice at later time points showed the RFP+ stem cells of both morphotypes (Fig. 6D-E, n = 10 ventricular walls from 10 mice per time point). Thus, both subsets of the stem cells originated from a single pool of postnatal radial glial cells. Commencing between postnatal day 0 and 4 weeks of age, the single pool bifurcated into two pools, one ventricle-contacting and less branched and one non-ventricle-contacting and highly branched. Indeed, we observed that some RFP+ postnatal radial glial cells at 1 week of age already showed more branches than the rest (Fig. 6F-G) consistent with the notion that the branched postnatal radial glial cells will later become the branched Lrig1-lineage stem cells.
Fate of the stem cell subset that proliferates during postnatal/juvenile development
We further analyzed the proliferative behaviors of the distinct subsets. The Lrig1T2A-iCreERT2/+; Rosa26Ai14/+ mice were tamoxifen induced at postnatal day 0, administered EdU at 3 weeks of age for 1 week, then they were perfused at 4 weeks or 13 weeks of age. When analyzed at 4 weeks, as above, the RFP+ EdU+ stem cells were again rare (6.7 ± 2.6 cells per mm2, 6.3 ± 1.7% RFP+ EdU+ cells / RFP+ cells, n = 15 ventricular walls from 15 mice). However, consistent with a proliferative history during postnatal/juvenile development, we observed doublets of RFP+ stem cells that were EdU- (Fig. 7A) or EdU+ (Fig. 7B). Next, when analyzed at 13 weeks, we observed extremely rare singlet RFP+ EdU+ double-positive cells (Fig. 7C, 3 cells from n = 22 ventricular walls from 22 mice). The morphology of the singlet RFP+ EdU+ cells differed from the stem cells already described above. They were not branched at the cell body and showed a more complex basal process.
Lrig1 knock-out results in hyperproliferation without increased apoptosis
Taken together, the lineage tracing analyses above suggested that Lrig1 is expressed in all known subsets of quiescent stem cells through postnatal and juvenile development. We then analyzed the lateral walls of Lrig1 knock-out mice for cellular phenotypes to assess its function. Two mouse lines with different Lrig1 null alleles were utilized. First was the Coffey Lrig1creERT2 allele [35] that replaces the first exon sequence encoding the membrane-localizing signal sequence with the creERT2 cDNA. Second was the EUCOMM Lrig1Δ allele that results in a non-sense frame-shift prior to the protein’s transmembrane domain because of cre recombination of a floxed Lrig1 allele [39]. The Coffey knock-out and wildtype littermate mice were analyzed at 1 year and 3 months and at 2 years and 4 months. The EUCOMM knock-out and wildtype littermate mice were analyzed at 1 year and 3 months. To assess any confounding effect of the genetic background, the Coffey knock-out and wildtype littermate mice from a later generation of backcross to the C57BL/6J background were analyzed at 1 year and 8 months.
Remarkably, we observed increased numbers of KI-67+ cells and ASCL1+ cells in every knock-out mouse at every age (Fig. 8A-J, Student’s t test). Although the numbers of all DCX+ neuroblasts could not be counted directly because of the compacted DCX staining pattern, the numbers of early neuroblasts (i.e., the KI-67+ cells that are ASCL1- DCX+) were increased in the knock-out mice. The GFAP immunoreactivity was unchanged in the knock-out mice (Additional file 1 A-B). The same phenotype was observed in the Coffey and the EUCOMM knock-out mice (Fig. 8A-E, Additional file 2 A-C). The phenotype was not altered by the increasing contribution of the C57BL/6J genetic background (Fig. 8A-C, K-M).
Then, we analyzed whether apoptosis is increased because of the increased proliferation in the lateral wall stem cell niche. Analysis of cleaved CASP3 revealed no difference between the Coffey and EUCOMM knock-out and wildtype littermate mice (Fig. 8N, Additional file 2 D, Student’s t test, p > 0.05). Finally, we did not observe any tumors in the brains of these mice.