Lgr5 Marks Post-Mitotic, Lineage Restricted Cerebellar Granule Neurons during Postnatal Development

Wnt signaling regulates self-renewal and fate commitment of stem and progenitor cells in development and homeostasis. Leucine-rich repeat-containing G-protein coupled receptor 5 (Lgr5) is a co-receptor for Wnt signaling that marks highly proliferative stem and progenitor cells in many epithelial tissue types. Wnt signaling instructs neural developmental and homeostatic processes; however, Lgr5 expression in the developing and adult brain has not been characterized. Here we report that Lgr5 is expressed in the postnatal cerebellum during the maturation and synaptogenesis of cerebellar granule neurons (CGNs), processes controlled by Wnt signaling. Using a transgenic reporter mouse for in vivo Lgr5 expression analysis and lineage tracing, we reveal that Lgr5 specifically identified CGNs and was restricted temporally to the CGN maturation phase within the internal granule layer, but absent in the adult brain. Cells marked by Lgr5 were lineage restricted, post-mitotic and long-lived. The ligand for Lgr5, R-spondin, was secreted in a paracrine fashion that evolved during the maturation of CGNs, which coincided with the Lgr5 expression pattern. Our findings provide potential new insight into the critical regulation of Wnt signaling in the developing cerebellum and support a novel role for Lgr5 in the regulation of post-mitotic cells.


Introduction
The discovery of Lgr5 (leucine-rich-repeat-containing G-protein-coupled receptor 5) as an adult epithelial stem cell marker led to interest in this previously orphan receptor. Since its discovery as a marker of intestinal crypt stem cells [1], Lgr5 has been shown to mark stem or progenitor cell populations across diverse epithelial tissues, including skin, stomach, intestine, mammary gland, and cochlear hair follicle, and acts as a Wnt co-receptor (reviewed in [2]). To date, characterized epithelial Lgr5-positive cell populations all demonstrate self-renewal and proliferative capacity.
Wnt signaling plays an important role in many processes during development and homeostasis in the brain [3,4]. However, Lgr5 expression and function as an important Wnt co-receptor in Wnt-dependent cell types in the brain has not been described. Cerebellar granule neurons (CGNs) make up the largest neuronal population in the mammalian brain, outnumbering all other neurons combined [5], and their development is dependent on Wnt signaling, suggesting that Lgr5 may contribute to CGN biology.
CGNs proceed through well-organized, sequential differentiation events during development [6]. During murine embryonic development, CGN precursors (CGNPs) from the rhombic lip migrate to form the external germinal layer (EGL), where they undergo extensive proliferation in response to Sonic hedgehog (Shh) secreted from Purkinje neurons. CGNP proliferation continues for the first two postnatal weeks, but within a few days of birth cells begin to exit the cell cycle and differentiate. CGNPs stop dividing to differentiate and migrate through the molecular layer into the internal granule layer (IGL) [7]. The final maturation phase occurs in the IGL when CGNs form branched dendrites and long axons -a Wnt signaling dependent process [8,9]. CGNs secrete Wnt-7a, which acts in an autocrine fashion through the Frizzled-5 receptor to mediate synapse formation with excitatory mossy fibers [9,10]. Proper development of CGNs is critically important to the overall development and architecture of the cerebellum. Abnormal development or loss of CGNs leads to severe cerebellar abnormalities in mice and several disease states in humans [11,12]. Aberrant Wnt signaling in CGN precursors leads to severe cerebellar alterations [13], while interruption of Wnt signaling leads to improper synapse formation [9].
In other cell types, the Wnt receptor complex, consisting of LRP and Frizzled proteins, is recruited by and bound to R-spondin-activated Lgr5. Once bound to Lgr5, the LRP-Frizzled complex binds Wnt ligands to increase signaling through the Wnt/b-catenin pathway [14,15]. However, the role of Lgr5 in CGN development is unknown.
Here we report that Lgr5 is expressed in CGNs exclusively during their Wntdependent maturation phase, and that the Lgr5 ligand, R-spondin1 (Rspo1) displays a spatio-temporal concomitant pattern of expression. These data indicate Lgr5 is involved in the orchestrated development of these non-stem neuronal cell populations, demonstrating a potential role for Lgr5 outside of epithelial stem cells.

Animals
This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. All animal experiments were approved by the Cleveland Clinic Institutional Animal Care and Use Committee or the Sanford-Burnham Medical Research Institute Institutional Animal Care and Use Committee. Lgr5-eGFP-IRES-CreER2 (Lgr5 EGFP-CreERT2 )(RRID:IMSR_JAX:008875) [1] and Rosa26-lox-STOP-lox-tdTomato (R26R tdTomato )(RRID:IMSR_JAX:007914) [16] mice were acquired from Jackson Labs. Standard PCR genotyping was performed according to the vendor's website. At least 3 Lgr5 EGFP-CreERT2 mice were examined for each time point and marker. Catnb flox(exon3) mice (RRID:MGI_MGI:2673882) [17] were previously described [18].

Tissue processing, histology and imaging
Tissues were fixed in 4% PBS-buffered paraformaldehyde overnight and then transferred to 30% sucrose (w/v) for another 24 hours. Tissue was then embedded in Optimal Cutting Temperature (OCT) compound and sectioned at 8 microns on a Leica cryostat. Sections were blocked (PBS, 5% goat serum, 0.1% TritonX-100), incubated with primary antibodies for 2 hrs at room temperature in humidity chamber, washed with PBS and incubated with secondary antibodies for 45 min room temperature. Primary

Image Analysis
Image quantification was done in conjunction with ImageIQ (http://www.imageiq.com/) using image analysis software. Briefly, for Lgr5-GFP/tdTomato overlap experiments, Lgr5-GFP and tdTomato thresholds were set as standard across images and individual cells were marked separately as either negative or positive for each marker and then overlaid to quantify single-and double-positive cells.

Lgr5 is expressed in the postnatal cerebellum and its expression is temporally restricted
We analyzed brain expression of Lgr5, a known marker of epithelial stem cells [2], to understand if Lgr5 was playing a role in neural stem and progenitor cells. We utilized the Lgr5 EGFP-CreERT2 mouse, in which EGFP expression is driven by the Lgr5 locus [1]. While we did not find significant expression of Lgr5 in the neural stem cell compartments (S1 Figure), we surprisingly found robust expression in the cerebellum and olfactory bulb in the developing postnatal brain (Fig. 1a). Cerebellar Lgr5 expression was temporally restricted to the period of postnatal development between postnatal day (P) 4 and P21, with maximum expression between P10 -P14 (Fig. 1b). This expression pattern was confirmed using RNA expression data from the Cerebellar Development Transcriptome Database (RRID:nif-0000-00008) [20], suggesting this phenomenon is not an artifact of the transgene (Fig. 1c). In contrast, Lgr5 is expressed in the olfactory bulb by postnatal day 4 and persists into adulthood (Fig. 1d). Overall, these results suggest specific and tight regulatory control of Lgr5 during development of the cerebellum.

Lgr5-positive cells in the postnatal cerebellum are exclusively cerebellar granule neurons
To identify the specific cell population expressing Lgr5, we probed cerebellar sections from the Lgr5 EGFP-CreERT2 mouse with a panel of cerebellar cell lineage markers (Fig. 2). Lgr5 expressing cells were morphologically and geographically consistent with CGNs, nearly exclusively located in the internal granule layer in the cerebellum. Lgr5-positive cells were also positive for the neuronal marker, NeuN, and the mature granule neuron marker, gamma-aminobutyric acid type A (GABA(A)) receptor, alpha6 subunit (Gabra6) (Fig. 2a). Lgr5-positive cells were universally negative for the Bergmann glial marker, BLBP (Fig. 2b), the Purkinje neuron marker, Calbindin (Calb1) (Fig. 2c), and the oligodendrocyte lineage marker Olig2 (Fig. 2d) in all sections analyzed (Fig. 2e). Together, these findings indicate that Lgr5 is expressed in CGNs during postnatal development and that Lgr5 expression is exclusive to CGNs in the cerebellum during this period.

Lgr5-positive cells in the postnatal cerebellum are lineage restricted
In many tissues, Lgr5 marks multipotent stem cells; in contrast, we observed labeling of CGNs, which are terminally differentiated and lineage-restricted. To determine whether Lgr5-positive cells in the postnatal cerebellum included multipotent cells, we conducted lineage-tracing experiments by crossing Lgr5 EGFP-CreERT2 mice with Rosa26-loxP-STOP-loxP-tdTomato reporter mice (Lgr5 EGFP-CreERT2 ; R26R tdTomato ) [16]. The resulting animals express tamoxifeninducible Cre recombinase (CreER T2 ) in Lgr5-positive cells, allowing permanent activation of a Cre-inducible tdTomato fluorescent protein in these cells and their progeny (Fig. 3a). The ability of Lgr5-positive cells to differentiate into other cells types can be determined by subsequent histological examination of the tdTomatopositive (lineage traced) cells. We initiated lineage tracing at P4, the first stage at which Lgr5 is detected in the internal granule layer, as well as at later time points (P7, P10, P14, P21 and P28). We analyzed cerebella from these mice 3, 7 and 14 days after lineage tracing initiation. Cells expressing Lgr5 at the time of tamoxifen administration generated only mature granule neurons at all time points (Figs. 3b and 4, and data not shown). Lineage tracing started at P21 or P28 resulted in no detectable tdTomato-positive cells at P28 or P35, respectively, indicating that the Lgr5 locus is no longer active in CGNs at these time points. Long-term lineage trace experiments revealed Lgr5-positive postnatal CGNs contributed to the adult cerebellar architecture, as expected, and continued to be lineage-restricted (  Fig. 3c). Taken together, these results provide further evidence that Lgr5 specifically marks CGNs during a temporally restricted window and that these cells are long-lived and lineage-restricted.

Lgr5-expressing CGNs are non-proliferative and post-mitotic
To date, Lgr5 has been shown to mark cells that have the capacity for self-renewal and proliferation. However, mature CGNs are known to be post-mitotic. Therefore, we interrogated the proliferative potential of Lgr5-positive CGNs to understand if Lgr5 was expressed in a terminally differentiated and post-mitotic cell. Lgr5-positive cerebellar granule neurons were uniformly negative for Ki67, a marker of proliferation, at both early (Fig. 5a) and late stages of Lgr5 expression (  Fig. 5b). Cells that were previously Lgr5-positive CGNs, as identified by lineage tracing, were also negative for Ki67 (Fig. 5c). However, immature granule neurons in the EGL, which are not Lgr5 positive, are highly proliferative and accordingly stained intensely for Ki67 (Fig. 5a-c).
To further investigate if Lgr5-postive CGNs had the capacity for self-renewal or proliferation, we interrogated if over-activation of Wnt signaling could stimulate proliferation of Lgr5-positive CGNs. In other tissue types, constitutively active Wnt signaling increases proliferation of Lgr5-positive populations [21,22] and can cause tumor formation [23]. We crossed Lgr5 EGFP-CreERT2 mice with Catnb flox(exon3) mice [17] to yield Lgr5 EGFP-CreERT2 ; Catnb flox(exon3) mice, in which Cre recombinase initiates expression of a constitutively active form of b-catenin in Lgr5-positive cells following treatment with tamoxifen (Fig. 5d). b-catenin is the key downstream node for canonical Wnt signaling and increasing its activity increases Wnt signaling-dependent cellular responses through transcriptional regulation. We found that tamoxifen administration at P4 led to a robust and prolonged expression of Lgr5 in CGNs (Fig. 5e, f). However, this increase in Lgr5 and Wnt signaling did not disrupt normal architecture of the cerebellum (Fig. 5e), did not disrupt Lgr5-positive CGN lineage-restriction (Fig. 5f) or initiate proliferation in these cells as has been shown in other systems [22] (Fig. 5g). These results further support the notion that Lgr5 does not regulate proliferation or selfrenewal in CGNs.

Maturing CGNs progress through a transient Lgr5-positive phase
CGNs are a relatively homogenous population of neurons that progress through defined stages of development regulated by distinct phases of temporally regulated gene expression [6,24]. Lgr5 is expressed in CGNs once they enter the IGL, and at a population level Lgr5 is expressed in CGNs in the IGL from P4 to P21. However, as individual CGNs progress through maturation at different times during the postnatal time period based on when they exit the proliferative stage in the EGL, we sought to determine the duration of Lgr5 expression in individual CGNs during maturation in vivo. We again utilized Lgr5 EGFP-CreERT2 ; R26R tdTomato mice to conduct a lineage tracing time course initiated at the first signs of Lgr5 expression in CGNs at P4 and analyzed cerebella for tdTomato-positive cells at multiple time points until P21, when the Lgr5 locus is no longer active (Fig. 6a). We calculated the ratio of tdTomato-positive cells that were also positive for Lgr5 to those that were solely tdTomato-positive in order to understand how quickly Lgr5-expression was turned off in cells that were Lgr5-positive at the time of lineage trace initiation. When initiating lineage tracing at P4, we found that few cells had gone through both recombination and production of the tdTomato protein by P7, as anticipated. Once Lgr5-positive cells are exposed to tamoxifen, it can take up to 24 hours before recombination and transcription occurs, and tamoxifen can remain in the system, maintaining Cre recombinase activity for a further 48 hours [25]. Therefore, we expected Lgr5-positive cells to be labeled from P4-P7 (Fig 6a), and, with considerably more cells expressing Lgr5 at P7, we  anticipated most lineage-traced cells would be derived from the later days of this time period. Of the few cells that were tdTomato-positive at P7, nearly 20 percent were already Lgr5-negative. By P10, 50 percent of tdTomato-positive cells were Lgr5-negative (Fig. 6b,c) and this increased to nearly 90 percent negative at P14. These results suggest that the phase of Lgr5 expression in individual CGNs during maturation is transient (Fig 6c). Based on the length of Cre activity and the sharp drop in tdTomato-positive/Lgr5-positive cells, we predict that Lgr5 is expressed these mice die of other complications related to transgene expression. b-catenin over-activity led to prolonged Lgr5 expression in the IGL in Lgr5 EGFP-
doi:10.1371/journal.pone.0114433.g005 Lgr5 Temporally Expressed in Developing Cerebellar Granule Neurons during a period of 3-5 days in an individual CGN. Given that the half-life of the EGFP reporter is greater than 24 hrs [26], the length of time that Lgr5 is active in a given cell is likely even shorter than 3-5 days predicted by the EGFP reporter. This is consistent with the observation that Lgr5 acts in a short period to provide a temporary elevation in Wnt signaling activity [15]. The short period of Lgr5 expression in individual cells and the high percentage of CGNs that are Lgr5positive at any given point during the postnatal period supports the idea that Lgr5 expression occurs in most, if not all, CGNs during their maturation.

Expression of the Lgr5 ligand, R-spondin 1, coincides with Lgr5 expression in CGN maturation
In epithelial tissues, activation of the Wnt pathway through the Lgr5 receptor is initiated by the R-Spondin family of ligands [14]. Therefore, we stained for Rspo1 at various stages of CGN development to understand if this ligand was potentially regulating Lgr5 activation in CGNs. We found Rspo1 expression mirrored the onset and loss of Lgr5 expression (Figs. 7 and 8). The initial cellular source of Rspo1 appeared to be CGNPs in the EGL, with expression present by P4 (Fig. 7a). As the EGL shrunk due to CGNP differentiation and migration into the IGL, Purkinje neurons, which are known to secrete Shh during CGN development [5], began to express Rspo1, exhibiting maximal staining between P8 and P14 (  Fig. 7b-c). Rspo1 expression was barely detectable at P28, its expression loss coinciding with the loss of Lgr5 expression (Fig. 7d).
Purkinje neuron secretion of the Lgr5 ligand, Rspo1, and potentially a transition of Shh to Rspo1 secretion at these time points was supported by the Allen Developing Mouse Brain Atlas (RRID:nif-0000-00509) [27]. The Allen Developing Mouse Brain Atlas contains a database of RNA in situ hybridization (ISH) for nearly 2000 genes across seven stages of brain development. Using this database, we found that Rspo1 expression was low at P4 and much higher at P14, supporting our immunofluorescence data (Fig. 8a, d). In contrast, Shh was expressed in Purkinje neurons at high levels at P4, but this expression was lower at P14 (Fig. 8b, e). Purkinje neurons regained expression of Shh at P28 and continued expressing Shh through adulthood (P56) (Fig. 8h, k), while Rspo1 expression was completely abolished by P56 (Fig. 8j). Purkinje neurons secrete Shh, which acts to stimulate proliferation of CGN precursors in the EGL [5], but our results demonstrate that Purkinje neurons may also have a role in the next phase of CGN differentiation by activating the Wnt pathway through Rspo1 expression. Indeed, migrating CGNs must past directly through the layer of Purkinje neurons on their journey to the IGL, exposing them directly to this secreted ligand.

Discussion
Nearly 15 years ago, two reports using a panel of homogenized tissue RNA showed the then-orphan receptor Lgr5 (also termed FEX, HG38, GPR49, and GPR67) had detectable expression in the brain [28,29]. However no detailed characterization of Lgr5 expression in the developing or adult brain has since been published. What initially started as a search for Lgr5 expression in neural stem and progenitor cells led us to the surprising discovery that Lgr5 is expressed in maturing CGNs, the most prevalent neuronal population in the mammalian brain. This is surprising because Lgr5 is a well-characterized marker of epithelial stem and progenitor cells in somatic tissues, but mature CGNs are both nonepithelial and post-mitotic. However, CGNs progress through a well-organized differentiation process during postnatal development and are dependent on Wnt signaling for full maturation, indicating a potential purpose for this Lgr5 expression [3,6].
In this report, we demonstrate that Lgr5 is expressed in CGNs in the IGL during postnatal cerebellar development, and that Lgr5-positive CGNs are lineagerestricted and post-mitotic, even in the presence of forced activation of the Wnt/ b-catenin pathway. We find that Lgr5 is expressed from P4-P21 with no expression in the adult cerebellum. CGNs migrate into the IGL and mature continuously during the same period of Lgr5 expression, suggesting a role for Lgr5 in the maturation of CGNs. Based on the very high percentage of CGNs that are Lgr5-positive from P7 to P14 and the phasic nature of Lgr5 expression indicated by our results, we conclude it is likely that nearly all CGNs go through a short period of Lgr5 expression, which we speculate may initiate or enhance Wnt signaling during the Wnt dependent phase of synaptogenesis and maturation.
In further support of this possibility, Rspo1, the Lgr5 ligand that triggers downstream Wnt signaling is also expressed in the postnatal cerebellum transiently during CGN maturation. Interestingly, the cell types that express Rspo1 during this window vary with the changing architecture of the cerebellum during postnatal development. Early in development, when most granule neurons are immature precursors and the EGL and IGL are in close proximity, CGNPs express Rspo1. However, as CGNPs mature, the EGL diminishes and is separated from the IGL and the maturing CGNs by an expanding molecular layer. At this point, Purkinje neurons that are directly adjacent to the IGL begin to express Rspo1, suggesting that the Rspo1/Lgr5 program that controls Wnt signaling in other tissue types is also tightly regulated in the developing cerebellum.
During maturation, CGNs respond to Wnt signaling to form synapses with mossy fiber axons and Purkinje neurons [8,9]. This response is mediated through the binding of Wnt-7a to the Frizzled-5 (Fz5) receptor [10]. Lgr5 has been shown in other tissues to be a Wnt pathway co-receptor that recruits and binds the LRP-Frizzled receptor complex, and subsequently responds to Wnt ligands to increase signaling in the Wnt/b-catenin pathway [14]. Therefore, Lgr5 may be acting as a co-receptor within this Wnt-7a responsive complex to facilitate CGN maturation.
It is possible, however that Lgr5 may have a completely novel role in CGNs outside of its known roles in epithelial stem and progenitor cells.
These results raise the possibility that Lgr5 may not only function in stem and progenitor cells, but may have a more general role as a regulator of cell fate. In fact, a recent report from our group demonstrates that Lgr5 is also constitutively expressed in a subpopulation of mature retinal amacrine cells, interneurons in the mammalian eye, providing additional evidence of a role outside of epithelial stem and progenitor cells [30]. Lgr5 EGFP-CreERT2 homozygous mice die within 24 hours of birth, which currently limits our ability to study the effect of Lgr5 knockout on CGNs in vivo. Future studies using conditional Lgr5 knockout mice could be conducted to further study the role of Lgr5 in nervous system tissue.
Wnt signaling is critical in many developing and adult tissues, and disruption or over-activation of the Wnt pathway can lead to severe abnormalities in many tissues, including the cerebellum [11,13]. Disruption of Lgr5 in the intestine led to rapid and complete loss of intestinal crypts comparable to loss of Wnt signaling [14], while overexpression of Rspo1 led to rapid expansion of intestinal crypts, demonstrating how important these Wnt signaling regulators are to the tightly regulated Wnt pathway. Our study provides evidence that Lgr5 and Rspo1 have roles outside of epithelial stem and progenitor cells, and provides novel insight into the function and regulation of Wnt signaling in cerebellar development.