Transcriptome Analysis Did Not Show Endogenous Stem Cell Characteristics in Murine Lgr5+ Retinal Cells

Lgr5, an intestinal adult stem cell marker, was recently also found in neuronal tissues. We investigated whether retinal Lgr5+ cells express properties of neural stem cells (NSC) and/or of differentiated interneurons during retinal development. RNA was isolated from Lgr5+ and Lgr5− populations from postnatal day 5 (PN5) and adult retinas of Lgr5EGFP-Ires-CreERT2 knock-in mice sorted by fluorescence-activated cell sorting (FACS). Transcriptome analyses were performed on two RNA samples of each developmental stage (PN5 and adult). The online platform PANTHER (Protein ANalysis THrough Evolutionary Relationships) was used to determine overrepresented gene ontology (GO) terms of biological processes within the set of differentially expressed genes. The detailed evaluation included gene expression in regard to stem cell maintenance/proliferation, cell cycle, and Wnt signaling but also markers of differentiated retinal neurons. None of the enriched GO terms of upregulated genes of Lgr5+ cells showed a positive association to NSC. On the contrary, NSC maintenance and proliferation rather prevail in the Lgr5− cell population. Furthermore, results suggesting that Wnt signaling is not active in the Lgr5+ population. Therefore, our transcriptome analysis of Lgr5+ retinal cells suggest that these cells are differentiated neurons, specifically glycinergic amacrine cells.


Introduction
Lgr5 (leucine-rich repeat containing G-protein receptor 5) belongs to a family of G-protein-coupled receptors [1] and is a Wnt target gene, which specifically acts as a receptor for Wnt agonists called R-spondins. Wnt signaling is critical during development, controlling cell fate decisions and tissue patterning [2]. In the adult organism, Wnt signals regulate homeostatic regeneration in numerous tissues, including the retina [3]. Furthermore, aberrant Wnt signaling is implicated in several diseases, especially cancer. It is classified into two pathways, GSK-3β dependent (canonical) and GSK-3β independent (non-canonical) [4]. In both cases, Wnt binds to the membrane receptor Frizzled (Fz). In the canonical pathway, Wnt binds to Fz and its co-receptor low-density lipoprotein receptor-related protein 5/6 (LRP5/6). Disheveled (Dvl) is recruited to the membrane resulting in the endocytosis of the complex to form a signalosome. The β-catenin destruction complex consisting of Axin2, adenomatous polyposis coli (APC), casein kinase-1α (CK-1α), and glycogen synthase kinase-3 (GSK-3β) is sequestered and disassembled. The inactivation of the destruction complex hinders GSK-3β in phosphorylating β-catenin. Thus, β-catenin evades proteasomal degradation and accumulates in the cytosol. Subsequently, β-catenin translocates to the nucleus, displaces the transcriptional repressor Groucho and bind to the T-cell factor/lymphoid-enhancing factor (TCF/LEF) promoter/enhancer compared to adult Lgr5 + cells were analyzed. Differentially expressed genes were defined by a fold change ≥ 2 or ≤ −2, p-value < 0.05, and false discovery rate (FDR) < 0.05. The online tool PANTHER was used to determine overrepresented gene ontology (GO) terms of biological processes within the set of differentially expressed genes. Thereby, the input list was compared to the list of all genes analyzed by the microarray and the fold enrichment was calculated.
The Clariom S Mouse array analyzes a total of 22,206 genes. Of these, 2242 genes were differentially expressed in adult Lgr5 + compared to adult Lgr5 − , 1335 being upregulated and 907 being downregulated ( Figure 1A). At PN5, Lgr5 + presented 1821 differentially-expressed genes compared to Lgr5 − . Thereof, 972 genes showed significantly higher whereas 849 showed significantly lower expression compared to the negative populations ( Figure 1B). When comparing the positive populations of both developmental stages, significant differences in gene expression was found in 1884 genes. Thereof, 814 genes were upregulated and 1070 downregulated ( Figure 1C). change ≥ 2 or ≤ −2, p-value < 0.05, and false discovery rate (FDR) < 0.05. The online tool PANTHER was used to determine overrepresented gene ontology (GO) terms of biological processes within the set of differentially expressed genes. Thereby, the input list was compared to the list of all genes analyzed by the microarray and the fold enrichment was calculated. The Clariom S Mouse array analyzes a total of 22,206 genes. Of these, 2242 genes were differentially expressed in adult Lgr5 + compared to adult Lgr5 − , 1335 being upregulated and 907 being downregulated ( Figure 1A). At PN5, Lgr5 + presented 1821 differentially-expressed genes compared to Lgr5 − . Thereof, 972 genes showed significantly higher whereas 849 showed significantly lower expression compared to the negative populations ( Figure 1B). When comparing the positive populations of both developmental stages, significant differences in gene expression was found in 1884 genes. Thereof, 814 genes were upregulated and 1070 downregulated ( Figure 1C). Lgr5 + compared to PN5 Lgr5 − and (C) PN5 Lgr5 + compared to adult Lgr5 + . The horizontal axis is fold change ≥ 2 or ≤ −2, and the vertical axis represents the p-value after the negative logarithm conversion. The red dots indicate the upregulated genes, whereas the green dots represent the downregulated genes. The corrected p < 0.05 was an absolute threshold used to select DEGs. DEGs, differentially expressed genes.
The lists of up-or downregulated genes from DGE were used as input lists for the statistical enrichment analysis by PANTHER. Due to the high number of overrepresented GO terms, only the top ten overrepresented terms with the highest fold enrichments (FE) of upregulated (+) and downregulated (−) genes are shown (Tables 1-3). Complete lists of all analyses can be found in the supplementary materials section (Supplementary Tables S2-S7). Lgr5 + compared to PN5 Lgr5 − and (C) PN5 Lgr5 + compared to adult Lgr5 + . The horizontal axis is fold change ≥ 2 or ≤ −2, and the vertical axis represents the p-value after the negative logarithm conversion. The red dots indicate the upregulated genes, whereas the green dots represent the downregulated genes. The corrected p < 0.05 was an absolute threshold used to select DEGs. DEGs, differentially expressed genes.
The lists of up-or downregulated genes from DGE were used as input lists for the statistical enrichment analysis by PANTHER. Due to the high number of overrepresented GO terms, only the top ten overrepresented terms with the highest fold enrichments (FE) of upregulated (+) and downregulated (−) genes are shown (Tables 1-3). Complete lists of all analyses can be found in the supplementary materials section (Supplementary Tables S2-S7).   From upregulated genes in adult Lgr5 + , GO terms related to synaptic function were the most prevalent. Around a third of all enriched GO terms including seven of the top ten were associated to either pre-or postsynaptic organization, neurotransmitter signaling, action potential or synaptic plasticity. The GO term "Synaptic vesicle clustering" showed the highest FE (16.00; p = 6.89 × 10 −4 ; Table 1). Glycinergic, GABAergic, glutamatergic, as well as dopaminergic signaling showed significant overrepresentation. Yet, glycinergic signaling showed the highest FE of all neurotransmitters ("Synaptic transmission, glycinergic"; FE = 11.43; p = 3.91 × 10 −4 ). Furthermore, genes related to neuronal morphology, like dendrite and axon morphogenesis, were enriched. Downregulated genes in adult Lgr5 + showed an enrichment for GO terms associated to phototransduction, photoreceptor cell differentiation, and maintenance.
Approximately half of the overrepresented GO terms of upregulated genes in PN5 Lgr5 + cells were associated to synaptic signaling. Negative regulation of dopamine secretion showed the highest FE (Table 2; FE = 23.51, p = 2.37 × 10 −5 ). The highest FE of a positive reference to neurotransmitter signaling was the GO term "Positive regulation of AMPA receptor activity" (FE = 13.06; p = 1.64 × 10 −4 ) indicating glutamatergic inputs to Lgr5 + cells. Yet, the GO term "Glycine transport" (FE = 11.76, p = 1.2 × 10 −2 ) showed the highest FE for a presynaptic neurotransmitter. In addition, genes for GABAergic and cholinergic synaptic transmission showed significant enrichment. Like in adult Lgr5 + , genes related to neuronal morphology were also enriched. Of the 158 overrepresented GO terms of downregulated genes, "Retinal rod cell differentiation" showed the highest FE (Supplementary Table S5; FE = 24.18; p = 1.52 × 10 −4 ). Several terms associated to cell proliferation, such as cell cycle progression, DNA replication, chromatin segregation and cytokinesis, were found (Supplementary Table S5). Additionally, genes associated to the terms "neuronal stem cell maintenance" and "positive regulation of neural precursor cell proliferation" were significantly enriched with FE of 6.31 (p = 8.35 × 10 −4 ) and 3.45 (p = 4.2 × 10 −4 ), respectively (Supplementary Table S5). Furthermore, genes related to angiogenesis and the regulation of apoptosis were enriched.
Comparing Lgr5 + cells from PN5 to adult animals, three major groups of overrepresented terms of upregulated genes were found. One consisting of terms related to cell cycle processes, such as the term with the second highest FE "DNA damage response, signal transduction by p53 class mediator resulting in transcription of p21 class mediator" (Table 3; FE = 12.11; p = 1.92 × 10 −4 ) and different cell cycle phase transitions. The second group encompasses terms associated to development and neuronal differentiation. For example, "Regulation of development, heterochronic" (FE = 8.88; p = 6.04 × 10 −4 ). The last group includes terms connected to transcriptional and translational processes, with "ribosomal small subunit assembly", "cytoplasmic translation", and "ribosomal large subunit assembly" all in the top ten of overrepresented terms. Furthermore, genes associated to the term "positive regulation of Notch signaling pathway" showed significant enrichment (FE = 5.33; p = 2.97 × 10 −4 ). The 77 overrepresented GO terms of downregulated genes in Lgr5 + cells at PN5 were mostly related to synaptic signaling and transmission of nerve impulse. Synaptic signaling includes GO terms of synapse assembly, plasticity, as well as synaptic transmission. Whereas transmission of nerve impulse encompasses processes of action potential generation, hence the transport of different ions across the plasma membrane, regulation of intracellular ion concentrations and membrane repolarization.

Gene List Analysis
For better understanding of the PANTHER overrepresentation analysis, specific gene lists were analyzed for expression values and fold changes of Lgr5 + cells. The list of up-or downregulated genes of DGE is the input for the PANTHER tool. It does not take signal values into account. Therefore, genes that display low signal values of below 100 but show significant differential gene expression compared to the control population will still be analyzed by PANTHER.
Lgr5 marks stem cell populations in several tissues. Furthermore, it has been proposed that retinal Lgr5 + cells harbor regenerative potential. Regarding genes involved in NSC maintenance, in the PANTHER analysis only Prox1 was highly expressed in both PN5 and adult, showing mean values of 1138 and 889, respectively (Supplementary Figure S4).
On the other hand, our analysis revealed that most of the analyzed cell cycle genes were downregulated in PN5 Lgr5 + cells ( Figure 2). However, signal values of these genes were relatively low, mostly < 100. Still, confirming the PANTHER analysis, some of these genes showed significantly higher values than expression in adult Lgr5 + cells. For example, Cdk1 showed the highest fold change in signal values between PN5 Lgr5 + and PN5 Lgr5 − (FC: −16.09; p = 4.3 × 10 −6 ) with a mean signal intensity of 14 in positive cells. Furthermore, this was detected to be significantly higher than the signal value of Cdk1 in adult Lgr5 + cells (FC: 3.59; p = 4.7 × 10 −5 ) with mean signal of 4. Of the analyzed cell cycle genes, the highest expressed gene was Ccnd1 in PN5 Lgr5 − cells.
Lgr5 is part of the Wnt signaling cascade. Yet, the only overrepresented GO term related to Wnt signaling in the PANTHER analysis was "negative regulation of canonical Wnt signaling pathway" (FE: 3.51, p = 2.3 × 10 −5 ; Supplementary Table S5). The gene expression levels of members of the signaling cascade, target genes and Wnt antagonists were also investigated. At PN5 differentially expressed genes of the signaling cascade showed low signal intensities with mean values of below 40 ( Figure 3). The Wnt target gene Ccnd1 is a marker of cell cycle progression. As mentioned above Ccnd1 showed significantly lower expression in Lgr5 + compared to Lgr5 − cells at PN5 (FC: −6.16, p = 1.6 × 10 −6 ). Yet, Lgr5 + cells showed lower expression of Wnt signaling antagonists, for example Dkk3 (FC: −3.79, p = 1.9 × 10 −5 ). In contrast, these genes were significantly higher expressed in adult Lgr5 + compared to Lgr5 − cells (Dkk1; FC: 3.42; p = 1.4 × 10 −5 ).
Adult Lgr5 + cells have been characterized as glycinergic amacrine cells. By morphological analysis they have been categorized as narrow-field amacrine cells, in particular as AII amacrine cells [10]. Yet, molecular profiling of Lgr5 + is lacking. Therefore, gene expression levels of different known amacrine markers were analyzed. With more than 40 different subtypes, amacrine cells represent one of the most diverse neuronal classes. Still several markers have been identified to be expressed across amacrine subtypes. Lgr5 + cells showed significantly higher levels of all analyzed pan-amacrine markers compared to the negative population ( Figure 4A). The difference in expression levels varied greatly, the least expressed being Nrxn2 (mean: 88.3) and the highest Pax6 (mean: 1909). In addition, Slc32a1 showed low signal values (mean: 117).
On the other hand, our analysis revealed that most of the analyzed cell cycle genes were downregulated in PN5 Lgr5 + cells ( Figure 2). However, signal values of these genes were relatively low, mostly < 100. Still, confirming the PANTHER analysis, some of these genes showed significantly higher values than expression in adult Lgr5 + cells. For example, Cdk1 showed the highest fold change in signal values between PN5 Lgr5 + and PN5 Lgr5 − (FC: −16.09; p = 4.3 × 10 −6 ) with a mean signal intensity of 14 in positive cells. Furthermore, this was detected to be significantly higher than the signal value of Cdk1 in adult Lgr5 + cells (FC: 3.59; p = 4.7 × 10 −5 ) with mean signal of 4. Of the analyzed cell cycle genes, the highest expressed gene was Ccnd1 in PN5 Lgr5 − cells. Lgr5 is part of the Wnt signaling cascade. Yet, the only overrepresented GO term related to Wnt signaling in the PANTHER analysis was "negative regulation of canonical Wnt signaling pathway" (FE: 3.51, p = 2.3 × 10 −5 ; Supplementary Table S5). The gene expression levels of members of the signaling cascade, target genes and Wnt antagonists were also investigated. At PN5 differentially expressed genes of the signaling cascade showed low signal intensities with mean values of below 40 ( Figure 3). The Wnt target gene Ccnd1 is a marker of cell cycle progression. As mentioned above Ccnd1 showed significantly lower expression in Lgr5 + compared to Lgr5 − cells at PN5 (FC: −6.16, p = 1.6 × 10 −6 ). Yet, Lgr5 + cells showed lower expression of Wnt signaling antagonists, for example Dkk3 (FC: −3.79, p = 1.9 × 10 −5 ). In contrast, these genes were significantly higher expressed in adult Lgr5 + compared to Lgr5 − cells (Dkk1; FC: 3.42; p = 1.4 × 10 −5 ). Adult Lgr5 + cells have been characterized as glycinergic amacrine cells. By morphological analysis they have been categorized as narrow-field amacrine cells, in particular as AII amacrine cells [10]. Yet, molecular profiling of Lgr5 + is lacking. Therefore, gene expression levels of different known amacrine markers were analyzed. With more than 40 different subtypes, amacrine cells represent one Furthermore, molecular markers of the more widely studied amacrine cells have emerged. Lgr5 + cells showed signal values of <100 for all markers of GABAergic amacrine cells ( Figure 4B; Chat, Gad1, and Gad2). Likewise, Lgr5 + did not express the marker of excitatory amacrine cells Slc17a (mean: 4). Except for Gjd2, genes of glycinergic subtypes (Slc6a9, Dab1, Prox1) were highly expressed. Moreover, Ebf3 showed high expression values in Lgr5 + cells. Ebf3 marks non-GABAergic non-glycinergic (nGnG) amacrine cells as well as Satb2-Ebf3-Glyt1-postive (SEG) amacrine cells. All markers of glycinergic amacrine cells were significantly higher expressed in Lgr5 + compared to Lgr5 − ( Figure 4C) Gad1, and Gad2). Likewise, Lgr5 + did not express the marker of excitatory amacrine cells Slc17a (mean: 4). Except for Gjd2, genes of glycinergic subtypes (Slc6a9, Dab1, Prox1) were highly expressed. Moreover, Ebf3 showed high expression values in Lgr5 + cells. Ebf3 marks non-GABAergic nonglycinergic (nGnG) amacrine cells as well as Satb2-Ebf3-Glyt1-postive (SEG) amacrine cells. All markers of glycinergic amacrine cells were significantly higher expressed in Lgr5 + compared to Lgr5 − ( Figure 4C). However, Myf6, Plekhf1 and 4930444P10Rik showed signal values < 100. Since adult Lgr5 + cells highly expressed the nGnG and SEG amacrine cell marker Ebf3, other markers of nGnG cells were analyzed. The expression values of these markers were investigated in PN5 Lgr5 + cells since the transcriptional profile of nGnG cells has so far only been studied at PN6 [11].
In addition, at PN5 Lgr5 + cells highly expressed Ebf3 ( Figure 5; mean: 2213). Neurod6 showed a significant fold change of 9.08 in Lgr5 + compared to Lgr5 − (p = 2.9 × 10 −6 ). Yet, the signal values are low compared to the expression of Ebf3 (mean: 118). Similar expression values were seen for Satb2 (mean: 138), however, these were not significantly upregulated compared to Lgr5 − . In addition to Ebf3, Frem1 showed the highest signal values (mean: 280) and a significant fold change of 6.73 (p = 5.3 × 10 −7 ). Since adult Lgr5 + cells highly expressed the nGnG and SEG amacrine cell marker Ebf3, other markers of nGnG cells were analyzed. The expression values of these markers were investigated in PN5 Lgr5 + cells since the transcriptional profile of nGnG cells has so far only been studied at PN6 [13].
In addition, at PN5 Lgr5 + cells highly expressed Ebf3 ( Figure 5; mean: 2213). Neurod6 showed a significant fold change of 9.08 in Lgr5 + compared to Lgr5 − (p = 2.9 × 10 −6 ). Yet, the signal values are low compared to the expression of Ebf3 (mean: 118). Similar expression values were seen for Satb2 (mean: 138), however, these were not significantly upregulated compared to Lgr5 − . In addition to Ebf3, Frem1 showed the highest signal values (mean: 280) and a significant fold change of 6.73 (p = 5.3 × 10 −7 ). By single cell RNA-sequencing Macosko et al. discovered candidate markers of amacrine subpopulations [12]. Therefore, expression of these markers in Lgr5 + cells was analyzed. The highest expression and significant fold change was seen for the marker Igf1, with a mean signal value of 701 and fold change of 7.34 ( Figure 6; p = 9.9 × 10 −7 ). Two other markers showed signal values of above 100 and significant upregulation, namely Ptgds (mean: 201, FC: 6.73, p = 5.3 × 10 −7 ) and Igfbp5 (mean: 106; FC: 6.01, p = 1.3 × 10 −5 ). Amacrine cells are classified by the neurotransmitter they secret. Yet to integrate and modulate the visual signal they receive signals from different bipolar cells and other amacrine cells. For instance, the AII amacrine cell has been shown to receive at least nine different inputs arising from various cells and using diverse neurotransmitters [13]. Therefore, Lgr5 + cells were further characterized by their expression pattern of different neurotransmitter receptors. By single cell RNA-sequencing Macosko et al. discovered candidate markers of amacrine subpopulations [14]. Therefore, expression of these markers in Lgr5 + cells was analyzed. The highest expression and significant fold change was seen for the marker Igf1, with a mean signal value of 701 and fold change of 7.34 ( Figure 6; p = 9.9 × 10 −7 ). Two other markers showed signal values of above 100 and significant upregulation, namely Ptgds (mean: 201, FC: 6.73, p = 5.3 × 10 −7 ) and Igfbp5 (mean: 106; FC: 6.01, p = 1.3 × 10 −5 ). By single cell RNA-sequencing Macosko et al. discovered candidate markers of amacrine subpopulations [12]. Therefore, expression of these markers in Lgr5 + cells was analyzed. The highest expression and significant fold change was seen for the marker Igf1, with a mean signal value of 701 and fold change of 7.34 ( Figure 6; p = 9.9 × 10 −7 ). Two other markers showed signal values of above 100 and significant upregulation, namely Ptgds (mean: 201, FC: 6.73, p = 5.3 × 10 −7 ) and Igfbp5 (mean: 106; FC: 6.01, p = 1.3 × 10 −5 ). Amacrine cells are classified by the neurotransmitter they secret. Yet to integrate and modulate the visual signal they receive signals from different bipolar cells and other amacrine cells. For instance, the AII amacrine cell has been shown to receive at least nine different inputs arising from various cells and using diverse neurotransmitters [13]. Therefore, Lgr5 + cells were further characterized by their expression pattern of different neurotransmitter receptors. Amacrine cells are classified by the neurotransmitter they secret. Yet to integrate and modulate the visual signal they receive signals from different bipolar cells and other amacrine cells. For instance, the AII amacrine cell has been shown to receive at least nine different inputs arising from various cells and using diverse neurotransmitters [15]. Therefore, Lgr5 + cells were further characterized by their expression pattern of different neurotransmitter receptors.
In the retina, GABA is widely expressed by subpopulations of amacrine cells, bipolar cells, horizontal cells as well as retinal ganglion cells. Lgr5 + cells express a variety of GABA-receptors. The highest expressed is the GABA type A receptor subunit β3 (Gabrb3; mean: 1009). Moreover, both type B receptors are expressed (Gabbr1 and Gabbr2), as well as one type C receptor (Gabrr1).
In the retina, GABA is widely expressed by subpopulations of amacrine cells, bipolar cells, horizontal cells as well as retinal ganglion cells. Lgr5 + cells express a variety of GABA-receptors. The highest expressed is the GABA type A receptor subunit β3 (Gabrb3; mean: 1009). Moreover, both type B receptors are expressed (Gabbr1 and Gabbr2), as well as one type C receptor (Gabrr1).

Discussion
The statistical overrepresentation tests of differentially expressed genes of both developmental stages indicate that neither adult nor PN5 Lgr5 + cells harbor stem cell potential. None of the enriched GO terms of upregulated genes shows a positive association to NSC. On the contrary, downregulated genes in neonatal Lgr5 + cells compared to the negative population are enriched for genes associated

Discussion
The statistical overrepresentation tests of differentially expressed genes of both developmental stages indicate that neither adult nor PN5 Lgr5 + cells harbor stem cell potential. None of the enriched GO terms of upregulated genes shows a positive association to NSC. On the contrary, downregulated genes in neonatal Lgr5 + cells compared to the negative population are enriched for genes associated to NSC maintenance and proliferation, indicating these processes prevail in the Lgr5 − cell population. Furthermore, several genes of GO terms related to the cell cycle are enriched in the neonatal Lgr5 − cell population. The heat maps depicting the expression values and fold changes of genes associated to NSC and cell cycle affirm these findings. Of the NSC markers analyzed only the NSC maintenance marker Prox1 is expressed. During embryonic retinogenesis Prox1-expression regulates cell cycle exit of progenitor cells [18]. However, it is also expressed in a subtype of horizontal cells and in AII amacrine cells. In both of these cell types Prox1-expression is upregulated during maturation indicating different functions of Prox1 during embryonic development and in adulthood. In addition, the expression values and fold changes of cell cycle genes clearly show that in neonates the expression of these genes is significantly upregulated in the Lgr5 − cell population compared to Lgr5 + cells. While Müller cells, bipolar cells, and rods are still being generated from proliferating retinal progenitors at PN5, glycinergic amacrine cells exit the cell cycle around postnatal day 4 [19,20]. This coincides with the appearance of Lgr5-expression suggesting that Lgr5 is expressed after glycinergic amacrine cell-fate specification.
Since Lgr5 is a Wnt target gene and acts as a stabilizer of canonical Wnt signaling different genes of this pathway were analyzed. The PANTHER analysis identified one GO term associated with Wnt signaling: "negative regulation of canonical Wnt signaling pathway". These genes are enriched in the Lgr5 − cell population at PN5. However, from this result one cannot conclude that there is positive regulation of Wnt signaling in the Lgr5 + cell population. Indeed, the gene list analysis shows that Lgr5 + cells at PN5 express lower levels of Wnt antagonists/inhibitors. However, the genes show, overall, a low expression profile.
Regarding other members of the Wnt signaling cascade, none is significantly higher expressed in Lgr5 + cells compared to Lgr5 − cells at both developmental stages. Furthermore, Axin2 expression does not meet the signal threshold of 100. Axin2 is a transcriptional target of Wnt signaling and is generally regarded as an indicator of Wnt pathway activity [21]. These results suggest that even though Lgr5 is expressed, Wnt signaling is not active.
In this study, Lgr5 + cells of undamaged retinae were analyzed. Potentially injury to the retina might be necessary to induce stem cell properties of retinal Lgr5 + cells. In the liver a population of progenitors reside which only show regenerative capacity after acute injury, but do not contribute to homeostatic liver regeneration [22]. However, these cells only express Lgr5 after injury induction. That Lgr5 is expressed during homeostasis in retinal amacrine cells may suggest an alternative function in neurons than in stem cell populations.
In the murine retina, Lgr5 is expressed in glycinergic amacrine cells. However, this is not a homogenous cell population. Amacrine cells are the most diverse cell type of the retina and to date more than 40 different types have been described [23]. They are classified by their dendritic field (stratification and pattern), appearance of dendrites as well as neurotransmitter used. GABA and glycine are the two major neurotransmitters produced by amacrine cells [24]. Yet a small subpopulation remains for which the neurotransmitter is still unknown [13]. In addition to neurotransmitters, only a few molecular markers are known for the characterization of amacrine subpopulations. Recent advances in genetic profiling are changing this. For example, the efforts of Siegert et al. who generated a genetic address book of different retinal cell types [25]. By demonstrating that Lgr5 + cells display morphologies of different amacrine subpopulations, such as Flag A/B and AII cells, Sukhdeo et al. already suggested that the Lgr5 + amacrine cell population is heterogeneous [12].
Several receptors of the Lgr-family are alternatively spliced resulting in diverse transcript variants [26,27]. So far, three different splice variants of Lgr5 have been identified [8]. PCR was performed on adult mouse tissue to investigate whether Lgr5 is alternatively spliced in the retina (see Supplementary Data S1). Thereby, full-length transcript variants of variable sizes were found. However, these did not correspond to known variants Lgr5∆5 or Lgr5∆8 [28]. Osawa et al. demonstrated that Lgr5∆5 expressing cells show increased proliferation [8]. As no proliferation is seen in adult retinal tissue, it is not surprising that this variant is not expressed in the retina. Additionally, heterogeneity in the expression of Lgr5 splice variants, e.g., fetal tissue-specific variants or injury/disease-related modifications in adult tissue could possibly diminish the ability to detect progenitor capability. Follow up studies would be necessary to investigate alternative variants in the retina. Since Lgr5 is composed of 18 exons, several different splice variants are possible.

Animals
Adult (
Tissue was washed by centrifugation with Hank's Balanced Salt Solution (HBSS; Thermo Fisher Scientific, Waltham, MA, USA) + 2% fetal bovine serum (FBS) and gently triturated using a fire-polished Pasteur pipette. The filtered single cell solution was sorted according to their GFP fluorescence on a FACS Aria (Becton Dickinson, Allschwil, Switzerland) at the cytometry core facility of the Department for BioMedical Research (DBMR) of the Faculty of Medicine of the University of Bern. Both Lgr5 + and Lgr5 − cell fractions were collected in HBSS + 2% FBS.
Total RNA of FACS sorted cells was extracted using the PicoPure RNA isolation kit (Thermo Fisher Scientific) according to the manufacturer's instructions. RNA concentration, as well as quality was assessed by an Experion RNA HighSens Chip (Bio-Rad, Cressier, Switzerland).
Microarray processing was performed by ATLAS Biolabs in Berlin, Germany. RNA samples were analyzed by Affymetrix Clariom S Mouse Array. Raw intensity values were normalized by Robust Multiarray Averaging (RMA) and analyzed with Transcriptome Analysis Console (TAC) software (Version 4; Thermo Fisher Scientific).

Analyses of DGE
The online platform PANTHER (http://www.pantherdb.org) was used to analyze the results of DGE [29]. Statistical overrepresentation test of GO biological processes was performed on differentially expressed genes. Thereby, enriched GO terms were identified within a target list of genes compared to a background list. As target list either up-or downregulated genes from DGE analysis were used. The background list consisted of all analyzed genes of the transcriptome array. Enrichment was calculated by the following formula: Enrichment = b/n B/N N = total number of genes B = total number of genes associated with a specific GO term n = the number of genes in the target set b = the number of genes in the intersection The p-values for enriched GO terms were calculated using Fisher's exact test. Heat maps of signal values were generated in Excel.

Conclusions
In conclusion, the results of our transcriptome analysis suggest that Lgr5 + cells do not harbor stem cell properties. However, Lgr5 + cells are a heterogeneous population of amacrine subtypes. Possibly a small subtype of Lgr5 + cells do possess stem cell properties. Yet the employed techniques were not suitable in detecting them. Especially in the transcriptome analysis, the expression of stem cell markers in a small population may be masked. Therefore, single cell analysis to investigate such populations would have to be performed to further characterize properties of Lgr5 + cells in the retina.