Neural stem cells (NSCs) are the principal source of all cells found in the nervous system. Embryonic NSCs have the potential to generate the full array of neurons and glia in the brain. In contrast, adult NSCs are restricted in their potential and produce neurons in only specific regions of the forebrain. However, even adult NSCs can be coaxed to produce a broad range of neural cells in the appropriate culture conditions. In the postnatal mammalian brain, NSCs are retained in the specialized microenvironment called the neurogenic niche. The various environmental cues provided by the niche converge onto NSCs, which in turn utilize their intrinsic factors to respond appropriately for proper control of neurogenesis. In this study, we investigated the nature of the communication between NSCs and their environment that may control stem cell maintenance, differentiation, and lineage specification. We used the signal sequence trap method for the systematic exploration of autocrine/paracrine as well as cell-cell contact-based signaling factors derived not only from NSCs but also from their neighboring niche cells. Here, we present the SMEPs of NSCs and other niche cell types, including TAPs, non-neurogenic astrocytes, ependymal cells, vascular endothelial cells, and the choroid plexus.
Because multiple niche cell types are tightly associated in the SVZ, the isolation of each SVZ niche cell type at high purity is the critical first step for obtaining the cell type-specific expression profiles. There have been numerous attempts to isolate NSCs and surrounding niche cells in highly purified forms by FACS. Mostly they relied on the use of a single transgenic reporter mouse line 
and/or labeling cell surface markers or intracellular markers through permeabilization 
. Since NSCs and niche cells do not have a single bona fide marker, another study used a dual-labeling method to enrich the NSC population 
. These studies demonstrate the value of the transgenic mice for isolation of specific SVZ cell types, especially in combination with multiple overlapping markers. However, the additional steps required for labeling with ligands or antibodies result in a prolonged incubation period of freshly dissociated cells, raising the possibility of altering the dynamics of gene expression profiles in processed cells. Our approach combined the genetic inducible fate mapping system and transgenic reporter mice (Gli1CreER/+;hGFAP-GFP;R26tdTomato/+
, and Tie2-GFP
mice) to isolate NSCs, TAPs, astrocytes, ependymal cells and endothelial cell from the SVZ. It is a highly efficient and reproducible method to isolate specific cell types because it is based on the genetic labeling and nascent fluorescent protein expression and does not require additional processing steps. Using this method, we were able to isolate highly enriched cell types from the SVZ niche as our validation experiments demonstrate.
Previous studies demonstrated that a subset of FoxJ1+ cells in the SVZ express GFAP and these FoxJ1+/GFAP+ cells have NSC-like features such as neurosphere formation 
. One scenario is that a subset of dividing SVZ astrocytes (GFAP+) incorporates within the ependymal cell layer and those incorporated cells acquire antigenic and morphological properties of mature ependymal cells, which is more often observed in the aged brain 
. Interestingly, during this transition state, incorporated cells co-express GFAP as well as ependymal cell markers 
. Thus, it is likely that a subset of newly born SVZ astrocytes incorporate into the ependymal cell layer, gradually lose GFAP expression, while obtaining ependymal cell features 
. Together, our results indicate that the FoxJ1
-YFP+ cell population contains GFAP-expressing cells but the majority of FoxJ1
-YFP+ cells are ependymal cells expressing CD24 and S100β.
Our SMEP data include well-known or previously identified cell type-specific genes such as Ptgds
(choroid plexus) 
(endothelial cells) 
. Besides these two genes, the expression patterns of multiple genes we identified by SST-REX have been reported in certain SVZ cell types (Table S3
). Additionally, we identified some novel genes that are enriched in each cell type that could have potential relevance for the niche signals. For example, we showed that the exogenous treatment of Ttr on neurosphere culture resulted in less neurosphere formation and reduced cell proliferation. Although Ttr is the major thyroid hormone (T4
) supplier in the brain, there was only T3
in our neurosphere culture medium (B-27 supplement, Invitrogen). Because Ttr can also bind to T3
but with a lower affinity than T4
, Ttr might bind to and sequester T3
, thereby reducing bioavailability of T3
in the culture medium. In agreement with our result, it was shown that an antithyroid drug that induced experimental hypothyroidism reduces the NSC proliferation 
. Interestingly, T4
levels in the brain parenchyma of Ttr null mice were not reduced 
and the number of dividing cells in the SVZ of adult Ttr null mice is comparable to that of Ttr wild-type mice 
, suggesting the possibility of a thyroid hormone-independent role for Ttr. Because the function of Ttr on NSC behaviors is largely unknown, further studies are necessary to elucidate its role as a possible niche signal.
The intracellular role of CPE in pro-peptide processing and sorting is well documented 
, but little is known about the autocrine/paracrine function of secreted CPE. Our in
neurosphere model uncovered a novel effect of secreted CPE. Treatment with the exogenous CPE resulted in fewer neurospheres and lower cell proliferation, suggesting that CPE can act as a negative regulator of NSC cell proliferation. This effect seems to be independent of the CPE’s carboxypeptidase activity because the culture medium we used was at physiological pH (~7.4), which is not the optimal pH of 5.6 for the enzymatic activity of CPE in the secretory granules 
. Such enzymatic-independent function of CPE could be achieved through receptor-mediated signaling. Based on the structural homology between the catalytic site within the amino-terminal domain of the Shh protein (Shh-N) and the catalytic site common to all carboxypeptidases, including CPE, it was speculated that CPE may work in a similar manner as a signaling molecule 
, but no experimental evidence has been reported. Alternatively, because CPE has a binding affinity for proteins other than pro-peptides 
, CPE may bind to selective proteins/peptides in the culture medium such as insulin or growth factors, thereby interfering with ligand-receptor interactions. However, the detailed underlying mechanism of the autocrine/paracrine function of CPE remains to be determined.
Although SST-REX is a useful method to screen secretory or membrane proteins, there exist some technical limitations including the limitation in the profile size. Because SST-REX is not a full-scale screening process at a saturated level, i.e., to screen the entire cDNA library, we could only test a limited number of the surviving factor-independent clones. Continued and repeated screening of the same library could increase the profile size. We initially identified a total of 307 cDNAs with the homology to known genes. However, we further limited our analysis to only 151 genes in our final SMEP to focus on only secretory and membrane associated molecules. Besides secretory and membrane associated molecules, approximately 40% of the initially identified genes encoded nonsecretory proteins. In agreement with our result, previous signal sequence trap studies have reported 25 to 50% false-positive rate, i.e., encoding proteins that are known to be not secreted 
. One possible explanation for these false positives could be the transduced cDNAs that encode cell cycle regulatory proteins or cytoplasmic signaling molecules allowed IL-3-independent survival of Ba/F3 cells in our screen. For example, we identified cell cycle regulatory genes such as the cell division protein kinase 19 (Cdk19
) and cytoplasmic signaling molecules such as mitogen-activated protein kinase 1 (Mapk1
) in our SMEPs. Furthermore, the fact that subcellular organelle membrane proteins contain N-terminal hydrophobic residues, which might act as signal peptides 
, can further increase the false positive hits from the screening. Some of the expected molecules were not identified by our SST-REX (false-negative). For example, epidermal growth factor receptor (EGFR) protein is a type I transmembrane glycoprotein with the N-terminal signal sequence. EGFR is known for its expression in TAPs 
but was not identified in our screening. Possible reasons for the false negatives include, 1) not present in the library due to primer or extension failure, 2) the expression of these proteins interferes Ba/F3 cell growth, 3) not screened at a saturated level 
For a systematic approach to identify NSC niche signals, various screening methods can be applied and each method has advantages and disadvantages. For example, SST-REX is useful to screen secretory or membrane proteins, but there are technical limitations as described above. Although cDNA microarrays aim for the genome-wide gene expression profiling, it can only detect genes that are already represented on the chip. Thus, combining data from different approaches will further improve the magnitude of gene expression profiling. Indeed, the identified genes in our NSC SMEP obtained by SST-REX did not overlap substantially with the potential NSC SMEP acquired from our microarray method, indicating that each approach could complement each other. Therefore, we employed a bioinformatics tool to refine the final SMEP and compared the cDNA microarray results as an alternative approach to widen the profile of NSCs for the secretory and cell-surface molecules. By combining SST-REX and microarray data, we identified several signaling molecules, which are known to modulate NSC behaviors. We found signaling pathway related molecules such as Shh receptor patched homolog 1 (Ptch1
, Wnt receptor frizzled homolog 2 (Fzd2
, Notch gene homolog 2 (Notch2
, and bone morphogenic protein receptor (Bmpr1a
and Table S5
). We also found interesting molecules that may be linked to NSC function, for example pleiotrophin (Ptn
), midkine (Mdk
), and protein tyrosine phosphatase receptor type Z polypeptide 1 (Ptprz1
) ( and Table S5
). Ptprz1 is a known receptor for Ptn and Mdk 
and it has been shown that they are involved in the growth of human embryonic stem cells 
, cerebellar Purkinje cell differentiation 
, and embryonic neural stem/progenitor cell survival/growth in vitro
. Since their potential roles in adult NSCs are not known, they can be candidate niche signaling molecules acting in an autocrine or paracrine manner.
To understand the nature of niche signals, it is important to consider how each niche cellular component contributes to the signaling network for maintaining the niche homeostasis. For this, it is indispensible to obtain the collective profile of each niche cell type. In this study, we report the first systematic approach to provide the SMEP of NSCs as well as other neighboring niche cells in the adult SVZ. Our approach helps to understand the molecular signature of the NSC niche and how niche signals can be orchestrated for the maintenance of homeostasis. It further demonstrates the importance of understanding the role of each niche cell type individually as well as in an integrated manner. Combining our niche cell isolation methods with genomics tools such as microarray or RNA-sequencing technologies will provide us further insights into the nature of NSC niche signals.