Development and Notch Signaling Requirements of the Zebrafish Choroid Plexus

Background The choroid plexus (CP) is an epithelial and vascular structure in the ventricular system of the brain that is a critical part of the blood-brain barrier. The CP has two primary functions, 1) to produce and regulate components of the cerebral spinal fluid, and 2) to inhibit entry into the brain of exogenous substances. Despite its importance in neurobiology, little is known about how this structure forms. Methodology and Principal Findings Here we show that the transposon-mediated enhancer trap zebrafish line EtMn16 expresses green fluorescent protein within a population of cells that migrate toward the midline and coalesce to form the definitive CP. We further demonstrate the development of the integral vascular network of the definitive CP. Utilizing pharmacologic pan-notch inhibition and specific morpholino-mediated knockdown, we demonstrate a requirement for Notch signaling in choroid plexus development. We identify three Notch signaling pathway members as mediating this effect, notch1b, deltaA, and deltaD. Conclusions and Significance This work is the first to identify the zebrafish choroid plexus and to characterize its epithelial and vasculature integration. This study, in the context of other comparative anatomical studies, strongly indicates a conserved mechanism for development of the CP. Finally, we characterize a requirement for Notch signaling in the developing CP. This establishes the zebrafish CP as an important new system for the determination of key signaling pathways in the formation of this essential component of the vertebrate brain.


Introduction
The choroid plexus (CP) is a set of vital structures in the brain central to the formation, regulation and protection of the cerebral spinal fluid (CSF). Development and physiology of the CP has implications for many central nervous system pathologies and for the targeting of pharmaceutical treatment to the central nervous system [1,2]. In humans and other mammals, the CP consists of four independent structures, one in each of the four ventricles of the brain [1]. On the cellular level, the structure can be divided into three separate compartments, a vascular core, a stroma, and a layer of polarized epithelia [1]. The polarized epithelial cells surround the stroma and vascular core and exist as a monolayer, with their basal surface facing the stroma, and their apical surface extending microvilli and cilia into the CSF-filled ventricles [3].
Classic studies in human embryology place the onset of CP development at 6 weeks gestation and continuing past birth [1]. Comparative studies in mice place development of the CP beginning at embryonic day 11 [4]. In humans, the first of the choroid plexuses to develop is the fourth ventricle CP (4vCP), or its orthologue the myelencephalic choroid plexus (mCP), in other organisms including telosts [1,5]. However, despite its importance in neurobiology, little is known about how this important structure forms.
We use the transgenic line Et Mn16 , which express GFP in a subset of definitive mCP epithelial cells, to characterize the development of the zebrafish mCP in vivo. The subset of the mCP primordial epithelia are initially found in a diffuse pattern along the dorsal roof of the fourth ventricle and coalesce at the midline to form the definitive mCP. Upon coalescence, the dorsal longitudinal vein (DLV) provides the small vessel bed that invades the epithelium of the mCP. Time-dependent inhibition of Notch signaling defines a requirement for this pathway in CP development. Morpholino-based ligand/receptor inhibition studies map this critical signaling event to the notch1b receptor and the two candidate ligands, dla and dld. Together, this study provides the first description of CP development in vivo and demonstrates a role for Notch signaling in mCP development.

Identification of the zebrafish CP
Et Mn16 is a GFP expressing transposon-mediated enhancer trap line identified as part of an ongoing genomic annotation project [6]. At 30 hours post fertilization (hpf), cells of the fin bud, a symmetric set of cells that extend processes medially in the ventral hindbrain and spinal column, and the otic vesicles express GFP ( Fig. 1A-C). At 4 days post fertilization (dpf), we observe two additional structures expressing GFP. The first structure appears at the dorsal-most point of the fourth ventricle on the midline (Fig. 1D, arrowhead). The second structure is situated anterior to the eyes at the dorsal-most point of the third ventricle on the midline (Fig. 1D, arrow). Based on comparative anatomical studies in multiple organisms including telosts [1], we attribute the anterior GFP-expressing cells as those of the diencephalic choroid plexus Glial fibrillary acidic protein (Gfap) is an early transient marker of the CP epithelia [7]. We used immunohistochemistry on 6 mm longitudinal cryosections of 5 dpf mCP with a monoclonal antibody (zrf-1) directed against zebrafish Gfap [8]. We observe Gfap expression in the epithelial cells of the mCP (Fig. 1G) as defined by colocalization of Gfap (red) and GFP (Fig. 1H, merged 1J). This colocalization of Gfap and GFP-expressing cells in Et Mn16 larvae in conjunction with anatomical analyses indicates Et Mn16 labels the forming zebrafish mCP.

Development of the mCP as visualized in the Et Mn16 Larvae
CP formation was imaged in vivo using Et Mn16 zebrafish larvae. GFP-expressing presumptive mCP are first observed in the roof plate of the fourth ventricle at 2.5 dpf (Fig. 2B). At 3 and 4 dpf, the cells progressively coalesce (Fig. 2C,D) toward the midline forming a tight rounded structure by 5 dpf (Fig. 2E). A subset of the cells within the mCP do not express GFP at this time ( Fig. 2G, arrowhead). The structure remains visibly unchanged at 6 dpf, indicating the labeling of the definitive zebrafish CP (Fig. 2F).

The vasculature of the mCP is supplied by the DLV
The CP is tightly associated with vasculature in mammalian systems [9] as part of the blood brain barrier. We visualized functional blood vessel integration in the CP using Et Mn16 / Tg(gata1:dsRed) double transgenic fish, with the latter transgene expressing DsRed in the circulating red blood cells [10]. Blood vessel identities were established utilizing the zebrafish vasculature map [11]. At 5 dpf, the mCP is at the bifurcation of the Dorsal Longitudinal Vein (DLV) as this vessel progresses into the Posterior Cerebral Vein (PCeV and PCeV9, prime indicates opposing bilateral side) ( Fig. 3A-C). The bifurcation occurs shortly after the entry into the mCP domain. Three-dimensional renderings of the blood flowing through the DLV and mCP demonstrate that the vein is tightly associated with the CP epithelia ( Fig. 3D-F). In addition, the DLV makes a slight dorsal bend just before the bifurcation, locating this vessel on the dorsal side of the mCP opposite the ventricle (Fig. 3G-H). The DLV then flattens as it joins the PCeV and PCeV9, transiting the mCP. After the PCeV and PCeV9 exit the mCP, these vessels again turn ventrally.
We further characterized mCP vasculature by describing the development of the DLV in more detail utilizing the Tg(fli1:eGFP) zebrafish transgenic line that labels vascular endothelial cells with GFP [12]. The DLV begins development between 40-48 hpf by sprouting from the Medial Cerebral Vein (MCeV and MCeV9) (Fig. 4B). The DLV extends by angiogenesis extending a growth cone like structure (Fig. 4C), until it reaches a hypothesized decision point (Fig. 4D). At this time, blood can be seen within the lumenized vessel, but remains static as it lacks a functional vascular outlet (data not shown). The vessels of the PCeV and PCeV9 migrate dorsomedially to meet the DLV (Fig. 4D), appearing to be attracted to the locale independent of the DLV (Fig. S1E). The DLV then develops preferentially to one of the PceV's where this vessel will connect and establish blood flow (Fig. 4E). A branch will then proceed toward the other PCeV and connect (Fig. 4F). The connection of the DLV and PCeV's approximately corresponds to the initial point at which GFP-expressing cells are observed in the Et Mn16 line. At around the same time, the MCeV-DLV junction fuses with the Mesencephalic Vein (MsV) forming the Dorsal Midline Junction (DMJ) (Fig. 4E-G). Concurrently, both the MsV and the MCeV supply the DLV with blood, with the MsV becoming the primary supply by 120 hpf (data not shown). Lastly, a cross-vein will form between the DLV-PCeV junction and the DLV-PCeV9 junction crossing the mCP between 3 and 5 dpf (Fig. 4G,H). We label this previously unannotated cross-vein, the trans-choroid plexus branch (TCB) as it transits the mCP. Further elaboration of the small vessels that supply the mCP continues past 5 dpf. This network is highly variable with vessels extending between the DLV and TCB ( Fig. 4H and Fig. S1).

The Notch signaling pathway in mCP development
Multiple Notch ligands are expressed within the developing mammalian CP [13], with no known requirement for Notch signaling yet described. We used the gamma secretase inhibitor N-[N-(3,5-Difluorophenacetyl-L-alanyl)]-S-phenylglycine t-butyl ester (DAPT) [14] for a temporal-specific knockdown of Notch signaling during CP formation. DAPT was applied to embryos at 50 hpf to avoid indirect effects caused by early embryologic patterning defects. DAPT treatment resulted in a dramatic increase in mCP size in Et Mn16 larvae [from a mean of 3200 mm6160 s.e.m. in vehicle-treated Et Mn16 to a mean size of 5000 mm6150 s.e.m. (p = 1.98*10 211 ) in DAPT-treated larvae ( Fig. 5H)]. This appears to be due to a lateral expansion as the mCP remains a monolayer (Fig. 5F).
As DAPT is a pan-Notch signaling inhibitor, we sought to specifically identify the individual Notch receptor and its ligand(s) involved in mCP formation. We conducted a reverse genetic screen utilizing morpholino oligonucleotides previously designed and validated against ten known zebrafish Notch pathway members [15][16][17][18][19]. We first screened morpholinos targeting four Notch receptors: notch1a, notch1b, notch2, and notch3 (Fig. 5H). Treatment with 0.5 ng of notch1b-targeted morpholino leads to expansion of the mCP epithelia (mean 5700 mm6320 s.e.m. p = 3.8*10 29 ) ( Fig.  5D) compared to wild type (3300 mm6120 s.e.m.) (Fig. 5A). We then screened morpholinos targeting Notch ligands: jagged1a, jagged1b, jagged2, deltaA (dla), and deltaD (dld) (Fig. 5I). The mCP in dld-targeted morphant larvae is significantly larger (5100 mm6280 s.e.m) compared to untreated Et Mn16 controls (3000 mm695 s.e.m.)(p = 1.9*10 28 ) (Fig. 5E). Knockdown of dla also increases the mCP size significantly (4600 mm6630 s.e.m. p = 0.02), but this effect is much less consistent than the effect of dld (Fig. 5I). In addition, while the size of the mCP epithelial domain is not significantly different in notch1a-targeted morphants compared to controls, there is a visible effect on the mCP structure in morphant larvae (Fig. 5C) with a loss of a cohesive mCP epithelial sheet. Therefore, we conclude that knockdown of notch1b, dla, and dld significantly expands the domain occupied by the mCP epithelia. Together, these analyses implicate a novel Notch pathway critical role in CP formation.

Identification of the CP as visualized in the Et Mn16 larvae
We demonstrate that the GFP-expressing cells of Et Mn16 are dCP and mCP epithelium. The mCP in zebrafish is located within the fourth ventricle posterior to the cerebellum (Fig. 1E,F), consistent with the location of the 4vCP/mCP in mammals and other teleosts [1,20]. The mCP of zebrafish expresses Gfap, a transient fetal marker for human 4vCP [7]. This is the first description of the CP in zebrafish.

Development of the mCP epithelia
We describe Et Mn16 GFP-expressing cells that migrate laterally along the roof plate of the fourth ventricle, coalesce and become a subset of the definitive mCP epithelia. Garcia-Lecea and coworkers describe a second enhancer trap SqET33-E20 (Gateways) transgenic zebrafish that express GFP within cells that migrate along the anterior posterior axis at the midline, coalesce and form a distinct subset of definitive mCP epithelia [21]. These SqET33-E20-labeled cells are temporally and spatially separated from the cells we describe marked by Et Mn16 . The Et Mn16 transgenic line labels cells highly reminiscent of cell lineages described by Hunter and Dymecki that migrate out of the dorsal rhombomeric lip and directly populate the mouse mCP [22]. We conclude that these migrating cells found in fish and mice are orthologous and the resulting developmental model represents a likely shared and conserved mechanism for CP development in vertebrates.

The vasculature of the mCP is supplied by the DLV
We characterize the mCP-associated vasculature in zebrafish. The DLV supplies the blood to the mCP, bifurcating just inside the margin of the mCP epithelia, and forming junctions with both the PCeV and PCeV9 (Fig. 3A-F). The DLV bends dorsally just before crossing the mCP margin and is tightly associated with the dorsal surface of the mCP epithelial monolayer (Fig. 3G,H). The DLV sprouts from the MCeV and MCeV9 between 40-48 hpf via angiogenesis. The DLV reaches a hypothesized decision point at ,50 hpf, and the PCeV and PCeV9 grow dorsomedially to meet it. The trans-choroid plexus branch (TCB) develops between the DLV-PCeV junction and the DLV-PCeV9 junction between 55-120 hpf. This forms a variable structure that overlays the mCP epithelia, and produces a network of small vessels that connect the bifurcated DLV and TCB. This is the first description of the development of the vasculature associated with the mCP and has many similarities to the development of mammalian mCP development [9]. Future studies will be required to identify whether these small vessels that cross between the DLV and TCB are orthologous to the fenestrated capillaries of mammals.

Synthesis of a model for zebrafish mCP development
By comparing the developmental time points at which the DLV and mCP epithelia develop (Fig. 2, Fig. 4) [21], we can precisely order developmental events in the mCP. Garcia-Lecea and coworkers describe a subset of mCP epithelial precursor cells that are initially spread out at the dorsal midline from rhombomere 2-6 that coalesce between 48-72 hpf and express GFP in their SqET33-E20 line [21]. The DLV and PCeV develop to the location associated with the mCP synchronously with the GFPexpressing cells of SqET33-E20. Intriguingly, the developmental cue is independent of the vasculature (Fig. S1E) suggesting an attractant that promotes growth to the fourth ventricle roof plate. Previous work has suggested the existence of such an organizer in mouse [23][24][25][26]. After the DLV has completed junction formation with the PCeV and PCeV9, the subset of Et Mn16 defined mCP migrate laterally coalescing on the roof of the fourth ventricle, by 5 dpf. This model is the first to integrate both the cellular migration of the mCP epithelial precursors with that of the development of the vasculature and lays the groundwork for further vascular studies. For example, the zebrafish should be a very tractable model to identify, the dorsal roof organizer and the factors that localize both vasculature and epithelial development to the fourth ventricle roof plate.

The Notch signaling pathway is required for mCP development
We undertook a small-scale targeted chemical and morpholinomediated screen to investigate the role of Notch signaling in mCP development. The pan-Notch signaling chemical inhibitor (DAPT), when exposed to 50 hpf Et Mn16 embryos increased the size of the developing mCP epithelia. To specify those members of the Notch pathway that are involved, we utilized previously validated Notch and Notch ligand targeting morpholinos to investigate the individual effect of several notch receptors and ligands. The notch1b, dla, and dld-targeted mopholinos increase the size of the mCP similar to the DAPT treatment, while other Notch and Notch ligandtargeting morpholinos did not significantly alter mCP size. It should be noted, however, that while other morpholinos were unable to change CP size, there may have been other effects on the structure as exemplified by a notch1a knockdown mCP (Fig. 5c) that has an increase in GFP-negative surface area within the mCP structure. Previous reports on Notch signaling in the mammalian CP have shown that multiple Notch ligands are expressed in the CP [13], a novel Notch2-mediated ligand independent role for adult rat CP integrity [27], and that expression of a constitutively activated Notch1 protein causes an increase in the proliferative potential of the CP precursor cells by activating mitosis in normally quiescent cells [22]. This study presented here identifies an endogenous requirement for notch1b in development of the mCP and a functional role for any Notch ligand (dla and dld) in the mCP. Future studies will be required to determine the relationship of Notch1a, Dla, and Dld, the localization of these proteins, and the nature of the expanded mCP including the role of the vasculature in determining the domain occupied by the CP.

Conclusion
We provide the first description of the zebrafish CP by characterizing a transgenic line (Et Mn16 ) that expresses GFP in the zebrafish dCP and mCP epithelia. We observe a tight integration between the mCP and its vascular supply, primarily the DLV and TCB. We provide a model that integrates development of the epithelial and vascular components of the mCP. In addition, we demonstrate a functional role for endogenous Notch signaling in mCP epithelia. We identify a Notch receptor and two Notch ligands involved in the lateral expansion of the mCP epithelial layer. This model provides a unique system in which to rapidly expand the knowledge of the molecular signaling pathways that contribute to CP development.

Creation of the Et Mn16 line
Et Mn16 is a product of a Sleeping Beauty (SB) transposon mediated-enhancer trap screen previously described [6]. In brief, a weak EF1a enhancer element upstream of zebrafish optimized GFP (GM2) was placed in between the SB inverted-direct repeats (pT2/S2EF1a-GM2). Co-injection of this construct with SB transposase mRNA was performed into 1-2 cell embryos and propagated as described [29,30].

Microinjection of morpholinos
Zebrafish were collected and injected as described previously [33]. Morpholino oligonucleotides were obtained from GeneTools LLC. To investigate Notch signaling during CP formation, we used published oligonucleotides against Notch, Jagged and Delta transcripts from the zebrafish (Table S1) [15][16][17][18][19]. For each morpholino, we confirmed that there was no significant sequence overlap between oligonucleotides targeting these highly related proteins (Table S1). The notch1b-targeted morpholino was coinjected with tp53-targeted morpholino (1.56) to verify effects were not due to off-targeting [34].

Notch inhibitor treatment
We used the gamma-secretase inhibitor DAPT (Sigma) to inhibit all Notch signaling in zebrafish [14]. To identify the window of Notch sensitivity, we added DAPT at several different developmental time-points (data not shown). We determined treatment at 50 hours post fertilization was optimal. Fish were dechorinated and placed in medium Petri dishes with 6ml of embryo medium. 100 mM DAPT (suspended in DMSO) was added at various doses as described previously [35]. 6% DMSO in embryo medium (v/v) was used as a vehicle control.

Microscopy and Imaging
All images were taken on a Zeiss Axioplan 2 with Apotome utilizing the Axiocam mRM (Zeiss). For brightfield microscopy, fish were imaged under DIC optics. For imaging CP development, zebrafish were mounted in 1% low melting point and anesthetized using tricane [28]. Images were processed using Axiovision software with Image 4D (Versions 4.2 and 4.6; Zeiss), and Photoshop (Version 8; Adobe).

Measurements and Statistics
The surface area of the mCP (as defined by Et Mn16 fluorescence) was measured using the outline measurement module (Axiovision 4.2 and 4.6). Values were entered into graphing software, Prism (Version 4; GraphPad) for histograms and normality statistics. Normality was determined for each of the distributions utilizing the D'Agostino & Pearson omnibus normality test (alpha = 0.05) and Shapiro-Wilk normality test (alpha = 0.05) (data not shown) for each of the treatments to verify that the Student T-test was appropriate (Version 4; GraphPad). Student T-tests were performed to determine statistical significance (two tailed with unequal distribution) in Excel (version 11.4.1; Microsoft). Figure S1 Variation of the DLV-PCeV junction. The DLV (arrows) bifurcated to join the PCeV and PCeV9 (open arrowheads) and the TCB (small arrowheads) connects the two junctions. This structure is plastic. The most common structure includes a third branch that transits between the DLV and the TCB (A-C), but in some instances the third DLV-TCB branch does not form by 5 dpf (D). The most rare occurrence is that the DLV does not develop (E). The PCeV and PCeV9 still meet in the location of the pCP, but lack blood flow (E, open arrowheads). All images are shown with anterior to the left. Scale bar is 50 mm.