Directional cerebrospinal fluid movement between brain ventricles in larval zebrafish
- 3k Downloads
Cerebrospinal fluid (CSF) contained within the brain ventricles contacts neuroepithelial progenitor cells during brain development. Dynamic properties of CSF movement may limit locally produced factors to specific regions of the developing brain. However, there is no study of in vivo CSF dynamics between ventricles in the embryonic brain. We address CSF movement using the zebrafish larva, during the major period of developmental neurogenesis.
CSF movement was monitored at two stages of zebrafish development: early larva [pharyngula stage; 27–30 h post-fertilization (hpf)] and late larva (hatching period; 51–54 hpf) using photoactivatable Kaede protein to calculate average maximum CSF velocity between ventricles. Potential roles for heartbeat in early CSF movement were investigated using tnnt2a mutant fish (tnnt2a −/−) and chemical [2,3 butanedione monoxime (BDM)] treatment. Cilia motility was monitored at these stages using the Tg(βact:Arl13b–GFP) transgenic fish line.
In wild-type early larva there is net CSF movement from the telencephalon to the combined diencephalic/mesencephalic superventricle. This movement directionality reverses at late larval stage. CSF moves directionally from diencephalic to rhombencephalic ventricles at both stages examined, with minimal movement from rhombencephalon to diencephalon. Directional movement is partially dependent on heartbeat, as indicated in assays of tnnt2a −/− fish and after BDM treatment. Brain cilia are immotile at the early larval stage.
These data demonstrate directional movement of the embryonic CSF in the zebrafish model during the major period of developmental neurogenesis. A key conclusion is that CSF moves preferentially from the diencephalic into the rhombencephalic ventricle. In addition, the direction of CSF movement between telencephalic and diencephalic ventricles reverses between the early and late larval stages. CSF movement is partially dependent on heartbeat. At early larval stage, the absence of motile cilia indicates that cilia likely do not direct CSF movement. These data suggest that CSF components may be compartmentalized and could contribute to specialization of the early brain. In addition, CSF movement may also provide directional mechanical signaling.
KeywordsCerebrospinal fluid Brain ventricular system Fluid dynamics Zebrafish
2,3 butanedione monoxime
combined diencephalic/mesencephalic superventricle
volumetric flow rate
Selective Plane Illumination Microscopy
Cerebrospinal fluid (CSF) is contained in the brain ventricles (chambers derived from the brain lumen) and contacts neuroepithelial progenitor cells during brain development and in adulthood. The CSF contains a unique composition of proteins (CSF proteome) [1, 2, 3, 4, 5, 6], small molecules [5, 7] and lipid particles  for which limited studies have demonstrated function during brain development [1, 2, 3, 4, 5, 7, 8, 9]. Since the brain lumen forms before the choroid plexus (CP), ventricles exist with no requirement for a CP to be associated with each ventricle. In some cases early pre-CP fluid is called ependymal fluid . The cohort of proteins in zebrafish CSF is largely overlapping with early amniote CSF prior to CP development  and this conservation has not been investigated at later stages. Once the CPs have become functional they secrete most of the ventricular fluid, although there is clear data that fluid continues to be produced from ependymal cells even after CP differentiation is complete [11, 12, 13, 14]. The ventricular fluid is thus always a mix from different sources and the term CSF is most general and appropriate to refer to the ventricular fluid at all stages.
Restricted or directional CSF movement, either during production and drainage or coupled to heartbeat [15, 16, 17, 18, 19], could limit locally-produced factors to specific regions of the developing brain. Studies in Xenopus have investigated local intraventricular CSF dynamics in the developing brain [19, 20], but there has been no direct study of in vivo CSF movement dynamics between ventricles. Recent studies have suggested that cilia on neuroepithelial cells may detect a mechanical signal to regulate neuronal progenitor differentiation . The nature of this signal is unknown but could be CSF movement. CSF fluid dynamics have been examined in zebrafish after CP formation , but have not been studied earlier in development during the major period of developmental neurogenesis. To address these deficits, we investigate CSF movement in the accessible larval zebrafish brain, prior to CP formation. Our data demonstrate directional movement of CSF, which may function during brain development.
Fish lines and maintenance
Danio rerio fish were raised and bred according to standard methods . Embryos were kept at 28.5 °C and staged according to Kimmel et al. . Times of development are expressed as hours post-fertilization (hpf) or days post-fertilization (dpf). Analyses were performed at early larva (pharyngula stage; 27–30 hpf) stage, late larva (hatching period; 51–54 hpf) stage, and some at 5 dpf. Lines used were: AB; Tg(βact:Arl13b–GFP) transgenic line ; troponin T type 2a (cardiac) mutants tnnt2a −/− (R14GBT0031/silent heart; obtained from The Zebrafish International Resource Center (ZIRC; Eugene, USA)) . Larvae were incubated in 3 μg/mL 1-phenyl-2-thiourea (PTU) starting between 22 and 24 hpf to inhibit maturation of melanocytes and allow for clear visualization. All experimental procedures on live animals and embryos were reviewed and approved by the Institutional Animal Care and Use Committee of the Massachusetts Institute of Technology (Protocol# 0414-026-17) and were carried out in accordance with the recommendations in the National Institutes of Health (NIH) Guide for the Care and Use of Laboratory Animals.
Multiview SPIM brain ventricle imaging
Larvae were anesthetized in 0.1 mg/mL (0.38 mM) Tricaine (Sigma, St. Louis, USA) dissolved in embryo medium  prior to injection and imaging. At high doses, Tricaine can induce death or stop the heart [27, 28, 29], but these effects have not been shown at the concentration used here. Direct effects of Tricaine on CSF movement have not been investigated.
The rhombencephalic ventricle was microinjected with 2 nL of 2000 kDa dextran conjugated to fluorescein (ThermoFisher/Invitrogen Molecular Probes, Waltham, USA; D7137) diluted to 25 mg/mL in sterile ddH2O. Embryos were embedded in 1 % low-melting point agarose in a glass capillary, and selective plane illumination microscopy (SPIM) was performed using a Zeiss LightsheetZ.1 microscope within 60 min of injection. Six multi-view images were acquired (60° apart), and were processed using ZEN software (Zeiss, Oberkochen, Germany), FIJI [30, 31], and the Multiview reconstruction plugin [32, 33]. Images were analyzed using Imaris software (Bitplane, Belfast, UK) to visualize 3-D images and to generate volume measurements for ventricles at each age. The image threshold was set so that both the ventricle outline and the spinal canal were detected. For rhombencephalic ventricular volume measurements, the same threshold as above was used, but only the rhombencephalon was included in the region of interest. Injection does not affect development, survival or ventricle size over the time period studied ( and Additional file 1: Fig. S1).
Tissue histology and hematoxylin and eosin staining
Larvae were anesthetized as described above and fixed overnight in aqueous Bouin’s fixative (RICCA Chemical, Pokomoke City, USA). Larvae were washed in 70 % EtOH until supernatant was clear (at least six washes of 1 h each at room temperature). Larvae were oriented in Richard Allan Scientific HistoGel (ThermoFisher, Waltham, USA). The HistoGel discs containing oriented larvae were fixed for 1 h in Bouin’s fixative, and subsequently washed in 70 % EtOH until supernatant was clear. Samples were infiltrated with paraffin using a Tissue-Tek VIP5 Vacuum Infiltration Processor (Sakura Finetek, Torrance, USA; model 5215). Sections were cut at 4–5 μm and slides were stained with Harris acidified hematoxylin (3 min) and eosin Y (10 s) using the “standard” H&E protocol  on a Shandon Varistain Gemini Slide Stainer (ThermoFisher, Waltham, USA). Slides were imaged using an Axioplan 2 (Zeiss, Oberkochen, Germany) equipped with a Q-Capture MicroPublisher 5.0 RTV camera/software (QImaging Scientific, Surrey, Canada) using 10× and 40× air objectives and 63× and 100× oil-immersion objectives.
Live time-lapse confocal imaging
Image processing and fluid movement quantification
Q was then smoothed to negate technical noise using a 9-point moving window average. To control for individual variations in ventricle shape, Qt was converted to linear velocity (v) by dividing Q by the cross-sectional area of the aqueduct. Cross-sectional areas were measured using Imaris software (Bitplane, Belfast, UK) by creating surfaces of the unconverted (green) Kaede signal in the first scan (before photoconversion) at the narrowest point between the two ventricles being analyzed. The maximum velocity number (vmax) was used to analyze brain ventricle CSF movement (see Fig. 1c).
Acute inhibition of heartbeat
Heartbeat was acutely inhibited by soaking ventricle-injected embryos in 40 mM 2,3 butanedione monoxime (BDM; Sigma, St. Louis, USA) solubilized in embryo medium  for 10–30 min prior to imaging as previously shown . This treatment stops heartbeat, but does not change the gross ventricular anatomy within 2 h of incubation (Additional file 1: Fig. S2). BDM prevents action of all muscles (and can alter activities of connexins , potassium channels [40, 41], and L-type calcium channels ) and is therefore less precise for disruption of heartbeat than the tnnt2a −/− mutant.
The Tg(βact:Arl13b–GFP) transgenic line was used for visualization of cilia. Embryos were treated with 40 mM BDM, as described above for 10 min to stop heartbeat and enable completely steady, immotile embryos suitable for high-speed imaging. They were prepared for SPIM as described above and incubated in 20 mM BDM and 0.1 mg/mL Tricaine (Sigma, St. Louis, USA) in embryo medium during imaging. Images were collected at 57.22 frames per second (fps) in the telencephalon, diencephalon/mesencephalon, rhombencephalon and spinal canal. Time-lapse movies were registered using FIJI [30, 31] and the StackReg plugin . The ciliary beat frequency (CBF) was calculated using the following formula: [CBF = (number of frames per second)/(average number of frames for single beat)] . Images were processed by overlaying sequential frames in pseudocolor using Photoshop (Adobe).
The early zebrafish brain ventricular system is complex and dynamic
Ventricular shape changes over the first 2 days of development; with the combined diencephalic/tectal (D/Te) ventricle developing a more complex shape as development proceeds allowing the two regions to be discernible (Fig. 2d, e). The ventricular volume increases over development and the total volume of the ventricular system is ~4 nL at early larval stage, ~5.5 nL at late larval stage and ~6.5 nL at 5 dpf. The rhombencephalic ventricular volume increases from ~1.5 to ~3.5 nL from 30 to 54 hpf, but stabilizes at ~3.5 nL from 2 to 5 dpf. At the early larval stage, the isthmic (D/M-to-R) canal is significantly narrower than the anterior (T-to-D/M) canal (cross-sectional area: anterior: 1667 ± 131.3 μm2; n = 18 vs. isthmic: 1159 ± 61.9 μm2; n = 17; p = 0.0016). At the late larval stage, the isthmic (D-to-R) and anterior (T-to-D) canals are smaller than at early larval stage, but are equal to one another (cross-sectional area: anterior: 665.3 ± 118.0 μm2; n = 17 vs. isthmic: 716.2 ± 129.5 μm2; n = 21; p = 0.78). By 5 dpf, the adult connection from the R ventricle to the SC is forming (Fig. 2g–iʹ). These data add detail to the anatomical understanding of the developing ventricular system and ventricular dimensions may also play a role in influencing CSF movement.
CSF moves from anterior to posterior in early larval zebrafish brain ventricles, but partially reverses direction at late larval stages
CSF directional movement is dependent on heartbeat at early larval stages
Brain ventricle cilia are non-motile at 27 hpf and cannot contribute to directional CSF movement
In this study, we expanded understanding of CSF dynamics through quantification of CSF movement. Using the zebrafish model, we focused on the major period of developmental neurogenesis at two stages: early larva [pharyngula stage; 27–30 h post-fertilization (hpf)] and late larva (hatching period; 51–54 hpf). Our study leads to four conclusions: (1) CSF moves preferentially anterior to posterior from the diencephalic (D) (or combined diencephalic/mesencephalic superventricle (D/M)) to rhombencephalic (R) ventricles. (2) Direction of CSF movement between the telencephalic (T) and diencephalic (D) ventricles reverses from early to late larval stages. (3) Directional CSF movement is partially dependent on heartbeat. (4) At early larval stages no motile cilia are present in the brain and therefore are not likely to direct the CSF movement observed. Together these data indicate that locally restricted CSF factors or physical signals associated with fluid movement may contribute to specialization of the early brain.
The primarily unidirectional CSF movement observed between D (or D/M) and R ventricles was associated with a large vmax from D (D/M) to R ventricles. Reverse movement was present, but with about 25 % of the observed forward vmax. This raises two questions: (1) where does the excess volume of CSF entering the R ventricle go and (2) what mechanisms underlie directional CSF movement? First, with respect to CSF “destination”, the brain-CSF barrier is quite tight at early developmental stages , however small molecules (<70 kDa) can pass between the ventricles and the neuroepithelium . Water is also likely selectively transported since aquaporins 1a.1, 3a, 7 are expressed by the neuroepithelium . Therefore some CSF components could pass in or out of the ventricular system. Additionally, we note that R ventricle volume more than doubles over this timeframe of development, perhaps due to CSF influx from more anterior ventricles as movement proceeds from D → R ventricles. Second, with respect to potential mechanisms of directional CSF movement at the stages examined, the choroid plexuses (CP) are not yet mature and there is unlikely to be active secretion from CP precursors. It is not known at what stage the CPs become mature and secrete CSF in zebrafish. Ventricles inflate at approximately 20 hpf  with no evidence of CPs. Henson et al.  claim that the zebrafish CP are mostly mature by 3 dpf, but Garcia-Lecea et al.  report more complete maturation and coalescing of the CP in the R ventricle by 5 dpf. However, even without secretory CP, different regions of the neuroepithelium could produce CSF at different rates to drive directional movement. Alternately, osmolarity differences between the ventricles could drive directional CSF movement, a mechanism that contributes to diffusion gradients in adult brain .
Directional CSF movement from D (D/M) to R ventricles could result in factors from the D (D/M) ventricle entering the R ventricle, but factors unique to the R ventricle may remain largely distinct from those of D (D/M). The stages examined encompass a major period of neurogenesis, when the detailed patterns of the diencephalon, mesencephalon, and rhombencephalon begin to read out as distinct neuronal and glial differentiation programs. Thus differentially localized CSF factors could modulate differentiation of the optic tectum (which is derived from the mesencephalic epithelium) or the cerebellum and medulla (which is derived from the rhombencephalic epithelium). At the stages examined, the spinal canal connects to the D (D/M) ventricle. We observed no converted Kaede protein in the spinal canal (data not shown), indicating that the vast majority of D (D/M) CSF movement is into the R ventricle, consistent with the smaller diameter of the spinal canal connector relative to the isthmic canal.
A surprising contrast to the consistent direction of movement between D (D/M) and R ventricles was the reversal of direction between T and D (D/M) ventricles when early and late larval stages were examined. What would lead to this reversal? One possibility is that at late larval stages, CP cells present along the D ventricle, which arise later but complete differentiation first, have begun to coalesce and are producing CSF, while the CP cells in the R and T ventricles have not yet completed differentiation [50, 52]. This could lead to greater pressure from in the D ventricle from more mature CP cells and promote CSF entry into the T ventricle, where the CP is the last to form . Therefore, posterior-to-anterior CSF movement between D and T ventricles could result from active CSF secretion and greater D ventricular CSF pressure. The strong ciliary movement in D neuroepithelium relative to T neuroepithelium at late larva stage may also contribute to reversal of flow. Another possibility is that the reverse flow could result from gradually increasing pressure in the D ventricle as result of ventricular shape change and the subsequent reduction in D ventricular volume as development proceeds. Whatever the mechanism, the outcome of this change in direction would be that at early larval stages CSF in the D ventricle would contain factors derived from the T, whereas at late larval stages CSF in the T ventricle would contain some components derived from the D. Thus distinct cohorts of factors in T or D ventricles could be present at each stage.
Heartbeat plays a role in CSF dynamics early in brain development, as has also been shown in Xenopus . After tnnt2a loss of function, a mechanism independent of heartbeat, perhaps CSF production, maintains directional CSF movement from T and D/M ventricles, but not for D/M to R ventricular movement, although there is a trend toward less significant directionality. Consistently, acute loss of heartbeat with BDM treatment disrupts all directional CSF movement. However, BDM prevents action of all muscles (and can alter activities of connexins , potassium channels [40, 41] and l-type calcium channels ) and some of these side effects may contribute to the phenotypes observed.
At early larval stages no motile cilia are present in the neuroepithelium and are therefore unlikely to contribute to the CSF movement observed. The non-motile cilia at this time may be able to detect CSF movement, potentially to activate planar cell polarity signaling described by Ohata and colleagues , or to guide migrating neurons or extending axons . At late larval stages, cilia along the D ventricle are highly motile and may indicate differentiating CP, as CP cells have motile cilia . As discussed above, these ciliary movements or CSF production may drive CSF movement from D to T ventricles. Mutants that disrupt the primary cilia present with hydrocephalus , which would preclude clear comparison of CSF movement. However, future studies acutely disrupting cilia are needed to fully address potential roles of cilia in CSF movement.
In summary, this study demonstrates directional CSF movement during early brain development. One outcome of this may be differential concentrations of CSF factors in different ventricles, which may contribute to regional specialization of the brain. Directional CSF movement may also act as a physical signal to modulate neuroepithelial fates. The zebrafish offers an excellent system with which to address these intriguing possibilities.
Embryonic CSF in the zebrafish model moves directionally during the major period of developmental neurogenesis. Specifically, CSF moves preferentially from the diencephalic into the rhombencephalic ventricle. In addition, direction of CSF movement between telencephalic and diencephalic ventricles reverses from early [pharyngula stage; 27–30 h post-fertilization (hpf)] to late (hatching period; 51–54 hpf) larval stages. CSF movement is partially dependent on heartbeat. At early larval stages, the absence of motile cilia indicates that cilia likely do not direct CSF movement. These data suggest that CSF components may be compartmentalized and could contribute to specialization of the early brain and CSF movement may also provide directional mechanical cues.
RMF completed all experimental components and wrote the paper. AH contributed to experimental data collection. JTC and NAS contributed to the design and feasibility of experiments and produced the Kaede protein. HLS contributed to experimental design, interpretation of results, and to writing the paper. All authors read and approved the final manuscript.
The Tg(βact:Arl13b–GFP) transgenic fish line was a kind gift from Brian Ciruna (Sick Kids/University of Toronto). The authors would also like to thank Rotem Gura, Jeremy England, and Mark Bathe (MIT) for thoughtful discussions about fluid dynamics; the Koch Institute Swanson Biotechnology Center, specifically the Hope Babette Tang Histology Facility, for technical support; Wendy Salmon (The Whitehead Institute W. M. Keck Microscopy Facility) and Douglas Richardson (The Harvard Center for Biological Imaging) for valuable insight into appropriate imaging methods; members of the Sive lab for helpful comments; and Olivier Paugois for expert fish husbandry.
The authors declare they have no competing interests.
Availability of data and materials
All supporting data are contained within the manuscript or included in supplemental data.
Consent for publication
This work was supported by NSF IOS-1258087 (H. Sive PI), the HHMI-funded MIT Biology Summer internship program (NAS), and by the Balkin-Markell-Weinberg Postdoctoral Fellowship (RMF).
- 14.Davson H. The physiology of the cerebrospinal fluid. London: Churchill; 1967.Google Scholar
- 23.Westerfield M. The zebrafish book. a guide for the laboratory use of zebrafish (Danio rerio). 3rd ed. Eugene: University of Oregon Press; 1995.Google Scholar
- 27.Muntean BS, Horvat CM, Behler JH, Aboualaiwi WA, Nauli AM, Williams FE, et al. A comparative study of embedded and anesthetized zebrafish in vivo on myocardiac calcium oscillation and heart muscle contraction. Front Pharmacol. 2010;1:139. doi: 10.3389/fphar.2010.00139.CrossRefPubMedPubMedCentralGoogle Scholar
- 35.Sheehan DC, Hrapchak BB. Theory and practice of histotechnology. 2nd ed. Columbus: Battelle Press; 1980.Google Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.