Characterization and classification of zebrafish brain morphology mutants

Abstract

The mechanisms by which the vertebrate brain achieves its three-dimensional structure are clearly complex, requiring the functions of many genes. Using the zebrafish as a model, we have begun to define genes required for brain morphogenesis, including brain ventricle formation, by studying 16 mutants previously identified as having embryonic brain morphology defects. We report the phenotypic characterization of these mutants at several time-points, using brain ventricle dye injection, imaging, and immunohistochemistry with neuronal markers. Most of these mutants display early phenotypes, affecting initial brain shaping, while others show later phenotypes, affecting brain ventricle expansion. In the early phenotype group, we further define four phenotypic classes and corresponding functions required for brain morphogenesis. Although we did not use known genotypes for this classification, basing it solely on phenotypes, many mutants with defects in functionally related genes clustered in a single class. In particular, class 1 mutants show midline separation defects, corresponding to epithelial junction defects; class 2 mutants show reduced brain ventricle size; class 3 mutants show midbrain-hindbrain abnormalities, corresponding to basement membrane defects; and class 4 mutants show absence of ventricle lumen inflation, corresponding to defective ion pumping. Later brain ventricle expansion requires the extracellular matrix, cardiovascular circulation, and transcription/splicing-dependent events. We suggest that these mutants define processes likely to be used during brain morphogenesis throughout the vertebrates.

Introduction

Organ function is dependent not only on proper tissue specification, but also on the three-dimensional organization of those tissues, directed by morphogenetic processes. In vertebrates, the embryonic brain originates from a simple columnar epithelium that forms a tube which will become the brain and spinal cord (Gray and Clemente, 1985). Brain morphogenesis occurs as the anterior neural tube undergoes a series of bends and constrictions to subdivide the brain into the future forebrain, midbrain, and hindbrain that allow it to pack into the skull. The first morphogenetic event is formation of a constriction at the midbrain-hindbrain boundary (Lowery and Sive, 2005). Another key event in brain morphogenesis is the opening of the brain ventricles, cavities inside the brain which contain cerebrospinal fluid (Lowery and Sive, 2005). Correct brain structure is intimately connected to normal brain function, as abnormalities in brain structure during development are correlated with a wide range of neurodevelopmental disorders (Kurokawa et al., 2000; Gilmore et al., 2001; Hardan et al., 2001; Rehn and Rees, 2005; Nopoulos et al., 2007).

Brain morphogenesis requires the function of many genes, but few have been well-characterized. The zebrafish is an ideal model system for analysis of brain morphogenesis, since embryos can be live-imaged at single cell resolution, and mutants can be identified. The early structure of the zebrafish brain is very similar to that of other vertebrates, including chicken, rat and human, further indicating that this is a useful model system (Tropepe and Sive, 2003; Lowery and Sive, 2004; Jo et al., 2005). In order to identify the genetic mechanisms that regulate brain morphogenesis, we have examined 16 zebrafish brain mutants previously suggested to have abnormal embryonic brain morphology, identified in two large-scale mutagenesis screens (Jiang et al., 1996; Schier et al., 1996). Together, Schier et al. and Jiang et al. identified 33 mutants with various embryonic brain morphology defects, 23 of which were described as having specific defects in embryonic brain ventricle morphology, through limited analyses. Through the generous sharing of the zebrafish community, we established 16 of these lines in our lab. In this report, we describe, in more detail, the brain phenotypes of these 16 mutants, including three mutants, nagie oko, snakehead, and whitesnake, on which we have published (Lowery and Sive, 2005; Lowery et al., 2007). Based on these data, we classify these mutants and describe some of the processes required for normal brain morphogenesis.

Materials and Methods

Fish lines and maintenance

Danio rerio fish were raised and bred according to standard methods (Westerfield, 1995). Embryos were kept at 28.5°C and staged as described previously (Kimmel et al., 1995). Times of development are expressed as hours post-fertilization (hpf). All procedures on live animals and embryos were approved by the Massachusetts Institute of Technology Committee on Animal Care.

Lines used were: snk^to273a, atl^tc234b, ott ^ta76b, wis^tr241, vip^tw212e (Jiang et al., 1996), nok^m227, has^m567, ome^m98 (Malicki et al., 1996), zon^m163, ful^m133, lnf^m551, log^m673, esa^m725, sly^m86 (Schier et al., 1996), gup^hi1113B(Amsterdam et al., 2004), nat^tl43c (Trinh and Stainier, 2004), mot^mot_m807 (Guo et al., 1999). As mot and ott are allelic, these two alleles were used interchangeably in our mutant analysis.

Brain ventricle imaging

Brain ventricle imaging was performed as described previously (Lowery and Sive, 2005). Briefly, embryos were anesthetized in 0.1 mg/ml Tricaine (Sigma) dissolved in embryo medium prior to hindbrain ventricle microinjection with 2–10 nl dextran conjugated to Rhodamine (5% in 0.2 mol/l KCl, Sigma), and then embryos were imaged by light and fluorescence microscopy with a Leica dissecting microscope, using a KT Spot Digital Camera (RT KE Diagnostic Instruments). Images were superimposed in Photoshop 6 (Adobe).

Immunohistochemistry

Whole-mount immunostaining was carried out using mouse anti-acetylated alpha tubulin (Sigma, 1:1000), mouse anti-neurofilament RM044 (Zymed #13–0500, 1:50), mouse anti-zn8 (Developmental Studies Hybridoma Bank, 1:20), and rabbit anti-Mpp5 polyclonal antibody (Wei and Malicki, 2002) (1:500). Goat anti-mouse Alexa Fluor 488 and anti-rabbit Alexa Fluor 488 (Molecular Probes, 1:500) were used as secondary antibodies.

For labeling with acetylated tubulin and RM044 antibodies, dechorionated 36 hpf embryos were fixed in 2% tricloroacetic acid for 3 hours at room temperature, washed in PBS, permeabilized in 0.5 % Triton X in PBS and blocked in 0.5 % Triton X, 10% normal goat serum, 0.1% BSA in PBS for 3 hours, prior to incubation in antibody. Brains were flat-mounted in glycerol and imaged with confocal microscopy.

For labeling with zn8 antibody, dechorionated 30 hpf embryos were fixed in 4% paraformaldehyde overnight at 4 degrees, then rinsed in phosphate buffer and permeabilized in 0.5 % Triton X in phosphate buffer. Blocking was done for 4 hours at room temperature in 0.5 % Triton X, 10% normal goat serum, 0.1% BSA in phosphate buffer. Brains were flat-mounted in glycerol and imaged with confocal microscope.

For labeling with anti-Mpp5 antibody and phalloidin-Texas Red, embryos were fixed in 4% paraformaldehyde for 2 hours at room temperature, then rinsed in PBS and dechorionated. Blocking was done for 4 hours at room temperature in 0.5% Triton X, 4% normal goat serum, in phosphate buffer. Phalloidin conjugated to Texas Red (Sigma, 1:1000) was used to label actin filaments.

To block pigmentation, embryos were incubated in 0.2mM 1-phenyl-2-thiourea in embryo media beginning at 22 hpf.

Live confocal imaging

Bodipy ceramide (Fl C5, Molecular Probes) was dissolved in DMSO to a stock concentration of 5 mmol/l. Embryos were soaked in 50 nmol/l bodipy ceramide solution overnight in the dark. The embryos were then washed, dechorionated and placed in wells in 1% agarose for confocal microscopy. Confocal imaging was performed using a Zeiss LSM510 laser-scanning microscope, using standard confocal imaging techniques (Cooper et al., 1999). Confocal images were analyzed using LSM software (Zeiss) and Photoshop 6.0 (Adobe).

In situ hybridization

RNA probes containing digoxigenin-11-UTP were synthesized from linearized plasmid DNA for pax2.1 (Krauss et al., 1991), krox20 (Oxtoby and Jowett, 1993), zic1 (Grinblat et al., 1998), and shh (Krauss et al., 1993) as described (Harland, 1991). Standard methods for hybridization and for single color labeling were used as described elsewhere (Sagerstrom et al., 1996). After staining, embryos were fixed in 4% paraformaldehyde overnight at 4 degrees C and washed in PBS prior to mounting in glycerol and imaging with a Nikon compound microscope.

Results and Discussion

Mutants can be classified based on brain morphology and timing of phenotype onset

We previously suggested that early brain morphogenesis in zebrafish occurs in two phases (Lowery and Sive, 2005). The first phase, occurring between 17 hours post fertilization (hpf) and 24 hpf, includes the shaping of the brain epithelium, as the straight neural tube undergoes regionally-specific bends and opens to form the brain ventricles (Lowery and Sive, 2005). During the second phase, which occurs between 24 and 36 hpf, along with the onset of heartbeat and circulation, both the amount of brain tissue and the volume of the brain ventricles increase substantially (Lowery and Sive, 2005; Mueller and Wullimann, 2005; Bayer and Altman, 2007).

In order to examine the phenotypes of brain morphology mutants, we first analyzed initial brain morphogenesis of each mutant between 17 hpf and 36 hpf using brightfield microscopy (Table 1; data not shown). One criterion for calling each mutant a “brain morphology” mutant is that it makes a healthy neural tube with no visible necrosis through at least 20 hpf (not shown), implying that earlier stages of neural development, including neurulation, are normal. We observed that the abnormal brain phenotypes of 13 mutants are obvious by 20–21 hpf, indicating that initial brain shaping and opening are perturbed. The remaining three mutants (vip, nat, wis) appear to have wild-type brain morphology until 28 hpf, at which point brain morphology defects become apparent, indicating that these mutants are defective in later brain expansion (Table 1).

Table 1.

New Classifications of Brain Morphology Mutants

Brain Phenotype (Class)	Locusallele	Gene	Gene Function	Brain pheno- Type Onseta	Neuronal Abnormalitiesb			Reference
					fbc	hbd	coe
Initial Shaping and Inflation
Midline Separation Defects (Class 1)	nokm227	mpp5	Junctions/ epithelium	20 hpf				(Schier et al., 1996; Wei and Malicki, 2002)
	hasm567	prkci	Junctions/ epithelium	21 hpf				(Schier et al., 1996; Horne-Badovinac et al., 2001)
	omem98	crb2	Junctions/ epithelium	20 hpf				(Schier et al., 1996; Omori and Malicki, 2006)
	zonm163	ND	ND	20 hpf		hb		(Schier et al., 1996)
	atltc234b	ND	ND	21 hpf				(Jiang et al., 1996)
Reduced Ventricle Size (Class 2)	ottta76b	med12	Transcription	20 hpf		hb	co	(Jiang et al., 1996; Wang et al., 2006)
	fulm133	ND	ND	20 hpf		hb	co	(Jiang et al., 1996; Schier et al., 1996)
	lnfm551	ND	ND	20 hpf		hb	co	(Schier et al., 1996)
	logm673	ND	ND	20 hpf		hbg	co	(Schier et al., 1996)
	esam725	ND	ND	20 hpf	fbh			(Schier et al., 1996)
MHB Abnormalities (Class 3)	slym86	lamc1	Extracellular matrix	21 hpf	fb	hb	co	(Jiang et al., 1996; Schier et al., 1996; Parsons et al., 2002)
	guphi1113B	lamb1	Extracellular matrix	21 hpf		hb	co	(Jiang et al., 1996; Schier et al., 1996; Parsons et al., 2002)
Absence of lumen inflation (Class 4)	snkto273a	atp1a1	Na+K+ ATPase	19 hpf				(Jiang et al., 1996; Schier et al., 1996; Lowery and Sive, 2005)
Later Brain Ventricle Expansion
Reduced Ventricle height	viptw212e	ND	NDi	28 hpf	ND	ND	ND	(Jiang et al., 1996)
	nattl43c	fn1	Extracellular matrix	28 hpf	ND	ND	ND	(Jiang et al., 1996; Trinh and Stainier, 2004)
Abnormal morphology	wistr241	sfpq	Splicing/ transcription	28 hpf		hbj	ND	(Schier et al., 1996; Lowery et al., 2007)

^aOther embryonic phenotypes may be visible earlier.

^bIt is noted when obvious neuronal abnormalities are detected by whole mount immunohistochemistry. Cells left blank indicate no abnormalities are detected, although there may be subtle defects which are not apparent.

^cAxon scaffolds in the forebrain are labeled with acetylated tubulin antibody at 36 hpf.

^dReticulospinal neurons in the hindbrain are labeled with RM044 antibody at 36 hpf.

^eCommissural neurons in the hindbrain are labeled with zn8 antibody at 30 hpf.

^fDescribed in (Bingham et al., 2003; Riley et al., 2004).

^gThe reticulospinal neurons of log were usually similar to wild-type, as shown in Fig 5Q, but there were occasional missing neurons at low frequency (10–20%), with phenotypes similar to the other mutants in this group.

^hThe phenotype of esa mutants were quite variable, ranging from severely disrupted axon scaffolds to wild-type-like scaffolds.

ⁱAs the only obvious mutant phenotype other than reduced ventricle height is lack of heartbeat, we speculate that the vip gene is important for some aspect of heart development and function.

^jMany reticulospinal neurons of wis were absent, as shown in Lowery et al 2007.

ND = not determined; MHB = midbrain-hindbrain boundary

Brain morphogenesis occurs on neuroepithelium that has already acquired initial anteroposterior and dorsoventral pattern. We therefore asked whether any early brain phenotypes result from patterning defects. We extended analyses of patterning performed by Schier et al., 1996 and Jiang et al., 1996 using in situ hybridization for the anteroposterior markers krox20 and pax2a, and the dorsoventral patterning markers shh and zic1. No obvious abnormalities are obvious, although we cannot exclude subtle perturbations (data not shown).

All 13 early mutants appear to have brain ventricles of reduced size, as assayed by brightfield microscopy. For these mutants, we extended analyses at two time-points by injecting a fluorescent dye into the brain cavity to highlight the brain ventricular space (Lowery and Sive, 2005). From observation of the brain morphology defects, we determined that the 13 early brain mutants can be divided into four phenotypic classes. We did not use known genotypes for this classification, but based it solely on the specific morphological phenotypes we observed. It was notable, however, that where affected genes are known, mutants with defects in related genes cluster together, showing similar phenotypes. The four classes are:

• Class 1 - Midline separation defects

• Class 2 - Reduced ventricle size

• Class 3 - Midbrain-hindbrain boundary abnormalities

• Class 4 - Absence of lumen inflation

Initial brain shaping – Class 1 - “Midline separation” defects (nok, ome, has, zon, atl)

In Class 1 mutants, dye injection into the ventricles highlights that distinct locations along the brain tube midline appear to remain shut. We have termed this a defect in “midline separation”. After neurulation in wild-type zebrafish, the neural tube is closed but shows a distinct midline (Lowery and Sive, 2004). Subsequently, the tube opens at the midline, leaving the ventricular space centrally (that is filled with fluid) (Fig 1A). But in the five mutants that comprise this class, nok, ome, has, zon, and atl, brain ventricle injections show that midline separation is perturbed, and the hindbrain, in particular, does not open uniformly (Fig 1B–C,E–L). In several of these mutants, opening of the forebrain and midbrain is also perturbed. At 24 hpf, either the brain midline does not open at all (as in nok, Fig 1B, and in forebrain and midbrain of zon, Fig 1H), or there are localized regions where the hindbrain tube does not separate at the midline (arrows, Fig 1C,G,H,I). These phenotypes persist through at least 36 hpf (Fig 1E–F,J–L). Thus, the defect is not simply a lack of lumen inflation, but may indicate that the neuroepithelial cells are abnormal and remain touching in discrete locations throughout the tube. This is visible both by brightfield microscopy after ventricle injection and also at higher resolution using live confocal imaging after soaking the embryos in BODIPY-ceramide (Fig 2B,E).

Figure 1.

Brain ventricle injections of midline separation defects mutants (Class 1, described in the text). Dorsal views of living, anesthetized embryos are shown, anterior to right, at 22 hpf (A–C,G–I) and 32–36 hpf (D–F,J–L) with brightfield microscopy. Ventricles are injected with Rhodamine-dextran. Compared to WT (A, D), the left and right sides of the brain tube do not open uniformly in the midline separation mutants (B–L). In nok (B,E) dye injected into the hindbrain ventricle does not move. In the other mutants, (C,F) ome, (G,J) has, (H,K) zon, (I,L) atl, there are regions where the tube opens separated by places where the sides appear to be touching (arrows). The ventricles of WT are labeled for comparison. F: forebrain ventricle, M: midbrain ventricle, H: hindbrain ventricle.

Figure 2.

Brain morphology comparison of Class 1 and Class 2. (A–C) Hindbrain dorsal views of living, anesthetized embryos are shown, anterior to right, at 22 hpf with brightfield microscopy, after dye injection. While Class 1 has locations along the brain tube where the left and right sides are opposed (B, has mutant), the sides of Class 2 brain tube separate normally (C, lnf mutant). (D–F) Hindbrain horizontal confocal sections after soaking embryos in BODIPY-ceramide to highlight cell outlines. Bright green identified hindbrain ventricle space. Neuroepithelial cells of Class 1 touch at the midline, or are perhaps fused (E, arrows, zon mutant), however fluid separates left and right sides of Class 2 (F, lnf mutant).

As will be discussed further below, it is not clear why midline separation is abnormal in these mutants. One possibility is that cells are abnormally adhesive across the midline. Apicobasal polarity is disrupted in some of these mutants (see below) and this may be required for a loss of adhesion and cell separation at the apical surface. A second, related, possibility is that with loss of apicobasal polarity, cell junctions form across the midline, preventing midline separation. A third possible reason for midline separation defects is that cell shape and/or ability to move is abnormal. In general, high pressure injection into the brain ventricles of these mutants cannot completely separate the neural tube midline, which therefore seems stuck shut in places (data not shown), however, the specific reasons that the midline does not separate will require additional analyses.

Three of the mutants in this group (nok, ome, has) correspond to mutations in genes previously implicated in epithelial polarity and junction formation. The most severe midline separation mutant is nagie oko (nok), which has a mutation in the mpp5 gene encoding a MAGUK protein localized to apical junction complexes (Wei and Malicki, 2002). This mutant has an almost straight brain tube with no or little midline opening at 24 hpf (Fig 1B). When dye is injected into the midline where the hindbrain ventricle normally opens, the dye does not diffuse into other areas of the brain, suggesting that the brain tube is stuck shut (Fig 1B). This defect persists through 36 hpf (Fig 1E). We have previously observed in histological sections that nok mutants have a disorganized epithelium with no continuous midline, although there are small, intermittent regions with a midline present (Lowery and Sive, 2005). We suggest that these obstructions in the midline correspond to locations where the brain tube is stuck shut.

Two mutants, oko meduzy (ome) and heart and soul (has), which have mutations in the crb2a and prkci genes, respectively (Horne-Badovinac et al., 2001; Omori and Malicki, 2006), display almost identical brain phenotypes. Both have relatively normal forebrain and midbrain shaping, but disrupted hindbrain opening, with several small openings instead of one large opening at 24 hpf (Fig 1C,G), persisting through 36 hpf (Fig 1F,J). Unlike the nok mutant (in which regions of the midline do not form), both ome and has appear to form a continuous midline throughout the neuroepithelium. Yet, the midline still does not separate in certain locations throughout the hindbrain.

The Mpp5, Crb2a, and Prkci proteins co-localize at the apical surface of neuroepithelia and control apical junction formation and epithelial apicobasal polarity (Horne-Badovinac et al., 2001; Hsu et al., 2006; Omori and Malicki, 2006), although the mechanisms underlying the brain phenotypes in these mutants have not yet been thoroughly explained. While the adherens junctions of the nok mutant are disrupted, as assayed by localization of adherens junction-associated actin foci (Lowery and Sive, 2005), the junctions in the brain epithelium of both ome and has mutants appear normal upon analysis of adherens junction-associated actin foci, even in the hindbrain regions (Fig 3B,C), consistent with previous reports in other regions of the neuroepithelium (Horne-Badovinac et al., 2001; Omori and Malicki, 2006). However, localization of the apical junction protein, Mpp5 (mutated in nok), is partially disrupted in the ome mutant (Fig 3E), but not in the has mutant (Fig 3F). Whereas Mpp5 localizes solely to the apical junctions in wild-type neuroepithelia (Fig 3D), in the ome mutant, Mpp5 is also mislocalized throughout the cells, as assayed by immunohistochemistry (Fig 3E). Thus, while the adherens junctions of these mutants appear normal according to actin localization, at least one junction component, Mpp5, is not properly localized in the ome mutant.

Figure 3.

Epithelial junction analysis of ome and has mutants. Confocal images of 24 hpf flat-mounted embryos in horizontal section through midbrain and hindbrain. (A–C) Phalloidin-Texas Red labels adherens junction-associated actin of wild type (A), ome mutant (B), has mutant (C). Actin is enriched at the apically localized adherens-junctions, and actin localization is normal in the mutants. (D–F) Mpp5 antibody labeling (green) of wild type (D) ome mutant (E), and has mutant (F) with phalloidin-Texas Red as counterstain (red). Mpp5 is apically-localized in wild-type. In ome, while some Mpp5 localizes normally to the junctions, it is also present throughout the entire neuroepithelium (E). Localization is normal in has (F). Part of the midbrain ventricular surface is not visible in the plane of section, but junctions are normal in those planes. M midbrain, H hindbrain.

How do apical junction complex proteins regulate brain morphogenesis? In the case of the nok mutant, the neural tube midline is defective immediately after neurulation (Lowery and Sive, 2005), suggesting that Mpp5 may be required prior to neural tube closure, for normal epithelial integrity, apicobasal polarity, and formation of a midline corresponding to a plane of cell separation. In the case of ome and has, the neural tube does open in places, suggesting that the midline forms normally, and the adherens junction-associated actin localization appears normal. However, we observed that at least one apical component, Mpp5, is mislocalized in the ome mutant (Fig 3E), and it is possible that other apicobasal polarity components are also disrupted. Thus, apicobasal polarity defects may lead to a neuroepithelium with compromised integrity and inability to separate at the midline. For example, during gut tube formation, has often shows multiple small lumens rather than one large one, due to a lack of apical clustering of adherens junctions (Horne-Badovinac et al., 2001). Further investigation will be required to determine the precise mechanisms that are disrupted in each mutant, and how these regulate midline separation. However, our analysis indicates that the apical junction components, Mpp5, Crb2a, and Prkci, may function together to promote neuroepithelial midline separation, and that Mpp5 apical localization requires function of Crb2a, but not Prkci.

Two additional mutants, zonderzon (zon) and atlantis (atl), corresponding to unknown mutations, also show variable midline separation defects. While the zon midline separation defect can be severe at 24 hpf (Fig 1H) and persist through 36 hpf (Fig 1K), the expressivity is variable. The atl mutant consistently shows a mild midline separation defect, with at least one point of contact at the midline within the hindbrain between 24 and 30 hpf (Fig 1I,L, arrows). This mutant is the least severe of the group, and in some, the brain appears wild type by 36 hpf. The atl mutant is also the only mutant described in this paper in which some homozygous mutants are viable. Further analysis of zon and atl and identification of their corresponding genes will be required to understand the underlying cause of the midline separation phenotype and if they are related to epithelial integrity.

One question we have considered is the relationship between early brain morphology and neuronal development and function. In particular, do any of these midline separation mutants also display neuronal abnormalities? We examined several axon tracts in the early embryo, looking at the early axon scaffolds in the forebrain and midbrain at 36 hpf, which can be visualized with an antibody to acetylated tubulin (Chitnis and Kuwada, 1990), at the reticulospinal neurons in the hindbrain at 36 hpf with an antibody to neurofilament M (Pleasure et al., 1989), and at commissural neurons in the hindbrain at 30 hpf with the zn8 antibody (Trevarrow et al., 1990). Within this mutant class, all neuronal tracts appear normal, although the zon mutant displays occasional defects of the reticulospinal neurons (data not shown). These data suggests that the midline separation defects in these mutants and formation of the early axon scaffold are under independent control.

Initial brain shaping – Class 2 - Reduced ventricle size (lnf, ful, ott, log, esa)

The second class of early mutants, including landfill (lnf), fullbrain (ful), otter (ott), logelei (log), and eraserhead (esa), exhibit reduced brain ventricle size and occasional misshapen midbrain tissue at 24 and 36 hpf, but show no other significant phenotypic abnormalities of brain morphology (Fig 4A–L). After dye injection into the ventricles, the phenotypes of these mutants may appear similar to those in class 1 above, although several comparisons indicate that these classes are distinct. Ventricle injection at high pressure showed that the brain ventricles of all of these Class 2 mutants can open normally, and the midlines of these mutants appear to separate normally (Fig 2C). Similarly, the adherens junction-associated actin foci localize normally in these mutants, and BODIPY-ceramide outlining of cells shows that the two sides of the brain tube do not touch as in Class 1 (data not shown, and Fig 2F).

Figure 4.

Brain ventricle injections and antibody labelings for reduced ventricle size Class 2. Dorsal views of living, anesthetized embryos are shown, anterior to right, at 22 hpf (A–F) and 36 hpf (G–L) with brightfield microscopy. Ventricles are injected with Rhodamine-dextran. The brain ventricles of lnf (B,H), ful (C,I), ott (D,J), log (E,K), and esa (F,L) are all similarly reduced compared to wild-type (A,G). (M–R) Dorsal views of 36 hpf hindbrain flatmounts, anterior is to the top, after labeling with the RMO44 Ab (reticulospinal neurons) shows reduced number of cell bodies and axons in lnf (N), ful (O), ott (P), and log (Q), compared to wild-type (M), although esa (R) appears similar to wild-type (M). (S–X) Dorsal views of 30 hpf hindbrain flatmounts, anterior is to the top, after labeling with the zn8 Ab (hindbrain commissural neurons) shows various levels of reduced commissures in lnf (T), ful (U), ott (V), and log (W), although esa (X) is indistinguishable from wild-type (S). (Y,Z) Lateral views of 36 hpf forebrain and midbrain flatmounts, anterior is to the left, dorsal is to the top, after labeling with acetylated tubulin Ab, which identifies the early axon scaffolds, shows that esa axonal pathfinding is severely disrupted, with the axons having a “feathered” appearance rather than fasciculating normally (Z), compared to wild-type (Y).

The brain phenotypes of lnf, ful, ott, and log are very similar, although at 24 hpf, ott ventricles are generally the most reduced in size, and log ventricles are the least reduced, relative to wild type (Fig 4D,E,A). The esa phenotype, while similar to the others in ventricle size reduction, is more variable (Fig 4F,L) and is occasionally accompanied by additional brain and body phenotypes, which do not occur elsewhere in this class. These abnormalities include gastrulation defects (data not shown), which sometimes result in a twisting of the brain tissue (not shown).

Since the phenotypes of ott, ful, lnf, and log appear similar, complementation crosses were performed between them. Three individual crosses of log and ott heterozygote carriers resulted in 100% wild-type embryos (n=154), demonstrating that log and ott do complement each other genetically. Crosses of all other mutant combinations also showed genetic complementation, indicating that each of these corresponds to a distinct locus. This is in contrast to a note added in the proof of (Schier et al., 1996) indicating that logelei does not complement ott.

We observed that the brain phenotypes of this mutant class resembles that of the mutants motionless (Guo et al., 1999) and kohtalo (Hong et al., 2005), which both have mutations in med12, a subunit of the mediator complex (Hong et al., 2005; Wang et al., 2006). A complementation cross between mot and ott resulted in 14 mutants (21%) and 54 wild-types (79%), suggesting that ott and mot are allelic, although the specific mutation in the ott mutant has not yet been reported. What role does the mediator complex play during development, and what could be responsible for the ott brain morphology phenotype? The mediator complex is a multi-protein complex that regulates transcription by acting as a bridge between DNA-binding transcription factors and RNA polymerase II (Conaway et al., 2005). Several mediator subunits, including med12, possess gene-specific activity (Yoda et al., 2005; Rau et al., 2006; Loncle et al., 2007), and med12 has been shown to interact with beta-catenin and transduce Wnt signaling (Kim et al., 2006). In the zebrafish med12 mutant, the morphogenesis of many organ systems are affected as tissue extension, cell movements, and generation of tissue architecture are disrupted in various tissues (Hong et al., 2005). Interestingly, polymorphisms of the med12 gene in humans are associated with an increased risk for schizophrenia (Philibert et al., 2007), a disorder which is correlated with abnormal brain structure and increased ventricle size (Antonova et al., 2004; Crespo-Facorro et al., 2007).

Zebrafish mutants deficient in med12 display specific neuronal defects, though not all neurons are affected (Guo et al., 1999; Wang et al., 2006). We find that lnf, ful, and ott all show strong defects in the hindbrain axons, with reticulospinal and commissural neurons reduced or missing (Fig 4N–P,T–V), although the early axon scaffolds in the forebrain and midbrain look normal (not shown). The log mutant has reduced commissural neurons (Fig 4W) but all others appear normal (Fig 4Q and not shown). The esa mutant phenotype is different than the others in that the reticulospinal and commissural neurons appear normal (Fig 4R,X), but the axons in the forebrain and midbrain are severely (but variably) affected, having a “feathered” appearance (Fig 4Z).

Initial brain shaping – Class 3 - Midbrain-hindbrain boundary abnormalities (sly, gup)

We defined Class 3 mutants as those which showed midbrain-hindbrain boundary (MHB) abnormalities, in addition to reduced ventricles. Two mutants were placed into this class: sleepy (sly) and grumpy (gup). In these mutants, all brain ventricles are reduced in size compared to wild-type at 22 hpf (Fig 5B,C). Additionally, the MHB appears abnormally shaped, in particular, the MHB fold that occurs by 22 hpf does not form normally (Fig 5B,C), although by 36 hpf, MHB shape and hindbrain ventricle size are partially recovered (Fig 5E,F). The sly and gup loci both encode components of the extracellular matrix proteins, laminin gamma1 and laminin beta1, respectively. These genes have previously been shown to play a number of roles during zebrafish developmental processes including notochord differentiation (Parsons et al., 2002), retina morphogenesis (Biehlmaier et al., 2007), blood vessel formation (Pollard et al., 2006), and retinotectal axon pathfinding (Karlstrom et al., 1996). As laminin in the basement membrane outlines the brain epithelium (data not shown), we hypothesize that loss of laminin in the basement membrane results in the inability to undergo normal brain epithelium shaping processes, particularly at the MHB (Gutzman et al, in press).

Figure 5.

Brain ventricle injections and neuronal antibody labelings for MHB abnormalities Class 3. Dorsal views of living, anesthetized embryos are shown, anterior to right, at 22 hpf (A–C) and 36 hpf (D–F) with brightfield microscopy. Ventricles are injected with Rhodamine-dextran. At 22 hpf, both sly and gup (B,C) show an abnormal midbrain-hindbrain boundary. By 36 hpf, the sly and gup (E,F) boundary region has mostly recovered compared to WT (D), although the forebrain and midbrain ventricles are not as large as in WT. The ventricles of WT are labeled for comparison. F: forebrain ventricle, M: midbrain ventricle, H: hindbrain ventricle. (G–I) Dorsal views of 36 hpf hindbrain flatmounts, anterior is to the top, after labeling with the RMO44 Ab (reticulospinal neurons) shows disruption in axon pathfinding in both sly (H) and gup (I), compared to wild-type (G). (J–L) Dorsal views of 30 hpf hindbrain flatmounts, anterior is to the top, after labeling with the zn8 Ab (hindbrain commissural neurons) shows reduced commissures and disruption in axon pathfinding in both sly (H) and gup (I), compared to wild-type (G). (M–O) Lateral views of 36 hpf forebrain and midbrain flatmounts, anterior is to the left, dorsal is to the top, after labeling with acetylated tubulin Ab, which identifies the early axon scaffolds, shows that sly axonal pathfinding is disrupted (N), although gup (O) looks similar to wild-type (M).

Consistent with the previously reported role of laminin during axon guidance, we observed that the hindbrain reticulospinal neurons are disrupted in both mutants (Fig 5H,I), as are the commissural neurons (Fig 5K,L). Although the early axon scaffolds in the forebrain and midbrain are disorganized in the sly mutant (Fig 5N), they are virtually indistinguishable from wild-type in the gup mutant, suggesting that gup/lamb1 function is not essential for axonogenesis (Fig 5O). It is possible that the mechanisms by which the axon defects arise in the sly and gup mutants are distinct from those regulating brain morphology.

Initial brain shaping – Class 4 – Absence of lumen inflation (snk)

One mutant was in a class by itself, based on its unique phenotype. The snakehead (snk) mutant lacks visible brain ventricles at 22 hpf by brightfield microscopy (Fig 6B), and thus we did not perform ventricle dye injection at this stage. By 30 hpf, however, small ventricles are visible (Fig 6D), and dye injection shows that snk brain morphology is similar to wild-type, with normal hingepoints indicative of normal brain morphogenesis. However, upon injection, all 3 ventricles are much smaller than normal. snk corresponds to a mutation in the atp1a1 gene, encoding a Na⁺K⁺ ATPase alpha subunit (Lowery and Sive, 2005). This pump is likely required for embryonic CSF secretion by creating an ionic gradient across the membrane, that results in water flow into the luminal space (Lowery and Sive, 2005). It is likely that the absence of fluid inside the brain ventricles leads to the snk brain phenotype (Lowery and Sive, 2005). At the stages analyzed, we detected no obvious abnormalities in the neuronal populations examined (Table 1 and data not shown).

Figure 6.

Brightfield microscopy images of absence of lumen inflation mutant Class 4. Dorsal views of living, anesthetized embryos are shown, anterior to right. While the snk mutant at 22 hpf (B) has no visible ventricles and thus ventricles are not injected with dye, by 30 hpf (D), there are small ventricles in which dye can be injected, showing smaller but relatively normal shaping. F: forebrain ventricle, M: midbrain ventricle, H: hindbrain ventricle.

Later brain ventricle expansion mutants

In addition to the 13 early brain shape mutants described above, three brain morphology mutants, viper (vip), natter (nat), and whitesnake (wis), show only later defects in brain morphology. All three display normal brain ventricles at 22 hpf (data not shown), although by 28 hpf, it is apparent that the dorsoventral height of the hindbrain ventricle is reduced (Fig 7B,C,D, bars). All three also lack heartbeat and circulation. It was previously shown that circulation is required for later brain ventricle expansion (Schier et al., 1996; Lowery and Sive, 2005), and thus, it is possible that the brain defects of these mutants are secondary to a lack of circulation. As the vip mutant shows no phenotypes other than reduced brain ventricle height and lack of heartbeat/circulation, and as the brain phenotype is similar to that of the silent heart mutant corresponding to a cardiac-specific troponin (Lowery and Sive, 2005), it is possible that the brain phenotype of this mutant is solely due to lack of circulation.

Figure 7.

Later brain ventricle expansion class. Lateral views of living, anesthetized embryos are shown, anterior to right, at 28 hpf, with brightfield microscopy. The vip mutant shows reduced hindbrain ventricle height (B, red bracket) compared to wild-type (A). The nat mutant shows more severe brain ventricle height reduction (C, red bracket). The wis mutant also shows significantly reduced hindbrain ventricle height (D, red bracket), in addition to reduced pigmentation and other brain morphology abnormalities not shown in this figure.

Conversely, the nat brain defect is more severe than the vip phenotype (compare Fig 7B and C), and it is likely that the brain phenotype of this mutant is due to brain-specific effects as well as lack of circulation. The nat mutant corresponds to the fibronectin gene fn1, a component of the extracellular matrix (ECM), indicating that the ECM is essential for normal brain morphology, consistent with the requirement for laminin function (class 3 mutants, Fig 5) for initial brain shaping. However, the phenotypes of the laminin mutants sly and gup are different from nat, and appear earlier. This indicates that not all ECM components are required at the same time or that the maternal contribution of fibronectin persists longer than that of laminin proteins.

The wis mutant has abnormalities in other aspects of embryonic brain morphology (not shown) in addition to the reduction in brain ventricle height (Fig 7D), and further analysis of this mutant and its corresponding gene, sfpq, is described elsewhere (Lowery et al., 2007). This mutant has defects in neural development, accompanied by increased cell death (Lowery et al., 2007), and these abnormalities may later result in abnormal brain morphology.

Summary

Analysis of brain morphology mutants has allowed us to define several steps and corresponding gene functions required for brain morphogenesis (Fig 8). Processes involved in initial brain shaping include midline separation, brain lumen inflation, midbrain-hindbrain boundary formation, and other mechanisms affecting brain morphology. Midline separation requires epithelial integrity and the apical junction components, mpp5, prcki, and crb2. The transcription regulator med12, which can affect early neuronal development (Guo et al., 1999; Wang et al., 2006), also has an effect on early brain morphology. Midbrain-hindbrain boundary formation requires the extracellular matrix protein laminin, specifically the lamc1 and lamb1 genes which encode the gamma and beta chains, respectively, components of the laminin heterotrimer. Brain lumen inflation requires the Na⁺ K⁺ ATPase, atp1a1. Later brain ventricle expansion requires the extracellular matrix protein fibronectin in order to maintain normal ventricle height, as well as the splicing/transcription factor sfpq, which contributes to normal brain morphology. There are certainly many more genes whose function is required for brain morphogenesis, but were not identified in the mutant set we examined.

Figure 8.

Gene functions required for early brain morphogenesis. Processes involved in initial brain shaping and inflation include midline separation (requiring epithelial integrity/junctions), other mechanisms affecting brain morphology (requiring transcription, among other unknown factors), midbrain-hindbrain boundary formation (requiring extracellular matrix), and brain lumen inflation (requiring Na⁺ K⁺ ATPase activity). Later brain ventricle expansion requires the extracellular matrix in order to maintain normal ventricle height, as well as splicing/transcription, which contributes to normal brain morphology. F:forebrain ventricle, M:midbrain ventricle, H:hindbrain ventricle.

In conclusion, this detailed phenotypic characterization of 16 zebrafish brain mutants has enabled us to determine some of the various processes that are required for early brain morphogenesis. As embryonic brain ventricle morphology is conserved throughout the vertebrates, we suggest that these processes and their underlying mechanisms are also conserved.