Small heat shock proteins are necessary for heart migration and laterality determination in zebrafish

Abstract

Small heat shock proteins (sHsps) regulate cellular functions not only under stress, but also during normal development, when they are expressed in organ-specific patterns. Here we demonstrate that two small heat shock proteins expressed in embryonic zebrafish heart, hspb7 and hspb12, have roles in the development of left-right asymmetry. In zebrafish, laterality is determined by the motility of cilia in Kupffer’s vesicle (KV), where hspb7 is expressed; knockdown of hspb7 causes laterality defects by disrupting the motility of these cilia. In embryos with reduced hspb7, the axonemes of KV cilia have a 9+0 structure, while control embyros have a predominately 9+2 structure. Reduction of either hspb7 or hspb12 alters the expression pattern of genes that propagate the signals that establish left-right asymmetry: the nodal-related gene southpaw (spaw) in the lateral plate mesoderm, and its downstream targets pitx2, lefty1 and lefty2. Partial depletion of hspb7 causes concordant heart, brain and visceral laterality defects, indicating that loss of KV cilia motility leads causes coordinated but randomized laterality. Reducing hspb12 leads to similar alterations in the expression of downstream laterality genes, but at a lower penetrance. Simultaneous reduction of hspb7 and hspb12 randomizes heart, brain and visceral laterality, suggesting that these two genes have partially redundant functions in the establishment of left-right asymmetry. In addition, both hspb7 and hspb12 are expressed in the precardiac mesoderm and in the yolk syncytial layer, which supports the migration and fusion of mesodermal cardiac precursors. In embryos in which the reduction of hspb7 or hspb12 was limited to the yolk, migration defects predominated, suggesting that the yolk expression of these genes rather than heart expression is responsible for the migration defects.

Introduction

The small heat shock proteins (sHsp) are a phylogenetically ancient family that is represented in all kingdoms of life (Sun and MacRae, 2005). The sHsp family is defined by a conserved, approximately 80 amino acid alpha-crystallin protein binding domain, flanked by divergent carboxy- and amino-terminal regions. The sHsps were first recognized as low molecular weight chaperones that prevent aggregation of denatured proteins (Sun and MacRae, 2005). However selected members of the family have diverse functions beyond classically defined chaperone activity. Some sHsps regulate cytoskeleton assembly or associate with myofibrils (Fujita et al., 2004; Golenhofen et al., 2004; Houck and Clark, 2010; Krief et al., 1999; Tucker and Shelden, 2009).

Nine of the thirteen zebrafish sHsps are expressed in temporal- and organ- specific patterns (Elicker and Hutson, 2007). Three sHsps, hspb1 (hsp27), hspb7 and hspb12, are expressed in the ventricle, atrioventricular canal (AVC) and outflow tract of the heart (Marvin et al., 2008) at 24 – 48 hpf (Marvin et al., 2008). In contrast to hspb1, which is ubiquitously expressed until 12 hpf and is upregulated by heat shock, hspb7 and hspb12 are expressed specifically in heart and skeletal muscle and their transcription is unaffected by heat shock (Marvin et al., 2008).

Human HSPB7 (cvHsp) is primarily localized to the cardiovascular system and skeletal muscle, and is upregulated by disease and age (Doran et al., 2007; Doran et al., 2006). Genome-wide association studies link mutations in HSPB7 to dilated cardiomyopathy (Matkovich et al., 2010; Stark et al., 2010). HSPB7 inhibits aggregation of poly-glutamine proteins but lacks chaperone activity (Vos et al., 2010). hspb12 is found only in birds and fish and is most closely related to hspb7; however there is little homology between the two proteins outside the conserved α-crystallin domain (Elicker and Hutson, 2007). Remarkably, although hspb1 is dispensable for normal heart development in zebrafish (Tucker et al., 2009), reduction of hspb1 in Xenopus causes cardia bifida (Brown et al., 2007).

Asymmetry of the internal organs is a universal characteristic of vertebrate development and is critically important for normal embryogenesis. The establishment of left-right asymmetry proceeds in three main phases; the breaking of symmetry relative to the other axes, the initiation and expansion of the asymmetric gene expression cascade, and the interpretation of laterality signals by the organs through differential proliferation, migration and cell polarity (Matsui and Bessho, 2012). The genetic programs that establish left-right asymmetry in vertebrates are largely conserved, but the timing of laterality signals and structure of the organs that generate asymmetry vary between vertebrate classes (Beyer et al., 2012). In many vertebrates, motile cilia in a transient embryonic structure – the ventral node in mouse, Kupffer’s vesicle in zebrafish or the gastrocoel roof plate in Xenopus (Essner et al., 2005; Kramer-Zucker et al., 2005; Nonaka et al., 2002; Schweickert et al., 2007) – generate a net leftward fluid flow that leads to the commitment of each side to a left or right identity (Buceta et al., 2005; Cartwright et al., 2004; Nonaka et al., 2002; Nonaka et al., 2005; Okada et al., 1999). Asymmetric activity of ion channels at cleavage stages specifies lateral fates in some species (Aw et al., 2008; Fukumoto et al., 2005; Kawakami et al., 2005; Levin et al., 2002); this early activity may specify but does not determine laterality except in chick. Mutations that affect cilia polarity (Kishimoto et al., 2008; May-Simera et al., 2010; Serluca et al., 2009; Wang et al., 2011), outgrowth (Bisgrove et al., 2012; Bisgrove et al., 2005; Ravanelli and Klingensmith, 2011) or movement all disrupt laterality determination (Essner et al., 2005; Olbrich et al., 2002).

In all vertebrates examined, the asymmetric activation of second messengers on the left side of the laterality organ triggers expression in the left lateral plate mesoderm of a Nodal-related gene such as southpaw (spaw) in zebrafish (Long et al., 2003; Levin et al., 1995; Lowe et al., 1996). Nodal expression expands rostrally on the left side of the embryo (Wang and Yost, 2008) through a self-enhancement and lateral inhibition (SELI) mechanism (Nakamura et al., 2006). Left-sided Nodal activity leads to expression of the transcription factor pitx2 on the left side of the viscera, brain and heart (Campione et al., 1999; Essner et al., 2000; Saijoh et al., 2000; Shiratori et al., 2001; Yoshioka et al., 1998) and activates lefty1 and lefty2 on the left side of the brain and heart, respectively (Bisgrove et al., 1999; Long et al., 2003; Meno et al., 1997; Meno et al., 1996). In the zebrafish embryo, disruption of cilia motility in Kupffer’s vesicle (KV) alters southpaw expression (Kishimoto et al., 2008; May-Simera et al., 2010; Serluca et al., 2009; Wang et al., 2011; Bisgrove et al., 2012; Bisgrove et al., 2005; Ravanelli and Klingensmith, 2011). If the wave of spaw expression first reaches its rostral-most extent on one side, downstream laterality genes may be expressed on that side only, resolving the asymmetry but randomzing its direction (Wang and Yost, 2008). The misdirection of the global vector of asymmetry leads to the concordant reversal of the internal organs and brain asymmetry (Chen et al., 2001; Bisgrove et al., 2003). However, embryos that express laterality genes on both sides may fail to develop asymmetry, leading to heterotaxy, in which organs have two left or right sides, and fail to migrate correctly. The expression of midline antagonists to Nodal signaling is essential to prevent the contralateral expansion of spaw. Disruption of the Nodal inhibitors charon in the peri-KV region (Lopes et al., 2010), lefty1 in notochord, and lefty2 in heart (Lenhart et al., 2011; Saijoh et al., 2000), or of BMP4 near KV (Chocron et al., 2007) all lead to discordant directions of asymmetry between the organs (Chen et al., 2001; Chin et al., 2000). In the final phase of the establishment of asymmetry, individual organs interpret the molecular laterality cues by executing organ-specific asymmetric behaviors. Asymmetric gene expression leads to asymmetric migration, growth or tension on the organ precursors, resulting in a leftward migration (jog) of the atrial end of the heart (Glickman and Yelon, 2002; Smith et al., 2008), a leftward bend of the gut (Horne-Badovinac et al., 2003; Kurpios et al., 2008), and asymmetric growth of the epithalamus and parapineal organ in the brain (Gamse et al., 2005). At this stage, the concordance of asymmetry depends on the interpretation of laterality signals by each individual organ.

In this paper we explore the early expression and function of hspb7 and hspb12 during the establishment of laterality and the migration of the heart fields. Both genes are expressed in the yolk syncytial layer (YSL) and in the bilateral heart fields, and hspb7 is transiently expressed in Kupffer’s vesicle. We find that hspb7 and hspb12 are required for the establishment of left-right asymmetry of the heart, as well as of the brain and the gut, although they have different effects on the concordance of asymmetry between these organs. We demonstrate that reduced hspb7 inhibits cilia motility in KV by converting motile 9+2 KV cilia to 9+0 cilia.

Methods

Fish lines

Morpholino oligonucleotide (MO) injections were performed in the transgenic Danio rerio line Tg(cmlc2:GFP) (Burns et al., 2005), which expresses green fluorescent protein in heart muscle under the myl7 (cmlc2) promoter. A small number of morpholino experiments were performed in Tg(cmlc2:GFP) crossed to a pet store line (Flanagan-Steet et al., 2005). In situ hybridizations (ISH) for hspb7, hspb12, Nkx2.5 and myl7 were performed in the AB* line on uninjected embryos. ISH for spaw, lefty1, lefty2, pitx2, foxa3 and myl7 was performed on MO injected Tg(cmlc2:GFP) embryos. Injected embryos were raised at 28.5°C in E2 medium (15 mM NaCl, 0.5 mM KCl, 1.0 mM CaCl₂, 1.0 mM MgSO₄, 0.15 mM KH₂PO₄, 0.7 mM NaHCO₃) supplemented with 10 μM gentamycin (Westerfield et al., 2009), and staged by somite count or by equivalent hours post-fertilization (hpf) (Kimmel et al., 1995).

Morpholino oligonucleotide injections

hspb7, hspb12, and control antisense morpholino oligonucleotides (MOs) were obtained from GeneTools (Philomath, OR). Hspb7MO1 targeted the exon1/intron1 boundary TGCATATAGCTTTCCACTCACTTGG. Hspb7MO2 spanned the intron1/exon2 boundary, GCCTGTTCTGTCTGATGAAAAACAT. Hspb12MO1 targeted the exon1/intron1 boundary, CTAGCCACATAAAATCTCGACCTTGG, and hspb12MO2 targeted the exon2/intron2 boundary, GCTCCTCGCTTACTCTTTATGTGGT. Control embryos were injected with standard negative control human β-globin MO (Genetools). Morpholinos were diluted to 0.5 mM in nuclease-free water containing 0.1% phenol red and injected into embryos in a volume of approximately 1 nL. Injections were delivered with MPPI-2 air-powered injectors (Applied Scientific Instrumentation, Eugene, OR). Yolk injections were performed between 3–4.3 hpf. Fluorescein-conjugated control MO (0.125 mM) was included in MO mixtures for yolk injection. Successfully yolk-injected embryos retained the phenol red color for 5 minutes post-injection; an immediate change of the injected fluid to yellow predicted the failure of the MO to be distributed through the yolk.

Quantitative PCR

Quantitative PCR for hspb7 and hspb12 was performed as previously described (Elicker and Hutson, 2007).

In situ hybridization

Following fixation in 4% formaldehyde at the appropriate developmental stage, in situ hybridization (ISH) was performed as previously described (Thisse and Thisse, 2008). spaw, lefty1, lefty2, and pitx2c probes were the kind gifts of Joe Yost; nkx2.5, and myl7 (cmlc2) were the generous gifts of Deborah Yelon; foxa3 probe was obtained from ATCC (Manassas, VA, USA). hspb12 probe was previously described (Marvin et al., 2008). A 1057 bp hspb7 probe was amplified by PCR from IMAGE clone 7233563, and transcribed from a T7 promoter incorporated into the 3′ primer. Constructs containing the probes were linearized and transcribed in the antisense direction with the appropriate prokaryotic RNA polymerase (T7: Promega; SP6; Roche Applied Science) in the presence of DIG labeling mix (Roche Applied Science).

Cilia motility and fluid flow in Kupffer’s vesicle

To analyze motile cilia or fluid flow in Kupffer’s vesicle (KV), live embryos were dechorionated and mounted in 1% low melting point agarose. Beating cilia were imaged between the 8–10 somite stages and fluorescent beads (Polysciences, Inc.) injected into the KV lumen as described (Amack and Yost, 2004; Essner et al, 2005) were imaged between the 6–10 somite stages to visualize fluid flow. Movies were captured using a Zeiss Axiocam high-speed monochromatic camera on a Zeiss AxioImager MI microscope with a 63x water-dipping objective. These movies were analyzed and edited using AxioVision (Zeiss), ImageJ (NIH) and Quicktime (Apple) software.

Transmission Electron Microscopy

Embryos were fixed at 10 somites in 2.5% glutaraldehyde and 2% formaldehyde in 0.1M sodium cacodylate solution at 4°C, overnight and washed in 0.1M sodium cacodylate. Secondary fix was 3 hours in 1% OsO₄. Embryos were washed in deionized water, dehydrated through an ethanol series, followed by a propylene oxide/resin series. Resin-embedded embryos were cured at 60°C overnight. Thin sections through KV were stained for 7 min in 2% uranyl acetate in a 50:50 mix of ethanol and methanol, followed by 2% lead citrate for 2 min (Bozzola and Russell, 1992). Sections were mounted on copper grids and imaged on a Philips CM 10 transmission electron microscope.

Rescue constructs for hspb7 and hspb12

Three Danio rerio hspb7 expression constructs were created from an EST (IMAGE clone 7623563; GenBank accession number: BC083373). Tol2-CMV-hspb7: The coding region of hspb7 was amplified by PCR with nested primers containing a consensus Kozak sequence and attB1 and attB2 sites and cloned into a Tol2 plasmid using the Gateway system (Invitrogen). CS2+hspb7: The coding region of hspb7 was amplified using primers engineered to create a BamH1 site upstream of the ATG and an Xba1 site at the stop codon, and cloned into corresponding sites on CS2+. CS2+hspb7EST: The hspb7 EST was cut with Not1 and blunted, then EcoR1. The 1150 bp fragment containing the entire cDNA was cloned into CS2+ cut with EcoR1 and Stu1. CS2+Hu-hspb7-myc: Frt-TO-myc-Huhspb7 (Vos et al, 2009; kind gift of Dr. Harm Kampigna) was cut with Dra1 and ligated to CS2+ cut with Stu1. CS2+hspb12: the coding region of hspb12 was amplified from an EST (Genbank accession number: CK689325) with primers conferring a BamH1 and Xba1 site at the 5′ and 3′ ends respectively, and ligated to CS2+ cut with the same enzymes. In vitro transcripts were generated using mMessage Machine (Ambion) from linearized plasmids using SP6, and purified with MegaClear (Ambion).

Results

Expression of hspb7 and hspb12 in the early zebrafish embryo

Our previous survey of the expression of small heat shock proteins demonstrated that three zebrafish shsps are expressed in heart at 24–48 hours post fertilization (hpf) (Marvin et al., 2008). Here, we extend the description of hspb7 and hspb12 expression patterns to earlier stages. Transcripts for hspb7 and hspb12 were not detected at 2 or 4 hpf by in situ hybridization (ISH) or quantitative RT-PCR. hspb7 and hspb12 expression was first detected by ISH at 5.7 hpf in the yolk syncytial layer (YSL; Figure 1A, E). Transient hspb7 expression was observed in the developing KV (arrows in Figure 1B, C) at the one-somite stage, but this KV expression was undetectable by the 4-somite stage (Figure 1D). In contrast, hspb12 was not observed in KV (Figure 1F, G). Both hspb7 and hspb12 were expressed in the YSL from 5.7 hpf through the 4-somite stage (Figure 1D, H), but disappeared by 10 somites. hspb7 expression began in the bilateral heart fields at the 12 somite stage (Figure 1I) and remained concentrated in the precardiac mesoderm during its migration toward the midline (Figure 1J). hspb12 expression commenced in the bilateral cardiac precursors several hours later, at the 18 somite stage (Figure 1K); at this stage the hspb7 and hspb12 expression patterns resembled that of nkx-2.5, which is localized to the ventricular precursors (Figure 1L). After fusion of the bilateral cardiac fields at 21–22 somites, expression of both hspb7 (Figure 1M) and hspb12 (Figure 1N) continued. The hspb12 domain appeared larger on the left in some embryos (Figure 1N); this asymmetry was transient. Compared to the expression domain of myl7 at 22 somites (Figure 1O), hspb7 and hspb12 were localized to a subset of the heart field. Their expression was maintained in ventricle until at least 48 hpf, and intensified in the atrioventricular canal and outflow tract between 24–48 hpf (Marvin et al., 2008).

Figure 1. Expression patterns of hspb7 and hspb12 during the first 22 hours of development.

A. hspb7 was first detected by in situ hybridization at 5.7 hpf in the yolk syncitial layer (YSL) B. hspb7 expression continued in the YSL and was transiently expressed in the forming Kupffer’s vesicle (KV) at the tailbud-1 somite stage. C. Posterior view of hspb7 expression in KV (arrow) at tailbud stage (dorsal to the left). D. hspb7 disappeared from KV (arrow) by the 4 somite stage but remained in the YSL. E. hspb12 was similarly expressed in the YSL at 5.7 hpf. F. hspb12 was expressed in the YSL at tailbud stage, but was not detected within KV. G. Posterior view showing that hspb12 hybridization was not visible in KV. The surface of the YSL is depressed beneath the tailbud. This causes an edge-on view of the YSL surface, but the apparent staining was not visible at other angles. H. hspb12 was absent from KV in 4 somite stage embryos, but YSL expression continued until approximately the 10 somite stage. I. hspb7 was expressed in the bilateral heart fields beginning at the 12 somite stage. J. hspb7 in the heart field and somites at the 18 somite stage. K. hspb12 expression began in the bilateral cardiac precursors at the 18 somite stage. L. Nkx-2.5 is expressed in ventricular precursors and corresponds to the expression sites of hspb7 and hspb12 at 18 somites. M. hspb7 and N. hspb12 were expressed in the fused heart fields as the leftward jog of the primitive heart tube begins at 21 somites. O. Expression of myl7 (cmlc2) at 21–22 somites overlapped hspb7 and hspb12 hybridization.

Global depletion of hspb7 alters heart development

To investigate the roles of hspb7 during embryonic development, we designed a splice-blocking morpholino (hspb7MO1) against the exon 1 – intron 1 boundary of hspb7 to disrupt processing and translation of hspb7 transcripts. At 26 hpf, the direction of the heart jog was randomized in embryos injected with 0.5 pmole of hspb7MO1 (Figure 2A), while the hearts of control MO injected embryos consistently jogged to the left (Figure 2A,B). In situ hybridization for myl7 demonstrated that the heart tubes that jogged to the right in hspb7MO1-injected embryos were abnormally shaped. The atrial ends were more rounded, and the heart was shorter and wider compared to control MO injected embryos (Figure 2C). At 24 hpf, control MO-injected embryos had a normal axis and straight trunk (Figure 2D), but hspb7 morphants developed a downwardly curved body axis (Figure 2E), a phenotype characteristic of cilia defects (Kishimoto et al., 2008; Serluca et al., 2009). Despite the axis curvature, the length of hspb7 morphants (n=6) from the otic vesicle to the tip of the tail was 101 (+/− 2%) of control MO embryos (n=5), suggesting that convergent extension defects did not contribute to the phenotype. Injection of 0.5 pmole of hspb7MO1 reduced hspb7 mRNA to 25–61% of control levels at 24 and 26 hpf, respectively (Supplementary Figure 1).

Figure 2. Heart jog direction was randomized by hspb7 knockdown.

A. Two splice-blocking morpholinos that target hspb7 both significantly increased right and midline heart jogs at the expense of leftward heart jogs (χ² test, p < 0.001). B, C. In situ hybridization of myl7 (cmlc2) in embryos injected with hspb7MO1. B. Left jog, C. Right jog. D. Live embryo at 26 hpf, injected with control MO. E. Live embryo at 26 hpf, injected with hspb7MO1, showing characteristic downward curl of the body. F. The combination of hspb7MO1 (0.25 pmole) and hspb7MO2 (0.17 pmole) prevented heart field fusion when targeted to the whole embryo at 1–4 cell stages. G–I. In situ hybridization for myl7 (cmlc2). G. Embryo injected with control MO at 1 cell stage had normal leftward heart jogs; the anterior boundary of the heart is midway across the eye at 26 hpf. H. Midline heart in embryo injected with hspb7MO1+MO2 at 1 cell stage, showing slowed anterior migration, with the anterior boundary of the heart at the posterior end of the eye. I. hspb7MO1+MO2 injected embryo, with cardia bifida and delayed anterior migration. J. hspb7MO1+MO2 injected embryo, with normal leftward jog but delayed migration and wide heart. K. Yolk injection between 3–4 hpf of control MO produced a small number of midline hearts; hspb7MO1 (0.5 pmole) induced midline hearts; hspb7MO1 (0.25 pmole) and hspb7MO2 (0.25 pmole) induced cardia bifida and midline hearts.

To test the specificity of hspb7MO1 phenotypes, we designed a second MO (hspb7MO2) to target the intron 1 – exon 2 splice acceptor site. At a dose of 0.25 pmole, hspb7MO2 also disrupted asymmetry (Figure 2A) and induced downward-curling tails at 24 hpf. Higher doses of hspb7MO2 were generally lethal; the surviving embryos’ hearts failed to jog and remained at the midline (data not shown). A third morpholino, which blocks hspb7 translation (Musso et al., in press) caused similar degrees of midline and right heart jogs (data not shown). Thus, three morpholinos against hspb7 produced similar laterality phenotypes.

Greater degrees of gene knockdown can be obtained by combining two MOs directed at the same gene (Draper et al., 2001). We therefore combined a reduced dose of hspb7MO1 with hspb7MO2 (0.25 pmol each) and injected this combination into embryos at the 1–4 cell stage. The combination of the two MOs reduced hspb7 transcripts to 12% of normal levels at 24 hpf (Supplementary Figure 1). This stronger knockdown led to a further reduction in the number of embryos with normal leftward-jogging hearts than was induced by either of the hspb7 MOs alone (Figure 2F,G). The midline hearts resulting from the combined hspb7 MOs were short and wide (Figure 2H). Even in embryos with leftward-jogging hearts, there appeared to be a defect in heart migration; the leading edge of the cardiac primordia just reached the posterior extent of the eye, while the leading edge of the heart field in control embryos lay at the anterior-posterior midpoint of the eye (Figure 2I). Moreover, the combined hspb7 MOs caused cardia bifida in some embryos, in which the bilateral heart fields fail to migrate to the midline and fuse (Figure 2J). Cardia bifida was never observed in control MO or single hspb7 MO-injected embryos. These data suggest that hspb7 supports cardiac mesoderm migration, in addition to its function in establishing left-right asymmetry of the heart. However, asymmetry is more sensitive to the dose of hspb7 than is heart precursor migration.

hspb7 depletion in the yolk disrupted cardiac movements

The fusion of cardiac primordia depends upon the integrity of the YSL extracellular matrix; mutations and morpholinos that disrupt either the YSL or definitive endoderm cause cardia bifida (Alexander et al., 1999; Arrington and Yost, 2009; Dickmeis et al., 2001; Kikuchi et al., 2001; Sakaguchi et al., 2006). Both the YSL and the precardiac mesoderm express hspb7, so depletion of hspb7 in either tissue could be responsible for the cardiac migration defect. To distinguish between these possibilities, we took advantage of the finding that cytoplasmic bridges between the yolk and most embryonic tissues close by the 256-cell stage (Kimmel and Law, 1985). MO injected to the yolk prior to sphere stage (4hpf) is taken up by the yolk syncyctial layer, and at a lower efficiency by the dorsal forerunner cells (DFC) that give rise to KV (Amack and Yost, 2004; Arrington and Yost, 2007; Essner et al., 2005). Therefore injecting MO into the yolk cell after the 256-cell stage (~2.5 hpf) restricts the effect of the MO to the YSL and KV. To identify embryos in which yolk-targeted MO was successfully distributed throughout the YSL, we included 0.1 pmol of fluorescein-conjugated control MO as a tracer. Over 60% of injected embryos showed fluorescent MO distributed throughout the YSL exclusively at 50% epiboly; the remainder of the embryos had the MO trapped in yolk granules. Distribution of the MO to the embryo itself was never observed, confirming previous studies on yolk injection (Amack and Yost, 2004; Arrington and Yost, 2009). Injection of hspb7MO1 (0.5 pmol) into the yolk cell at 3–4 hpf increased the frequency of abnormal heart jogs compared to control MO (Figure 2K). A strong yolk-targeted knockdown of hspb7 by the combined hspb7MO1 and hspb7MO2 (0.25 pmole each) induced cardia bifida, midline hearts, and altered anterior migration. Cardia bifida precludes analysis of laterality; the direction of jogging in bifid hearts cannot be determined. The distribution of heart phenotypes differs signficantly (χ² = 58.3, p < 0.001) between hspb7MO1+MO2 injected to the yolk compared to injection of this MO pair at the 1–4 cell stage (Figure 2F,K), suggesting that restricting the injection to the yolk causes a more severe migration defect by concentrating the morpholino there.

hspb12 is necessary for normal cardiac laterality

Depletion of hspb12 also caused cardiac left-right asymmetry defects (Figure 3A). A morpholino that targets the exon1 – intron 1 boundary (hspb12MO1) reduced expression levels of the endogenous transcript to 12–21% of control levels at 26 hpf (Supplementary Figure 1). hspb12MO1 (0.5 pmol) caused an increase in midline and right-sided hearts, compared to control MO (Figure 3A). In embryos injected with a second MO that targets the hspb12 exon 2 – intron 2 boundary (hspb12MO2), the frequency of midline and right hearts also differed significantly from control MO (Figure 3A). A dose of 0.125 pmol hspb12MO2 was optimal; higher doses had toxic effects. A combination of the two hspb12 MOs resulted in incomplete gastrulation even at low combined doses, so the effects of a stronger hspb12 knockdown could not be assessed.

Figure 3. Heart jog direction was randomized by hspb12 knockdown.

A. Two morpholinos that target hspb12 significantly increased right and midline heart jogs at the expense of leftward jogs compared to control MO (χ² test; p < 0.001 for hspb12MO1 and p < 0.05 for hspb12MO2). B. Yolk injections of hspb12MO1 increased midline hearts but not rightward jogs. Combinations of both hspb12 MOs were lethal. C. Combined hspb7MO1 and hspb12MO1 (0.25 pmol individually or in combination) injected at the 1-cell stage increased midline hearts and rightward jogs significantly compared to control (χ² test; all p < 0.001). Combined hspb7MO1 + hspb12MO1 (0.25pmol each) gave significantly more abnormal jogs than 0.25 pmol hspb7MO1 alone (χ² test; p < 0.05).

hspb12 and heart migration

YSL-targeted injections of hspb12MO1 increased the proportion of midline hearts compared to control MO (Figure 3B). There was no significant difference in the distribution of heart laterality for hspb12MO1 injections into the yolk versus the whole embryo (Figure 3A, B), although the number of right heart jogs appeared to be lower for the yolk injections. hspb12MO1 did not induce cardia bifida in the Tg(cmlc:GFP) AB background. However, occasional cardia bifida was observed in yolk injections in a Tg(cmlc:GFP) in another wild type background, which is more sensitive to hspb12 knockdown (Supplementary Figure 2).

Functional overlap between hpsb7 and hspb12

Of all the sHsps, hspb12 is most similar to hspb7 in the conserved α-crystallin domain. Their expression patterns are similar, although hspb12 was not detected in KV, suggesting that there could be functional redundancy between the two genes. To investigate this possibility, hspb7MO1 and hspb12MO1 were co-injected into 1-cell embryos. At doses for which each MO individually had a submaximal effect, the combination caused an increase in laterality defects (Figure 3C). The combination of 0.25 pmol Hspb7MO1 and 0.25 pmol Hspb12MO1 increased the frequency of abnormal laterality to so that it did not statistically differ from that caused by 0.5 pmole of Hspb7MO1 alone (Figure 2A), suggesting that there is functional overlap between hspb7 and hspb12.

Reduced hspb7 expression affects laterality genes

The first asymmetrically expressed gene in zebrafish is southpaw (spaw) in the left lateral plate mesoderm. This nodal-related gene induces the transcription factor pitx2 in the left lateral plate and brain, as well as the Nodal inhibitors lefty1 and lefty2 on the left side of the brain and heart, respectively (Long et al., 2003). Embryos injected with control MO expressed spaw in the left lateral plate mesoderm (Figure 4A, B). However, hspb7MO1 induced primarily bilateral or left sided spaw expression at 14–18 somite stages (Figure 4A, C); few embryos had right-sided expresssion (Figure 4A).

Figure 4. hspb7 was necessary for the expression of laterality genes in the left lateral plate mesoderm (LPM).

A. Embryos injected with 0.5 pmol control MO had primarily left-sided expression of spaw in the lateral plate mesoderm. In contrast, hspb7MO1 induced bilateral spaw expression (χ² test; p < 0.001). B–E. 14–18 somite stage embryos. B. Embryos injected with control MO had normal spaw and lefty1 expression in left LPM and notochord, respectively. In the tail, bilateral peri-KV spaw domains were present (arrow). C. hspb7MO1-injected embryos expressed spaw bilaterally. lefty1 in the notochord and peri-KV expression of spaw were not disturbed. D, E. hspb7MO1-injected embryo with bilateral spaw with D. left side leading leading and E. right side leading. F. hspb7MO1 randomized pitx2 expression (χ² test; p < 0.001), with largely left or right sided expression and a small fraction of bilateral expression. G. lefty2 expression in the left heart field was randomized or made bilateral by hspb7MO1 injection (χ² test; p < 0.001). H. embryos injected with control MO express lefty1 and lefty2 on the left side of the heart and brain at 22–24 somite stages. I. hspb7MO1 injected embryos expressing lefty1 and lefty2 on the right. J. lefty1 expression in brain was randomized by hspb7MO1 injection (χ² test; p < 0.001).

There are two mechanisms that can result in bilateral expression of spaw. When KV cilia function is disrupted or signals from KV fail to transmit to the lateral plate, spaw is often expressed bilaterally (Bisgrove et al., 2003; Bisgrove et al., 2005; Francescatto et al., 2010; Hashimoto et al., 2004; Ravanelli and Klingensmith, 2011; Schottenfeld et al., 2007). A second mechanism would be the failure of the midline barrier to block induction of ectopic spaw in the right lateral plate by the normal expression of spaw on the left. This midline barrier consists in part of KV-independent expression of lefty1 in notochord (Bisgrove et al., 1999; Meno et al., 1998). Embryos injected with either control MO or hspb7MO1 showed normal expression of lefty1 in notochord and around KV (Figure 4B, C), indicating that the midline barrier remained intact.

Many hspb7 morphant embryos expressed spaw bilaterally, but the anterior expansion of spaw expression was not equal on the two sides (Figure 4D, E), such that either the left or the right lateral plate domain was leading. In embryos that expressed spaw bilaterally, the lagging domain was typically greater than 75% of the length of the leading side. Of the embryos that expressed spaw bilaterally 16.4% had the left leading, 25.4% were equal, 10.4% had the right domain leading. Such unequal bilateral domains have previously been reported in pkd2 mutants and casanova, lefty1, and charon morphants (Schottenfeld et al., 2007; Wang and Yost, 2008).

Because hspb7 knockdown caused bilateral spaw expression, we would expect a disruption of lateralized gene expression downstream of spaw. pitx2 is a target of lateral plate nodal signaling and, consistent with this prediction, the expression of pitx2 was randomized in hspb7MO1-injected embryos, with equal numbers expressing pitx2 on the left or right, and a small fraction showing bilateral expression (Figure 4F). The expression of lefty2 in the heart field and lefty1 in the diencephalon also depend upon lateral plate spaw expression (Long et al., 2003). A much smaller fraction of embryos injected with hspb7MO1 had bilateral expression of lefty2 (Figure 4G) compared to the number of embryos expressing spaw bilaterally; lefty2 was primarily expressed on only the left or right side (Figure 4H, I). Similarly, expression of lefty1 in the brain was randomized at approximately the same rate as lefty2 in the heart field in hspb7MO1 morphant embryos (Figure 4J), and there was far less bilateral expression than for spaw. This frequency was consistent with the heart jog data (Figure 2A), suggesting that the bilateral expression of spaw resolves to unilateral lefty1 and lefty2 expression, and results in heart jogs with randomized direction, instead of failure to jog (Figure 2A).

Reduced hspb12 affects laterality genes

Embryos injected with hspb12MO1 showed an increase in bilateral or right-sided expression of spaw at the 14–18 somite stage (Figure 5A). The degree of this response was dependent on genetic background. Spaw expression in Tg(cmlc:GFP) with an AB background (Burns et al., 2005) and had a small but statistically significant increase in bilateral expression of spaw in response to hspb12 knockdown (data not shown). However, fish from this line outbred to another wild-type strain (Flanagan-Steet et al., 2005) had a more robust response (Figure 5A). Embryos injected with hspb12MO1 expressed lefty1 normally in the notochord, suggesting that the midline barrier remained intact in these embryos (Figure 5B, C).

Figure 5. hspb12 knockdown randomized spaw, pitx2, lefty2 and lefty1.

A. 0.5 pmol hspb12 MO1 induced bilateral spaw expression compared to control MO (χ² test; p < 0.001). B, C 14–18 somite embryos. B. Control MO injected embryo with left sided spaw and midline lefty1 expression. C. hspb12MO1-injected embryo with normal spaw and lefty1 expression. D. Hspb12MO1 randomizes pitx2 expression (χ² test; p < 0.001). E. hspb12 MO1 injection does not induce ectopic lefty2, but increases the number of embryos lacking lefty2 expression (χ² test; p < 0.001). F–H. lefty1 and lefty2 in situ hybridization at 22–24 somites. F. control MO with normal lefty1 and lefty2 in left brain and heart field respectively. G. in this embryo, Hspb12MO1 suppressed normal lefty1 and lefty2. H. in another embryo, Hspb12MO1 induced bilateral lefty1 and lefty2. lefty1 and lefty2 were never observed exclusively on the right. I. Hspb12MO1 injection caused bilateral or absent lefty1 expression in the brain (χ² test; p < 0.001). J. Injection of 0.25 pmol of hspb7MO1 or hspb12MO1 individually did not affect left-sided spaw expression in the LPM, but injection of both MO together induced bilateral and right-sided spaw. K. A dose of 0.25 pmol of hspb12MO1 or hspb7MO1 individually perturbed expression of lefty2 in heart, while the combination of 0.25 pmol each of hspb7MO1 and hspb12MO1 induced an even greater degree of abnormal expression. L. Similarly, 0.25 pmol of hspb7MO1 or hspb12 MO1 induce a similar alteration in the expression of lefty1 in the brain, while the combination induces further abnormal expression.

Despite the slight effect on spaw expression in the Tg(cmlc:GFP) line with an AB background, knockdown of hspb12 significantly altered the distribution of the downstream laterality genes pitx2, lefty1 and lefty2 in this background. Hspb12MO1 (Figure 5D) induced bilateral, right or absent pitx2 expression in approximately 38% of the embryos. In many embryos pitx2 expression in the lateral plate mesoderm was visible but reduced in intensity compared to either control MO embryos or hspb7MO1 injected embryos (data not shown). Injection of hspb12MO1 also perturbed expression of lefty2 (Figure 5E) and lefty1 (Figure 5I); however, the pattern of changes in gene expression differed notably from those of hspb7 morphants (Figure 4G, J). The majority of embryos expressed lefty1 and lefty2 on the left (Figure 5F), but these genes were frequently absent (Figure 5G), and they were never expressed on the right unless they were mirrored on the left (Figure 5H). In summary, hspb12 knockdown affected pitx2, lefty1 and lefty2 in both strains, but the effect of hspb12MO1 on spaw varied by strain.

Redundance of hspb7 and hspb12

At doses that individually had little effect on spaw, the combined knockdown of hspb7 and hspb12 caused an increase in bilateral and right-sided spaw (Figure 5J). This observation is consistent with the increased cardiac laterality defects in combined hspb7-hspb12 morphants (Figure 3C). Although reduced hspb12 alone did not increase bilateral spaw (Figure 5A), it potentiated the effect of a sub-optimal dose of hspb7MO1 (Figure 5J). The combined hspb7 and hspb12 knockdown also increased the frequencies of embryos with right-sided or absent lefty2 and lefty1 expression (Figure 5K, L), compared to knockdown of either gene alone. Suppression of right-sided lefty1 and lefty2 was characteristic of hspb12 but not hspb7 knockdown.

Concordance of laterality defects with reduced hspb7

The intact midline expression of lefty1 led us to investigate whether the laterality of brain, heart and gut are globally coordinated in hspb7 and hspb12 knockdown embryos. If the midline barrier function were disrupted, organs would be randomly oriented relative to each other (Bisgrove et al., 2000; Chin et al., 2000), and few of the embryos would show agreement between the laterality of any two organs. Discordant asymmetry could also occur if there were a disruption in the interpretation of the Nodal laterality signal by the downstream organs (Hirokawa et al., 2009; Ramsdell, 2005; Raya and Belmonte, 2006).

Expression of lefty1 and lefty2 in control embryos was concordant, with both genes localized on the left in 96.2% of the cases (Table 1). In embryos injected with hspb7MO1, expression of lefty1 in brain and lefty2 in heart at 22–24 somites were still largely concordant; lefty1 and lefty2 were both localized to the left or the right in 77.0% of the morphants (Table 1). The remaining embryos had either bilateral expression of lefty1 and lefty2, indicating an ongoing failure to specify a global vector of asymmetry, or they had discordant expression of lefty1 and lefty2. In these cases one lefty gene was expressed normally while the other was abnormal (Table 1).

Table 1.

Brain and Heart laterality at 22–24 somites in hspb7, hspb12 and double morphant embryos. lefty2 is normally expressed in the left heart field and lefty1 in the left epithalamic precursor at this stage. Concordant expression comprises both lefty genes on the same side of the embyro.

lefty1 brain, lefty2 heart	n=	L, L (%)	B, B (%)	R, R (%)	A, A (%)	Discordant (%)
control	26	96.2	3.8	0.0	0.0	0.0
hspb7 MO1	26	46.2	11.5	30.8	0.0	11.5
control	39	76.9	5.1	10.3	0.0	7.7
hspb12 MO1	28	53.6	10.7	0.0	25.0	10.7
uninjected	49	100.0	0.0	0.0	0.0	0.0
hspb7 MO1 + hspb12 MO1	39	51.3	0.0	0.0	33.3	15.4

To investigate concordance between heart and gut turning, we compared the expression of foxa3 in liver and gut to myl7 expression in heart at 48 hpf. In 95% of control MO-injected embryos, myl7 expression showed that the atrium was on the left, and that the ventricle had looped to the right, while foxa3 expression indicated the liver was on the left and the gut had a leftward bend. (Figure 6A, A′, Table 2). In 82.6% of embryos injected with hspb7MO1, the asymmetry of the gut was concordant with that of heart (Figure 6B, B′, Table 2). The discordant arrangements (17.4%) were due to hearts that failed to deviate from the midline. We did not observe any duplicated or midline livers or digestive tracts; in embryos with midline hearts, the gut was bent either to the left or the right (Table 2).

Figure 6. Concordance of asymmetry between heart and gut in hspb7 and hspb12 knockdown embryos.

A–D. Dorsal view, rostral toward bottom, left side as indicated in A: Expression of foxa3 is strong in liver (Lv) and weaker in gut and pancreatic bud (P) at 48 hpf. A′-D′. myl7 (cmlc2) is strongly expressed in ventricle (V) and more weakly in atrium (At) at 48 hpf. Anterior view, left side as indicated in A. A, A′. Control MO-injected embryo demonstrated the normal position of atrium to the left of the embryo’s midline and of the ventricle on the right (dextral loop). foxa3 is concentrated in the normal, left-sided liver and weaker in the pancreatic bud in control MO-injected embryo. B, B′. hspb7 morphant with right-sided liver and ventricle. C, C′. hspb12MO1-injected embryo with bilateral liver but normal position of the atrium and ventricle. D, D′. hspb12MO1-injected embryo with left-sided liver and a heart that failed to jog or loop (midline heart).

Table 2.

Heart and gut laterality at 48 hpf in hspb7 and hspb12 morphant embryos. The atrium jogs to the left and the ventricle undergoes a dextral loop by 48 hpf. The liver develops on the left and strongly expresses foxa3; the intestine expresses foxa3 at a lower intensity. Equal numbers of hspb7 morphants with midline hearts had foxa3 liver expression on the left or the right (8.7% each); these comprise the discordant embryos. hspb12 morphants with discordant phenotypes were also largely embryos with midline hearts and left livers (16.7%) or right livers (9.5%) but also included right sided hearts with left livers (4.8%), right heart without liver (2.4%) and left heart with bilateral liver (2.4%). Discordant control embryos had midline hearts with left or right livers.

Heart jog, liver side	n=	L, L (%)	M, B (%)	R, R (%)	A, A (%)	Discordant (%)
control MO	20	95.0	0	0	0	5.0
hspb7 MO1	23	52.2	0	30.4	0	17.4
control MO	33	84.8	0	0	0	15.2
hsbp12 MO1	42	57.1	2.4	2.4	0	38.1

hspb7 morphant embryos exhibited a high degree of concordance between both the laterality of the heart and brain, and between the heart and gut. While a fraction of embryos failed to specify a direction of laterality, the expression of lefty1 and lefty2, as well as the direction of heart jogging and gut turning were in agreement, indicating that laterality was not randomly determined for each organ, as would be the case if the midline barrier had been disrupted.

Concordance of laterality defects with reduced hspb12

In hspb12MO1 morphants, only 10.7% of the embryos had discordant brain and heart asymmetry; an equal number had bilaterally symmetric expression of lefty1 and lefty2 at the 22–24 somite stage (Table 1). The frequency of discordance and bilateral lefty1 and lefty2 expression was similar to that in hspb7 morphants. However, 25% of hspb12 morphants lacked both lefty1 and lefty2 expression, and there were no embryos with exclusively right-sided expression of the lefty genes. While the high frequency of absent expression of lefty1 and lefty2 was clearly abnormal, the lack of expression of both genes was still in agreement, indicating that the disruption of laterality determination caused by hspb12MO1 affects both heart and brain in the same manner.

The heart-gut laterality of hspb12 knockdown embryos was less concordant than heart-brain laterality. Instances of diverse arrangements of laterality were represented (Table 2). Bilateral or absent livers were rare; midline hearts were primarily associated with left or right livers (Table 2). Although the liver and gut both depend on hspb12 to establish the direction of laterality, the laterality of the visceral organs was less firmly linked to heart direction than for hspb7 knockdown. However, this increase in discordance was primarily due to the failure of midline hearts to jog or loop, while the visceral organs formed on either the left or right, and may therefore reflect a failure of heart migration rather than an uncoupling of heart and visceral laterality.

Concordance in hspb7-hspb12 double knockdowns

Double hspb7-hspb12 morphants also displayed primarily concordant asymmetry (Table 1) between heart and brain. Discordant double hspb7-hspb12 knockdown embryos consisted of those with bilateral lefty2 in the heart and left-sided lefty1 in the brain. The majority of embryos either expressed lefty1 and lefty2 on the left, or failed to express these genes at all. This pattern more closely resembled the effect of hspb12MO1 than hspb7MO1.

Attempts to rescue hspb7 and hspb12 morpholinos

We attempted to rescue the effects of the hspb7 and hspb12 morpholinos with in vitro transcripts of hspb7 and hspb12 RNAs. Three Danio rerio hspb7 constructs were made, containing different Kozak sequences: a consensus sequence, a BamH1 site or the endogenous 5′ untranslated region. None of the RNAs from these constructs rescued hspb7MO1 when injected to the cytoplasm at the 1-cell stage (Supplementary Figure 3A, B and data not shown). hspb7 RNA alone had no effect on laterality (Supplementary Figure 3B). A human hspb7 construct (Vos et al., 2009) also failed to rescue hspb7MO1-injected embyros. Both the zebrafish and human cDNA constructs produced Hspb7 protein in transfected cultured cells (not shown). qPCR demonstrated the survival of hspb7 RNA at 50 times the normal level; β-galactosidase enzymatic activity demonstrated the correct targeting of the RNA (data not shown). RNA encoding Danio rerio hspb12 also failed to rescue defects caused by hspb12MO1 (Supplementary Figure 3C). Yolk targeting of RNA encoding hspb7 did not rescue cardia bifida induced by hspb7MO1+MO2 injected to the yolk (Supplementary Figure 3D). Human Hspb7 is alternatively spliced (Krief et al., 1999), so we searched for alternative transcripts using 5′ and 3′ RACE on mRNA from 12 hpf zebrafish embryos, based on primers within the conserved α-crystallin domain in the second exon of hspb7. These procedures amplified only the known sequence of hspb7.

Kupffer’s vesicle cilia function

Because defects in the form or function of Kupffer’s vesicle can alter spaw expression and left-right asymmetry (Amack et al., 2007; Essner et al., 2005), we analyzed KV cilia in hspb7MO1-injected embryos. Many KV cilia in hspb7MO1 injected embryos failed to beat normally. While KV cilia in controls showed rapid vortical beating (Movie 1), cilia in hspb7MO1 injected embryos were often immotile or beat erratically (Movie 2 and Movie 3). Blind scoring of individual cilia revealed that 95% of KV cilia showed normal motility in control MO-injected embryos (n=183 from 32 embryos), whereas the number of normal cilia was reduced to 44% in hspb7MO1 fish (n=109 cilia from 30 individuals) and 57% in hspb7MO2 embryos (n=129 cilia from 26 individuals) (Figure 7A). To determine whether these cilia motility defects were sufficient to disrupt asymmetric fluid flow in KV, we injected fluorescent beads into the KV lumen. Beads injected into control MO embryos revealed a counterclockwise flow (n=7/7; Movie 4 and Figure 7B) as previously described in wild-type embryos (Essner et al., 2005). In contrast, this directional flow was lost in most hspb7MO1-injected embryos (n=6/9), in which beads moved randomly via Brownian motion (Movie 5 and Figure 7C). These results indicate that paralysis or abnormal motion of cilia in hspb7 MO embryos disrupts asymmetric fluid flow that is critical for normal L-R patterning.

Figure 7. Hspb7 knockdown disrupts cilia motility and asymmetric flow in Kupffer’s vesicle.

(A) Analysis of cilia motility revealed that many cilia in hspb7 MO embryos were immotile or beat abnormally relative to controls. (B–C) Maximum projections of fluorescent bead movements superimposed on images of KV. Beads moved in a counterclockwise direction in control MO embryos (n=7/7 embryos; B), whereas this directional flow was absent in most hspb7 MO embryos (n=6/9 embryos; C). D–F. Transmission electron micrographs showing: D. transverse section of cilium from embryo injected with control MO showing 9+2 configuration of microtubules. E. transverse section showing 9+0 configuration of microtubules in a cilium from embryo injected with hspb7 MO1+MO2. F. transverse section shows the abnormal central microtubule configuration in a cilium from embryo injected with hspb7 MO1. Arrows: outer dynein arms.

KV cilia central pair defect in hspb7 knockdown embryos

The abnormal or absent motion of cilia in KV suggested that there might be a defect ultrastructure of the cilia in embryos with reduced hspb7. Most KV cilia (7/8) in embryos injected with control MO had typical 9+2 configurations of microtubules within the ciliary axoneme (Figure 7D). However, one cilium in a control embryo lacked a central pair, confirming that both 9+0 and 9+2 cilia exist in the zebrafish KV (Gao et al., 2010). In contrast, all cilia (10/10) in hspb7 MO1+MO2 injected embryos lacked a central pair (Figure 7E). In addition, a cilium with a single central microtubule was detected in an hspb7MO1 injected embryo (Figure 7G), which has a lower degree of knockdown. Outer dynein arms were present in all cilia, whether or not there was a central pair. The loss of the central pair may account for the immotility of cilia in hspb7 morphant embryos.

Discussion

This work illustrates a role for two small heat shock proteins in supporting the establishment of laterality and the fusion of the bilateral heart fields in zebrafish through roles in Kupffer’s vesicle and the yolk syncitial layer.

Role of hspb7 in laterality

Knockdown experiments with two splice-blocking morpholino oligonucleotides against hspb7 revealed that it is required for the establishment of left-right asymmetry of the heart, gut and brain (Figure 2A, Table 1, 2). A third MO targeted to the AUG site of hspb7 (kindly shared by Gabriel Musso and Calum MacRae) produced similar rates of laterality disruptions. The high degree of concordance between the direction of asymmetry in the three organs (Table 1, 2) is consistent with a disruption of the global vector of asymmetry at the level of KV (Amack and Yost, 2004; Chen et al., 2001; Essner et al., 2005). Although KV appeared morphologically normal, the majority of the cilia were paralyzed or moved abnormally in hspb7 morphants (Figure 7A–C). hspb7 expression was never detected in otic vesicles or intermediate mesoderm; therefore it is not surprising that hspb7 knockdown embryos did not develop other hallmarks of cilia dysfunction such as supernumerary otoliths or enlarged nephric ducts. The timing and location of hspb7 expression in KV at the 1–4 somite period (Figure 1D) is consistent with a role in establishing cilia function (Amack et al., 2007).

Knockdown of hspb7 altered the ultrastructure of KV cilia. A 9+2 axoneme structure is found in the motile cilia that comprise the majority of KV cilia(Kreiling et al., 2007), although 9+0 cilia are also present (Afzelius, 2004; Fisch and Dupuis-Williams, 2011; Gao et al., 2010). In hspb7MO1+MO2 embryos all of the ciliary axonemes analyzed had 9+0 structure (Figure 7E). In the milder hspb7MO1 knockdown, the loss of one of the central pair of microtubules was observed (Figure 7F). These anomalous structures may account for the motility impairments in KV cilia of the hspb7MO1 knockdown embryos (Figure 7A–C).

This finding raises the question of how hspb7 influences the structure of the ciliary axoneme. The small heat shock proteins have a diverse set of cellular functions, and regulate cytoskeletal dynamics, protein aggregation and redox states even under conditions that do not provoke cellular stress (Boncoraglio et al., 2012; Christians et al., 2012; Sun and MacRae, 2005; Wettstein et al., 2012). Hspb7 prevents assembly of F-actin, conferring a cardioprotective activity by preventing RhoA-GTPase remodeling of the actin cytoskeleton (Ke et al., 2011), and other members of the sHsp family also regulate actin assembly (Sun and MacRae, 2005). Actin dynamics govern cilia outgrowth in KV (May-Simera et al., 2010; Ravanelli and Klingensmith, 2011). Thus a link between Hspb7 and the actin cytoskeleton is one possible mechanism for its effect on KV cilia structure. A yeast two-hybrid screen identified Filamin A, a protein that links actin to integrins, as a binding partner for human HSPB7 (Krief et al., 1999; Zhou et al, 2010). These data suggest avenues for further investigation of the molecular mechanisms of the developmental effects of sHsps.

hspb7 knockdown led to bilateral expression of southpaw (Figure 4A–C). This result is consistent with a requirement for hspb7 in KV cilia motility. Mutations and mRNA knockdowns that disrupt KV cilia motility prevent the restriction of spaw to the left lateral plate mesoderm (Bisgrove et al., 2012; Bisgrove et al., 2005; Kishimoto et al., 2008; Ravanelli and Klingensmith, 2011; Schottenfeld et al., 2007; Serluca et al., 2009). The self-propagation of spaw expression (Long et al., 2003; Wang and Yost, 2008) is checked at its midline, anterior and posterior borders by multiple barriers: antagonism by lefty1 in notochord (Meno et al., 1998; Nakamura et al., 2006), by lefty2 in the heart field (Lenhart et al., 2011), by both charon in the peri-KV region at 8 somite stage (Hashimoto et al., 2004; Lopes et al., 2010; Matsui and Bessho, 2012) and by competition between the BMP and Nodal pathways for their common downstream factor Smad4 (Furtado et al., 2008). Knockdown of hspb7 derepresses spaw expression in the right lateral plate, but does not disrupt these midline barriers to spaw expansion (Figure 4A, C); lefty1 in the notochord is intact, and the anterior and posterior extent of spaw are normal. Disruption of these midline barriers would result in discordant asymmetry (Bisgrove et al., 2000; Chin et al., 2000; Lenhart et al., 2011). However, there is a high degree of concordance between heart, gut and brain laterality in hspb7 morphants (Table 1, 2), indicating that hspb7 is not necessary for the integrity of the midline.

Genes downstream of spaw in the laterality cascade – lefty1 in brain, lefty2 in heart, and pitx2 in the left lateral plate (Long et al., 2003) – are also abnormally expressed in hspb7 morphants (Figure 4F–I). Although spaw becomes primarily bilateral, the majority of hspb7 knockdown embryos express the downstream genes on either the left or right. This resolution from bilateral spaw to a mixture of right, left, and bilateral pitx2, lefty1 and lefty2 is similar to that observed with other laterality mutants and morphants (Hashimoto et al., 2004; Hatler et al., 2009; Lin and Xu, 2009; Schottenfeld et al., 2007). Because the rate of anterior expansion of spaw determines the direction of asymmetry, this is a potential mechanism for resolving bilaterally symmetric spaw expression to unilateral (although randomized) expression of the downstream genes (Wang and Yost, 2008). In summary, these results indicate that hspb7 is required in KV to support cilia motility (Figure 8A), which is required for repression of spaw on the right, and restriction of the downstream genes pitx2, lefty1 and lefty2 to the left side.

Figure 8. Model for hspb7 and hspb12 function.

A. Mechanism of small heat shock proteins in laterality determination. Hspb7 protein is required for KV cilia motility, which leads to the upregulation of Ca²⁺ to the left of the node. This signal is propagated to the lateral plate mesoderm, where spaw expression expands in a positive feedback loop. Loss of hspb7 expression leads to impaired cilia motility, derepressing spaw on the right. hspb12 plays a supportive role in KV, but its role is revealed when hspb7 is reduced. B. hsp7 and hspb12 in the YSL are necessary for the migration of the precardiac mesoderm (PCM). Reduction of hspb7 and hspb12 specifically in yolk inhibits subsequent migration of the precardiac mesoderm to the midline, causing cardia bifida.

hspb12 and laterality hspb12

knockdown appears to have comparable effects on zebrafish laterality to those of hspb7 knockdown, although with lower penetrance (Figure 2A, 3A). Reduction of hspb12 mRNA perturbs pitx2, lefty1 and lefty2 expression (Figure 5D, E, I); the dependence of spaw on hspb12 depends upon the genetic background. Co-injection of sub-optimal doses of hsp7MO1 and hspb12MO1 induce a higher proportion of right and failed (midline) heart jogs than either MO alone (Figure 3C). Although hspb12 was not detected in KV in embryos of the AB line, it is likely that it is expressed in KV at low levels, which may vary by strain. Therefore, hspb12 may play a supporting role in KV cilia function (Figure 8A), which is primarily revealed when hspb7 is reduced. Hspb7 forms homo-multimers with up to 12 subunits (Yang et al., 2011), and it interacts with at least one other sHsp, Hspb8 (Sun et al., 2004). Hspb7-Hspb12 binding has not been studied because mammals lack hspb12, but they are plausible binding partners due to the homology of the α-crystallin domains. These results demonstrate redundancy between hspb7 and hspb12 in the laterality gene cascade.

There are some intriguing differences between hspb7 and hspb12. In hspb12 deficient embryos, the pattern of pitx2 expression differs from that of lefty1 and lefty2. Although ectopic pitx2 may be expressed on the left, bilaterally, or on the right (Figure 5D), lefty1 and lefty2 are expressed on the left, bilaterally or are completely absent (Figure 5E, I) in hspb12 morphants; lefty1 and lefty2 are never expressed on the right side alone. pitx2 is regulated not only by spaw but also by Bmp activity in the peri-KV region (Chocron et al., 2007). bmp4 expression near the node is normal in hspb12 knockdown embryos (data not shown). The heart and brain expression of lefty2 and lefty1 are largely concordant in hspb12 knockdown embryos; however, the heart and gut laterality are more frequently discordant (Table 2). The disparity between the lefty genes and gut migration may stem from the fact that only pitx2 is expressed in the gut precursors, and we observed that pitx2 expression levels are somewhat reduced in hspb12 morphants. In addition, hspb12 may support the migration necessary for the heart to jog at 26 hpf, and therefore the disparity between heart and gut migration may be due to a late failure of heart migration rather than uncoupling of heart and gut laterality. Taken together, these results suggest that hspb12 has functions in establishing laterality that do not entirely duplicate those of hspb7.

hspb7, hspb12, and heart migration

In addition to a role in KV cilia motility, hspb7 is also necessary to support migration of precardiac mesoderm along the YSL. Although a mild knockdown of hspb7 causes laterality defects (Figure 2A), a strong knockdown to 12% of control levels (Figure 2F, Supplementary Figure 1) causes cardia bifida and retards cardiac precursor migration (Figure 2G–J). hspb7 is expressed both in YSL from 6–12 hpf, and in the migrating precardiac mesoderm from the 12 somite stage (Figure 1A–D). To clarify which expression domain is responsible for the laterality and migration defects, we injected morpholino into the yolk between 3–4 hpf. Yolk injections restrict the MO to cells adjacent to the yolk, including the YSL. This injection method does not permit any MO to reach the embryo proper (Amack and Yost, 2004; Arrington and Yost, 2009; Essner et al., 2005). Yolk targeting of the combined hspb7MO1 and hspb7MO2 is sufficient to cause cardia bifida (Figure 2K), indicating that the failure of precursors to migrate is due to a requirement for hspb7 in the YSL rather than in the precardiac mesoderm (Figure 8B). In strong hspb7 morphants there is clearly a defect in the migration of cardiac precursors to the midline at the 18–20 somite stage. In yolk-injected hspb7 morphants that do form a fused heart tube, the failure of the atrial end of the heart to jog left by 26 somites may also be due to a migration defect, in this case blocking migration away from the midline. In the more sensitive line Tg(cmlc:GFP) x WT line, yolk injection of hspb12MO also causes a small amount of cardia bifida (8.7% vs 0% of control), and slows anterior migration (Supplementary Figure 2).

Together these lines of evidence indicate that hspb7, and to a lesser extent hspb12, are important in making the YSL capable of supporting precardiac mesoderm migration (Figure 8B). YSL-specific knockdowns and mutations have previously demonstrated that the YSL drives cardiac migration, through a common pathway that promotes fibronectin expression in the heart field (Arrington and Yost, 2009; Kupperman et al., 2000; Matsui et al., 2007; Osborne et al., 2008; Sakaguchi et al., 2006; Trinh and Stainier, 2004). Notably, knockdown of the small heat shock protein hspb1 (hsp27) results in cardia bifida in Xenopus (Brown et al., 2007), although it does not do so in zebrafish (Tucker et al., 2009). We confirmed that a second hspb1 MO translation-blocking MO did not affect laterality or heart fusion (MJM and LDH, unpublished data). In zebrafish, hspb7 may fill the role of hspb1 in Xenopus, despite their limited homology outside the α-crystallin domain (Elicker and Hutson, 2007).

Inability to rescue hspb7 and hspb12 morphants

Surprisingly, neither whole embryo nor yolk-specific replacement of hspb7 or hspb12 mRNA was sufficent to rescue the hspb7 or hspb12 morphants. We were unable to detect an alternative transcript of hspb7, suggesting that we had the only transcript present in the early zebrafish embryo. The constructs were expressed in cultured cells, the injected mRNA survived in embryos at least 24 hours, and injected β-galactosidase RNA was targeted to the whole embryo or yolk as expected. It is possible that post-transcriptional modifications to the RNA, or a very specific pattern of expression is necessary to obtain rescue of the hspb7 and hspb12 morphants.

This work demonstrates that hspb7 and hspb12 have partially overlapping roles in establishing embryonic lateral asymmetry in the zebrafish embryo through their effects on Kupffer’s vesicle and the downstream laterality genes. We show that hspb7 is necessary for the presence of the central pair of microtubules in the motile cilia in Kupffer’s vesicle. Furthermore, both hspb7 and hspb12 support the migration of the heart precursors to the midline through a role in the yolk syncitial layer. These findings add to a growing list of developmental roles for small heat shock proteins.

Supplementary Material

01. Movie 1.

Normal cilia beating in Kupffer’s vesicle in a control MO injected embryo. Arrows point to representative cilia.

02. Movie 2.

Cilia motility defects in an hspb7 MO-1 injected embryo. In this embryo, some cilia in Kupffer’s vesicle beat normally (green arrows), whereas others show abnormal beating (e.g. red arrow).

03. Movie 3.

Immotile cilia in an hspb7 MO-1 injected embryo. Arrows point to cilia that fail to beat in Kupffer’s vesicle.

04. Movie 4.

Normal directional flow in Kupffer’s vesicle of a control MO injected embryo. This movie starts with a brightfield image of Kupffer’s vesicle lumen and then switches to the fluorescent channel. Fluorescent beads injected into Kupffer’s vesicle flow in a counterclockwise direction. The green circle approximately outlines Kupffer’s vesicle lumen.

05. Movie 5.

Loss of flow in Kupffer’s vesicle of an hspb7 MO-1 injected embryo. This movie starts with a brightfield image of Kupffer’s vesicle lumen and then switches to the fluorescent channel. Fluorescent beads injected into Kupffer’s vesicle move randomly. The green circle approximately outlines Kupffer’s vesicle lumen.

Highlights.

• Small heat shock proteins hspb7 and hspb12 are necessary for embryonic asymmetry

• hspb7 and hspb12 expression in yolk syncytial layer supports heart field migration

• hspb7 is required for cilia motility in Kupffer’s vesicle

• Reduced hspb7 converts KV cilia to 9+0