|Home | About | Journals | Submit | Contact Us | Français|
The Nodal-related subgroup of the TGFβ superfamily of secreted cytokines regulates the specification of the mesodermal and endodermal germ layers during gastrulation. Two Nodal-related proteins - Squint (Sqt) and Cyclops (Cyc) - are expressed during germ-layer specification in zebrafish. Genetic sqt mutant phenotypes have defined a variable requirement for zygotic Sqt, but not for maternal Sqt, in midline mesendoderm development. However a comparison of phenotypes arising from oocytes or zygotes injected with Sqt antisense morpholinos has suggested a novel requirement for maternal Sqt in dorsal specification. In this study we examined maternal-zygotic mutants for each of two sqt alleles and we also compared phenotypes of closely related zygotic and maternal-zygotic sqt mutants. Each of these approaches indicated there is no general requirement for maternal Sqt. To better understand the dispensability of maternal and zygotic Sqt, we sought out developmental contexts that more rigorously demand intact Sqt signalling. We found that sqt penetrance is influenced by genetic modifiers, by environmental temperature, by levels of residual Activin-like activity and by Heat-Shock Protein 90 (HSP90) activity. Therefore, Sqt may confer an evolutionary advantage by protecting early-stage embryos against detrimental interacting alleles and environmental challenges.
During gastrulation key cell lineages are specified that undergo coordinated movements to establish the vertebrate body plan. This process is sensitive to environmental and genetic changes and its disruption can lead to a range of birth defects (Webster et al., 1988). Several signaling pathways have been identified with essential roles during vertebrate gastrulation (Schier and Talbot, 2005). Among these is the Nodal signaling pathway, which is required for the specification of endoderm and most mesoderm (Whitman, 2001).
Nodal signaling is initiated by the binding of Nodal-related ligands to Type I and Type II receptor serine-threonine kinases, leading to the activation of R-Smads 2 or 3 by phosphorylation. Phospho-R-Smads then assemble with Smad4 and other factors, such as the Fox family protein, FoxH1, forming DNA-binding complexes that regulate target gene transcription. There are three Nodal-related proteins encoded by the zebrafish genome: Squint (Sqt), Cyclops (Cyc) and Southpaw (Spaw) (Feldman et al., 1998; Long et al., 2003; Rebagliati et al., 1998b; Sampath et al., 1998). Sqt has a unique long-range signaling activity and the sqt gene has an evolutionarily-conserved 3′ untranslated region (UTR) that can target mRNA to the dorsal cell lineage (Chen and Schier, 2001; Gore et al., 2005). The sqt gene is also the earliest expressed of the three nodal-related genes, with maternal transcripts localized to the dorsal cell lineage as well as very early zygotic transcripts (Gore and Sampath, 2002; Rebagliati et al., 1998a). During the late blastula stage, sqt is expressed in mesendoderm precursor cells and cyc transcription initiates in these same cells (Rebagliati et al., 1998a). sqt expression is down regulated during gastrulation and is absent at later stages, whereas cyc expression persists in midline mesendoderm. During early segmentation stages, spaw expression commences in the paraxial mesoderm and during later segmentation, spaw and cyc are expressed within specific domains on the embryo’s left flank (Long et al., 2003).
Homozygous carriers of sqt and cyc mutations display cyclopia, among other phenotypes (Hatta et al., 1991; Heisenberg and Nusslein-Volhard, 1997). Cyclopia is also observed in human holoprosencephaly and the human and fish conditions have similar etiologies involving reductions in anterior midline mesendoderm and failed differentiation of the overlying ventral prosencephalon, or forebrain (Ekker et al., 1995)(Muenke and Beachy, 2000).
Cyclopia and its underlying anterior defects are generally lethal, such that affected sqt and cyc mutants fail to thrive (BF and WP, unpublished observations). However, while cyc mutations cause cyclopia in every embryo, sqt deficiency is incompletely penetrant, allowing embryos homozygous for the sqtcz35 allele or wild type (WT) embryos treated with antisense Sqt morpholinos to frequently escape cyclopia and even develop as viable and fertile adults (Aoki et al., 2002; Feldman and Stemple, 2001). This viability of sqt mutants likely represents a recovery rather than a failure to be affected, since an earlier sqt phenotype -,delayed formation of the dorsal organizer - shows complete penetrance (Dougan et al., 2003; Feldman et al., 1998). The ability of sqt mutants to survive and breed raises the question: why has this dispensable gene been evolutionarily conserved in zebrafish and other teleosts (Gore et al., 2005)? To address this question, in this study we have examined the parameters controlling sqt penetrance. We find that sqt penetrance is modulated by inherited factors and also by the embryo’s environment, in the form of developmental temperature. Thus, the Sqt protein may confer an evolutionary advantage by protecting embryos against deleterious allele combinations and temperature extremes. At the same time, the sqtcz35 mutant allele was isolated as a spontaneous mutation, and as such has been able to persist in laboratory stocks (Feldman et al., 1998). We considered two molecular pathways that might enable a deleterious sqt mutation to be tolerated in a population. Considering that sqt and cyc are partially redundant, we asked whether residual activity by Cyc or some other Activin-like signal can mask the sqt phenotype, and we find this to be the case. We also looked at Heat Shock Protein 90 (HSP90), which has been proposed to mask aberrant developmental phenotypes by chaperoning key client proteins (Rutherford and Lindquist, 1998). Under conditions of stress such as heat shock, it is believed that HSP90 is diverted to de novo mis-folded proteins, thereby abandoning its key clients, which leads to the expression of new phenotypes. In view of the temperature sensitivity of sqt mutants, we asked whether interfering with HSP90 function in sqt embryos at normal temperatures might also increase the incidence of cyclopia, and we find that it does.
It has been argued that Sqt is evolutionarily conserved due to its maternal activity (Gore et al., 2005). This conclusion is based on the ability of a human NODAL 3′ UTR to localize mRNA to the zebrafish dorsal cell lineage and on experiments targeting translation of maternal Sqt. When Sqt MOs are injected into fertilized eggs, the standard sqt phenotypic spectrum arises (Feldman and Stemple, 2001; Gore et al., 2005). However when Sqt MOs are injected into unfertilized oocytes followed by in vitro fertilization, a novel class of ventralized embryos is observed, in addition to standard cyclopean and WT phenotypes (Gore et al., 2005; Kishimoto et al., 1997). These experiments suggested that maternally-supplied Sqt protein is frequently required for dorsal specification. In contrast to this, maternal-zygotic sqtcz35/cz35 (MZsqtcz35/cz35) embryos have been reported to have the same range of phenotypes as zygotic sqt mutants (Aoki et al., 2002).
If maternal Sqt were commonly required for dorsal specification and sqtcz35 were a null allele as predicted, no maternal Sqt protein would be available in MZsqtcz35/cz35 embryos and one would expect to see the same phenotypes in early antisense-treated embryos and MZsqtcz35/cz35 embryos. However ventralization was not reported as an MZsqtcz35/cz35 phenotype. It is formally possible that this reflects a peculiarity of the sqtcz35 allele, so we asked whether or not MZsqt mutants for a second predicted sqt null allele, sqthi975, display ventralized phenotypes (Amsterdam et al., 2004). We found the phenotypic range of MZsqthi975/hi975 mutants to be very similar to that of MZsqtcz35/cz35 mutants. We also assessed the role of maternal Sqt by comparing the phenotypic profile of closely-related clutches of embryos containing zygotic sqtcz35/hi975 or MZsqtcz35/hi975 mutants, and saw no substantial difference. Our data therefore argue against a general requirement for maternal Sqt for dorsal specification or for any other process.
Exploiting the low penetrance of sqt phenotypes in the King’s Wild Type and Tübingen/AB hybrid lines (see below), we isolated sqtcz35/cz35 and sqthi975/hi975 parents by PCR genotyping adult offspring of heterozygous incrosses with primers specific for sqtcz35 (Feldman et al., 1998) and sqthi975 (forward 5′ tcc ata tca gca agc gat ga 3′; reverse 5′ ggt ctc ctc tga gtg att gac tac c 3′), yielding 9 out of 67 adult sqtcz35/cz35 and 14 out of 108 adult sqthi975/hi975 fish (13.4% and 13.0%, respectively, out of an expected Mendelian fraction of 25%). For the sqtcz35 crosses in this manuscript, all of the tested parents were siblings in the King’s Wild Type line, a discontinued lineage with mixed contributions from a London pet store line and more standard lines (Stephen Wilson, personal communication). For the sqthi975 crosses, all of the tested parents were cousins in the Tübingen/AB background.
WT and MZsqthi975/hi975 embryos were cultured to either the 8-cell stage (1.25 hours post fertilization [hpf] – Fig. 1E) or 40% epiboly (5 hpf – data not shown). Total RNA was extracted with Trizol (invitrogen). RT was then performed using oligo (dT)12–18 primers. cDNA generated from 80 ng (40% epiboly stage) or 160 ng (8-cell stage) of total RNA was used as the template for PCR with three primer pairs. Primer pair F1 (5′ gagct ttatt tcaat aactg cgtg 3′)/R1 (5′ gccag ctgct cgcat tttat tcc 3′) amplifies 163 bp from sqt’s 5′ UTR and pro-domain coding sequence. Primer pair F2 (5′ gttgg agcga ctgga ttgtt 3′)/R2 (5′ tga cca tct tgc cat tct ca 3′) amplifies 219 bp from the mature domain coding region. β-actin was used as an internal reference (β-actin-F: 5′ ttgtgaccaactgggatgac 3′, β-actin-R: 5′ agcact tcctgtgaacgatg 3′).
ARE-reporter assays were carried out using embryos from WT, low-, and high-penetrant MZsqthi975/hi975 crosses. Six one- to four cell embryos per data point were injected with 25 pg of reporter plasmids (pGL3-ARE3-Luc and pRL-CMV at a ratio of 10:1 w/w), and 30 pg of gap43-gfp or lefty1 synthetic RNA (Bisgrove et al., 1999). Injected embryos were lysed at the shield stage and luciferase activity was measured with the dual-luciferase reporter assay system (Promega).
For temperature shifts, embryos were placed in 4 ml Wheaton glass vials along with 2 ml of egg water, composed of purified (reverse-osmosis) water, supplemented with 0.006% w/v red sea salts and 0.0001% w/v methylene blue, pH ~6.4. For 34 °C or 15 °C incubations, these tubes were firmly inserted, top upwards, into inverted Styrofoam tube racks (15 ml) and floated with the vials immersed in a water bath. In some experiments, tubes were capped and a maximum of 12 embryos were placed per tube. In other experiments as many as 30 embryos were placed per tube, but the tubes were uncapped and aeration holes were included in the Styrofoam rack. Anoxia or hypoxia was achieved by incubating embryos in capped tubes in egg water under a layer of mineral oil and in the presence of E. coli membranes and sodium lactate (Jacobson et al., 1987). For experiments designed to slow development at 34 °C to standard (28 °C) rates, hypoxia treatment was limited to 1h during the late blastula stage, using 2–3 mM sodium lactate. Tests for the effects of pH, salinity or co-incubation with dead embryos were performed in standard petri dishes. Whole mount in situ hybridizations for cyc (Rebagliati et al., 1998a; Sampath et al., 1998), goosecoid (Stachel et al., 1993) and bhikhari (Vogel and Gerster, 1999) were done according to standard protocols (http://zfin.org/ZFIN/Methods/ThisseProtocol.html) assisted by an automated liquid exchanger (Biolane HTI) for post-hybridization washes.
To test the effects of drugs inhibiting HSP90, MZsqtcz35/cz35 embryos were injected with 0.7 nl of 27 mM radicicol in DMSO, 0.7 nl of 0.5 mM geldanamycin in DMSO, or with 0.7 nl DMSO alone. To test the effects of blocking translation of the HSP90A or HSP90B proteins, MZsqthi975/hi975 embryos were injected with 2 ng of either a control MO (TAGTTAAGCCTAGCTCTCATAAACT), an HSP90A MO (CGCTGTATTTCCTCT TCTCCACCTG) or an HSP90B MO (TCAGGCATCTTGGTTAGTTTCGTTG). All injected embryos were incubated overnight and scored for cyclopia on day two.
Embryos were scored as holoprosencephalic only when some degree of eye fusion (often uniquely along the ventral aspect) was detected on day two. In all experiments comparing treatments, sibling embryos were used for each genetic condition. WT and MZsqt embryos came from separate lines. To calculate P values, the numbers of embryos scored with WT eyes or with cyclopia under various conditions were compared by 2×2 chi-squared analysis using the following interactive site: (http://www.psych.ku.edu/preacher/chisq/chisq.htm).
The sqtcz35 allele (also called sqtz1) was isolated as a spontaneous insertional mutation from wild-type stocks of Darjeeling strain fish (Feldman et al., 1998; Heisenberg and Nusslein-Volhard, 1997). In a subsequent screen for retroviral insertions, a second sqt allele, sqthi975, was isolated (Amsterdam et al., 2004). sqthi975/hi975 embryos are reported to reproduce described phenotypes of sqtcz35/cz35 embryos. Both the sqtcz35 and the sqthi975 alleles have a large insertion within the pro-domain of Sqt, predicting truncated translation products that lack the mature TGFβ domain (Fig. 1A). Whole mount in situ hybridization of sqtcz35/cz35 embryos revealed highly reduced or absent levels of sqtcz35 (Dougan et al., 2003). Similarly, we see reduced or absent levels of maternal and zygotic sqthi975 transcripts in 8-cell stage and 40% epiboly stage MZsqthi975/hi975 embryos probed by reverse-transcription PCR (Fig. 1E and data not shown). Therefore, even if the sqthi975 allele can produce a functional Sqt ligand through an unrecognized mechanism, the reduction in transcript levels should still impair its function, and the same is likely true for the sqtcz35 allele.
sqthi975/hi975 fish were isolated by genotyping natural adult progeny from heterozygous sqthi975/+ crosses. The sqthi975/hi975 fish were then interbred to generate clutches of MZsqthi975/hi975 progeny. We used the same procedure to produce MZsqtcz35/cz35 embryos, as has been previously described (Aoki et al., 2002). Consistent with previous reports, we observed phenotypically WT and cyclopean MZsqtcz35/cz35 embryos, and no ventralized embryos, with the rate of cyclopia per clutch ranging from 0% (N=99) to 27% (N=100) over 27 crosses scored (Table S1).
It is formally possible that the observation of ventralized phenotypes in antisense-treated embryos, but not in MZsqtcz35/cz35 embryos is due to partial activity of the sqtcz35 allele. To address this possibility, we examined the phenotypes of MZsqt embryos carrying a second predicted null allele, sqthi975, which carries a large insertion in the exon encoding the N-terminal half of the mature Sqt ligand (Fig. 1A). We found a very similar phenotypic spectrum among MZsqthi975/hi975 embryos as among MZsqtcz35/cz35 embryos. The rates of cyclopia of the examined MZsqthi975/hi975 clutches ranged from 2% (N=57) to 49% (N=141) (Table S1). Affected MZsqthi975/hi975 embryos from 27 of 28 crosses displayed only cyclopean phenotypes, with a range of severity (Figure 1 C-C′, D-D). We did observe some ventralized embryos in the highest penetrant cross, however no ventralized embryos were seen in subsequent crosses of the same pair (Table S1). In summary, scoring 27 clutches of MZsqtcz35/cz35 and 28 clutches of MZsqthi975/hi975 embryos, we conclude that there is no general requirement of maternal Sqt for dorsal specification.
To further examine the role of maternal Sqt, we compared the phenotypic spectra in closely related embryos that either had or lacked WT maternal Sqt. We generated Zsqthi975/cz35 embryos by crossing two sqthi975/hi975 males with two sqtcz35/+ females, producing a total of 331 embryos. We generated MZsqthi975/cz35 embryos by crossing the same two sqthi975/hi975 males with two sqtcz35/cz35 females (siblings of the sqtcz35/+ females), producing a total of 286 embryos. One of the contributing males had previously sired 18% (N=607) cyclopean MZsqthi975/hi975 embryos when crossed to a sqthi975/hi975 female. Despite this high penetrance in another context, the rate of cyclopia in both sets of hybrid crosses were exceedingly low, with only three cyclopean Zsqthi975/cz35 embryos and zero cyclopean MZsqthi975/cz35 embryos detected. We suspect this low penetrance reflects hybrid vigor among the embryos, since the parents come from two different lineages. Regardless of the low penetrance, the incidence of cyclopia did not change with the loss of maternal Sqt and no novel phenotypes were observed. These experiments therefore provide direct evidence that loss of maternal Sqt does not exacerbate the phenotypic outcome of embryos lacking zygotic Sqt.
To better understand why Sqt has been evolutionarily retained despite its frequent dispensability, we went on to perform experiments aimed at identifying the parameters controlling the degree of sqt penetrance. We first asked whether heritable modulators might influence phenotypic penetrance. We crossed dozens of age-matched sqt-deficient fish and scored the incidence of cyclopia among their progeny. Under standard laboratory conditions, repeated crosses of the same parents produced consistent rates of cyclopia among progeny. This is demonstrated by the relatively tight error bars derived from repeat crosses of the sqt mutant pairs, as shown in Fig. 1F and 1G for MZsqtcz35/cz35 and MZsqthi975/hi975, respectively. By contrast, significantly different cyclopia rates can be transmitted from distinct pairs. All parents used for Fig. 1F were siblings from a single clutch and all parents used for Fig. 1G were age-matched cousins. Thus, heritable factors unevenly distributed in a particular lineage can determine sqt penetrance. Our ability to obtain varying allelic combinations in close relatives may be due to the both alleles are in hybrid backgrounds that likely carry polymorphic alleles (Stickney et al., 2002). Further studies in which the age-matched sqtcz35 mutant siblings were re-bred in various combinations revealed a complex transmission pattern of sqt susceptibility/resistance, ruling out simple models of inheritance, such as sex linkage or a single interacting allele (Table 1).
Apart from Sqt, the zebrafish genome encodes several other TGFβ-related ligands capable of inducing Smad2/3 phosphorylation: Cyc, Spaw, Dvr1, and several Activin subunits (DiMuccio et al., 2005; Dohrmann et al., 1996; Wang and Ge, 2003; Wu et al., 2000). The cellular responses generated by these proteins are classified as Activin-like. To examine whether residual Activin-like signaling might compensate for the lack of Sqt, we asked whether there is a correlation between the penetrance of the sqt phenotype and formation of an active Smad/FoxH1 complex. We did this by injecting embryos from high and low-penetrant pairs with a vector containing luciferase driven by a FoxH1-responsive promoter, the Activin response element (ARE), and comparing the accumulated luciferase activity at shield stage (6 hours post fertilization [hpf]) (Huang et al., 1995; Yan et al., 2002). WT embryos injected with lefty1 mRNA were used to determine baseline ARE activity in the absence of Nodal activity. These studies revealed an inverse correlation between endogenous Smad/FoxH1 activity and the penetrance of cyclopia in embryos lacking Sqt (Fig. 1H). Thus in the absence of functional Sqt, sufficient levels of residual Activin-like signaling can protect embryos against cyclopia and the level of Activin-like signaling is modulated by heritable factors.
In the above studies and elsewhere, the majority of Sqt-deficient clutches we examined had low penetrance. This suggested to us that Sqt might have been evolutionarily conserved to cope with environmental conditions not encountered under standard laboratory husbandry. To explore this idea, we examined the effects of five environmental parameters on the development of MZsqt embryos, namely, salinity, oxygenation, pH, signals from dying/decaying embryos and temperature.
Husbandry of zebrafish embryos outside of their aquaria is typically done in purified water that is supplemented with 0.006% w/v sea salt and 0.0001% w/v methylene blue and adjusted to pH 6.4. We observed no significant changes in the rate of cyclopia in embryos reared at a higher salt concentration (0.012% w/v), altered pH (pH 8), or in the presence of decaying or unfertilized embryos, although we did note an increased rate of morbidity in clutches with dead or unfertilized embryos (data not shown).
Hypoxia or anoxia caused decelerated embryonic growth (under hypoxia) or arrest at the shield stage (under anoxia), as has been described (Padilla and Roth, 2001). When returned to an oxygenated environment, however, treated embryos resumed development with no significant increase in cyclopia incidence. By contrast, overnight exposure to heat (34 °C rather than 28 °C) caused a dramatic increase in cyclopia incidence among MZsqt (Fig. 2B) but did not cause any cyclopia in WT embryos (Fig. 2A). In addition to increases in cyclopia incidence, we observed other heat-specific phenotypes in certain MZsqt embryos, namely similarity to sqt;cyc compound mutants (Fig. 2B) and midline bifurcations (Fig. 2D).
To better define the developmental processes that are heat labile and contribute to normal forebrain patterning, we incubated MZsqtcz35/cz35 embryos at 34 °C for varying windows of time during their first day of development. Assaying a few embryos per time point, we discerned two developmental phases when heat treatment corresponded to substantially increased cyclopia. Early induction of cyclopia was seen in MZsqtcz35/cz35 embryos heated for 2h or 1h within the 0–3 hpf time window, or for 0.5h within the 0–2.25 hpf time window (Figure 3A and Table S1). We also observed substantial lethality arising from these early heat treatments, due to heat sensitivity during early cleavage stages (Table S1 and data not shown). Later cyclopia induction was seen in MZsqtcz35/cz35 embryos heated for 2h within the 2–6 hpf time window or for 1h within the 2.5–5 hpf time window, but not for MZsqtcz35/cz35 embryos heated for only 0.5h within the 2–5.75 hpf time window (Fig. 3A and Table S1). No sensitivity was noted beyond 6 hpf, even in the extreme case of MZsqtcz35/cz35 embryos heated for 18h within the 6–24 hpf time window (Fig. 3A and Table S1).
We reproduced these findings with MZsqthi975/hi975 embryos, using larger numbers (Fig. 3B–C). For the early sensitive phase, 0.5 h of heat exposure prior to the onset of zygotic transcription significantly elevates cyclopia incidence in MZsqthi975/hi975 embryos (Fig. 3B, Table S1). By starting the heat treatment after early cleavage stages, lethality was minimized. For the later sensitive phase, 1h of heat is sufficient to induce cyclopia in MZsqthi975/hi975 embryos and we detected no significant heat sensitivity beyond 6 hpf, when gastrulation begins (Fig. 3C, Table S1). We further found that cold treatment (15 °C for 1h) of MZsqthi975/hi975 embryos during the later sensitive phase also causes a significant increase in cyclopia (Fig. 3D). Finally, to demonstrate that temperature-sensitivity is not particular to the sqt alleles or their genetic backgrounds, we determined that WT embryos from an independent background injected with a Sqt MO are also sensitive to heat (Fig. 3E). In summary, the above experiments have uncovered two early developmental phases sensitive to combined temperature stress and loss of Sqt and which are important for later patterning of the forebrain.
The second of the heat-sensitive phases defined in the above experiments coincides with the time that cyc is normally initiated, both in WT and sqtcz35/cz35 embryos (Dougan et al., 2003). To see whether temperature affects Nodal signaling activity in WT embryos and MZsqt mutants, we examined the expression of cyc and two Nodal-related protein targets - goosecoid (gsc) and bhikari (bik) - in heat-treated embryos. Whole-mount in situ hybridizations of shield-stage MZsqthi975/hi975 embryos heated prior to gastrulation consistently displayed substantial reductions in the expression of all three genes in the dorsal organizer domain, whereas expression of these markers in heat-treated WT embryos remained normal (compare Fig. 4G and 4H; 4I and 4J; 4K and 4L). As expected for embryos lacking Sqt, a general decrease in cyc, gsc and bik transcription was also apparent in untreated MZsqt embryos (compare Fig. 4A and 4G; 4C and 4I; 4E and 4K). ARE-luciferase assays comparing FoxH1-driven transcription in control and temperature-shifted MZsqthi975/hi975 embryos at shield stage also revealed a loss of Activin-like signaling, specifically in heat- or cold-treated MZsqt embryos (Fig. 4M and 4N). A significant loss in Activin-like signaling was also seen in cold-treated WT embryos, which never displayed cyclopia (Fig. 4N), suggesting that the activity of Sqt or the recovery of Cyc is high enough in WT embryos to protect them from cyclopia.
The molecular chaperone HSP90 has a select set of client proteins, many of which act in signaling pathways affecting growth and development (Sangster et al., 2004). Consistent with the roles of its clients, it has been observed that developmental phenotypes induced by environmental stress can be phenocopied by pharmacological inhibition of HSP90 (Rutherford and Lindquist, 1998). A proposed mechanism for this phenomenon is that under conditions of stress, HSP90 is diverted to de novo mis-folded proteins and abandons its key clients, leading to the expression of developmental phenotypes. In view of the increased cyclopia among MZsqt embryos subjected to temperature stress, we wondered whether HSP90 might normally chaperone a temperature-sensitive factor in MZsqt embryos, thereby masking cyclopia.
To test this idea, we injected MZsqtcz35/cz35 embryos with either of two specific HSP90 inhibitors: geldanamycin and radicicol. As an alternative strategy for diminishing HSP90 function, we injected antisense morpholinos targeting the translation of the two known zebrafish HSP90 orthologues – hsp90a and hsp90b - in MZsqthi975/hi975 embryos (Krone and Sass, 1994). All of these treatments caused a dramatic increase in the incidence of cyclopia specifically in MZsqt embryos (Fig. 5E–H) as well as a variety of defects common to MZsqt and WT embryos, including small heads, short axes, reduced notochords and compressed, U-shaped somites (Lele et al., 1999) (Fig. 5B, C, E, F). These data demonstrate that Sqt, HSP90A and HSP90B act together under normal developmental conditions to protect embryos against cyclopia. Furthermore, these results are consistent with the hypothesis that the temperature-sensitive effector of Sqt-independent Nodal signaling is an HSP90 client, but they do not rule out the possibility of indirect synergy between Nodal signaling and a distinct HSP90-dependent pathway.
Many of the MZsqt embryos that we scored for this study displayed WT or cyclopean phenotypes, but we only saw ventralized embryos like those reported by Gore et al. in a single cross (Gore et al., 2005). Over several years of breeding MZsqt embryos, we have encountered a handful of additional breeding pairs of MZsqthi975/hi975 fish and one breeding pair of MZsqtcz35/cz35 fish that yielded ventralized embryos. It is difficult to assess the significance of such sporadic findings, since we have also encountered rare pairs of fish producing “bad” embryos among WT breeders and carriers of other mutations (WP, PHW, MDC, DLS and BF unpublished observations). We therefore conclude that there is no general role for maternal Sqt in dorsal specification. One explanation for the consistent presence of ventralized embryos in the studies by Gore et al. and their general absence from our own studies is that it is a background-specific phenotype. To test this idea, it will be of interest in the future to cross the sqtcz35 and/or sqthi975 alleles into the background used by Gore et al. and re-examine the MZsqt phenotypic spectrum. Although our findings challenge the notion that maternal Sqt is generally essential, they do not exclude a role for maternal Sqt that can be compensated by other maternal factors in its absence.
By scoring the incidence of cyclopia among progeny from specific pairs of carriers, we have demonstrated that penetrance is relatively constant between clutches generated by the same parents. Furthermore, related pairs can yield significantly different rates of penetrance, indicating there are heritable modifiers with substantial effects on penetrance. The unequal distribution of these modifiers among siblings or cousins suggests there are a discrete number of loci that modify sqt. The observed distribution of cyclopia rates arising from reciprocal crosses, however, indicates a complex pattern of transmission (Table 1).
We identified two conditions – heat and cold exposure – that induced cyclopia in Sqt-deficient embryos, but not WT embryos. Our exploration of other environmental stresses, namely hypoxia and anoxia, elevated salt concentration, elevated pH and co-incubation with dead embryos, failed to identify a condition that induced cyclopia in either Sqt-deficient or WT embryos. Historically, embryological experiments in various species of fish, zebrafish included, identified heat, cold and salt treatment regimens that induce cyclopia and other “monstrosities” in WT embryos (Ingalls, 1962; Loeb, 1915). These treatments were harsh and included very high levels of mortality. The fact that WT embryos are insensitive to the treatment regimens we defined indicates that have remained within the range of Sqt’s buffering ability. The ability of Sqt to protect embryos against temperatures (34 °C and 15 °C) that arise regularly in the indigenous habitat of zebrafish (east India) provides a possible rationale for why this protein has been evolutionarily conserved. At the same time, Sqt’s dispensability under normal husbandry temperatures may explain why the spontaneous sqtcz35 allele was able to persist in laboratory stocks.
Looking at gene expression or activation of a FoxH1-reponsive plasmid, we found that there is a measurable level of residual Nodal-like signaling in MZsqt embryos. Clutches of embryos that have received genetic insults (high penetrance crosses) or environmental insults (heat and cold treatments) that increase phenotypic penetrance exhibit a decrease in this residual signaling. A straightforward explanation for this decrease in Nodal-like signaling would be that levels of Cyc are reduced. Our observation that cyc transcripts are indeed reduced in the organizer region of heat-treated MZsqt embryos (Fig. 4H) provides direct evidence for this in the case of heat treatment. By extension, perhaps variations in genetic background, cold treatment and loss of HSP90 function also affect sqt penetrance via reductions in cyc transcription. Depending on the degree of cyc reduction, by this model, MZsqt embryos might undergo WT development or phenocopy sqt;cyc compound-mutant embryos, as seen in Fig. 2B.
We also considered an alternate hypothesis – that Sqt’s long-range signaling capacity is specifically required to reach target cells during the accelerated growth that occurs at elevated temperatures (Kimmel et al., 1995). To address this question we subjected embryos to a hypoxia regimen that decelerated the rate of embryonic development at 34 °C to the standard rate of development at 28 °C. We then compared WT and MZsqtcz35/cz35 embryos that were subjected to either combined heat and hypoxia or heat alone for various times. We observed increases in cyclopia for MZsqtcz35/cz35 embryos (but not for WT embryos) under both conditions at indistinguishable rates (data not shown). We therefore conclude that the increased requirement for Sqt by heat-treated embryos is not related to changes in their rates of growth or morphogenesis.
Heat elevation reliably caused cyclopia in Sqt MO-injected embryos (Fig. 3E), MZsqtcz35/cz35 embryos (Fig 2B, ,3A)3A) and MZsqthi975/hi975 embryos (Fig. 2F, 3B, 3C). Heat sensitivity therefore appears to be a general phenomenon of Sqt-deficient embryos, rather than being allele or background specific. Of particular interest is our observation that heat treatment prior to the mid-blastula transition (MBT) is sufficient to perturb Nodal signaling (as measured by induction of cyclopia), implicating a maternal factor(s) as the heat-sensitive target. This factor is unlikely to be Sqt, which is dramatically reduced or absent from MZsqt embryos, nor Cyc, which is not maternally expressed (Rebagliati et al., 1998a; Sampath et al., 1998). It seems, rather, that the early heat-sensitive factor component lies upstream of Nodal signaling, and as such may be part of the postulated yolk-derived signal upstream of Sqt and Cyc (Mizuno et al., 1996; Rodaway et al., 1999) or the β-catenin pathway of dorsal organizer specification (Schneider et al., 1996). Furthermore, our ability to induce cyclopia by interfering with HSP90 function raises the possibility that the early and late heat-sensitive factors are HSP90 clients.
In conclusion, looking at three classes of Sqt-deficient embryos (MO-depleted and two separate alleles) we have found that under standard laboratory conditions both maternal and zygotic Sqt are generally dispensable. We have also found that sqt penetrance correlates with residual Activin-like signaling levels and cyc expression, and is substantially influenced by the distribution of genetic modifiers among closely-related fish, by temperature extremes encountered prior to gastrulation, and by loss of HSP90 function.
The authors wish to thank Adam Rodaway for sqtcz35 fish, Sarah Farrington and Nancy Hopkins for sqthi975 fish, Aaron Steiner and Daniel Kessler for the ARE-Luc and control plasmids, and to Michael Sargent for help with the anoxia/hypoxia studies. Thanks also to David Bodine and Nadine Peyrieras for helpful discussions and to Igor Dawid, Yingzi Yang, Shawn Burgess and Scott Dougan for critical reading of the manuscript. This research was supported in part by the Intramural Research Program of the National Human Genome Research Institute, National Institutes of Health.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.