The mammalian central nervous system (CNS) is derived from the dorsal ectoderm of postgastrulation embryos (Liu and Niswander,2005). Initially, the neuroectoderm thickens to form the neural plate. Later, it rolls up to form a closed neural tube in a process called neurulation. During neurulation, lateral cells move toward the midline, resulting in the elongation of the neural plate/tube along the anteroposterior axis of the embryo. This “convergent extension” movement is critical for the closure of the neural tube and it is under the control of an evolutionarily conserved molecular pathway, the planar cell polarity (PCP) pathway.
The PCP pathway was originally described in the fruit fly Drosophila melanogaster (Simons and Mlodzik,2008). In Drosophila, a group of “core” PCP proteins, including Frizzled, Dishevelled, Prickle, Flamingo, Van Gogh, and Diego, regulate cell and tissue polarity in multiple organs including the eyes, wings and bristles on the back (Vinson et al.,1989; Klingensmith et al.,1994; Taylor et al.,1998; Usui et al.,1999; Feiguin et al.,2001). The fly wing is a frequently used system for PCP studies because each cell contains a single actin-based wing hair projecting distally. In Drosophila mutants without normal function of core PCP proteins, the orientations of the wing hairs are altered, but the location of the wing hairs remains peripheral (Wong and Adler,1993). Most core PCP proteins are asymmetrically localized to the proximal or distal side of the wing cells and their proper localization depends on each other (Simons and Mlodzik,2008).
Another group of PCP regulators identified in Drosophila exhibit more limited roles in PCP regulation (Simons and Mlodzik,2008). These “PCP effector proteins,” including Inturned, Fuzzy, Fritz, and Multiple Wing Hairs, affect polarity of the wing cells and dorsal bristles, but not the eyes (Park et al.,1996; Collier and Gubb,1997; Collier et al.,2005; Strutt and Warrington,2008; Yan et al.,2008). In the wings, besides disrupting the normal orientation of the wing hairs, loss of PCP effector function also leads to the formation of additional hairs (Wong and Adler,1993). Genetically, PCP effector genes act downstream of core PCP genes (Wong and Adler,1993). Biochemically, although core-PCP proteins are known to physically associate with each other, there is no evidence that the PCP effector proteins physically associate with each other, or with the core PCP proteins (Simons and Mlodzik,2008).
In vertebrates, mutations in core PCP genes lead to craniorachischisis, an extreme form of neural tube defects (NTD) in which the entire spinal cord and hindbrain fail to close (Simons and Mlodzik,2008). Detailed analysis of homozygous Looptail (Vangl2Lp, a semi-dominant loss-of-function allele of Vangl2; one of the mammalian homologues of Van Gogh) mutants suggested that defective convergent extension cell movement during neurulation might underlie this extensive NTD (Ybot-Gonzalez et al.,2007). Vangl2 shows strong genetic interaction with other PCP regulators, such that removing one copy of Vangl2 on an otherwise normal-looking heterozygous or homozygous mutant for other PCP genes, leads to various degrees of NTD (e.g., Murdoch et al.,2001; Lu et al.,2004).
In contrast to the obvious roles of core PCP proteins in convergent extension, the functions of PCP effector proteins in this process have not been well established. In frogs, morpholino knockdown of Inturned (Xint) and Fuzzy (Xfy) causes a rather mild widening of the neural plate (Park et al.,2006). The mammalian Inturned (Intu) and Fuzzy (Fuz) loss-of-function mutants showed no clear sign of convergent extension defects (Gray et al.,2009; Heydeck et al.,2009; Zeng et al.,2010). Fritz appears to play a major role in convergent extension in frog gastrula but it is required for cell shape change rather than cell polarity determination (Kim et al.,2010).
Following neural tube closure, the CNS is patterned along its dorsal/ventral (D/V) axis with various neuronal and glial cell types arranged in stereotypical locations (Liu and Niswander,2005; Dessaud et al.,2008). At least in the ventral half of the CNS, the patterning is regulated by Sonic Hedgehog (Shh), a member of the Hedgehog family of secreted proteins produced by the notochord and floor plate cells. Studies showed that Shh acts as a morphogen to pattern the ventral CNS, especially the spinal cord, such that the highest level of Shh defines the ventrally located floor plate and intermediate levels of Shh define the ventral–lateral cell types, including the motor neurons and various groups of ventral interneurons (Dessaud et al.,2008).
Through the study of several mouse mutants, we previously showed that primary cilia, cell surface organelles present on most mammalian cells, are required for the normal patterning of the ventral spinal cord (Huangfu et al.,2003; Liu et al.,2005; Hoover et al.,2008). Of interest, the complete loss of cilia leads to the loss of floor plate, V3 interneurons and motor neurons in the spinal cord, whereas a partial loss of cilia leads to milder defects in the ventral patterning of the spinal cord. This ventral spinal cord phenotype, along with the decrease in Shh target gene expression and defects in other tissues such as extra digits in the limb buds, led us to the conclusion that cilia are required for normal Shh signaling. This conclusion was later confirmed by cell biology and genetic studies suggesting that the activation of Smoothened, a critical regulator of the Hh signal transduction, is dependent on cilia (Corbit et al.,2005; Huangfu and Anderson,2005).
Although Intu and Fuz do not exhibit noticeable function in convergent extension and spinal cord closure, our results showed that both proteins do play important roles in cilia formation and ventral spinal cord patterning (Heydeck et al.,2009; Zeng et al.,2010). In both Intu−/− and Fuz−/− mutants, the number and length of cilia are greatly decreased, suggesting that these two genes are both important but not essential for cilia formation. Consistent with the partial loss of cilia, both Intu−/− and Fuz−/− null mutants exhibit almost normal D/V patterning in the anterior spinal cord and loss of floor plate in the posterior spinal cord. This phenotype is relatively mild compared with the complete loss of cilia that leads to the loss of floor plate, V3 interneurons, and a great reduction of motor neurons in the entire spinal cord (Huangfu et al.,2003; Liu et al.,2005).
The surprising lack of convergent extension defects in Intu−/− and Fuz−/− mouse mutants and relatively mild spinal cord patterning defects raised a series of questions. Are the developmental defects, especially the spinal cord patterning defects, in Intu−/− and Fuz−/− mutants solely the result of partial loss of cilia? Are there potential roles of Intu and Fuz in PCP regulation that can only be revealed by removing one or two copies of a core PCP gene? Finally, is functional redundancy between Intu and Fuz responsible for the relatively mild cilia defects and lack of convergent extension defects in either mutant? In the current study, we address these questions through a genetic approach. We show that Intu−/−; Ift88−/− and Fuz−/−; Ift88−/− double homozygous mutants exhibit the same spinal cord patterning defects as Ift88−/− single mutants, suggesting that Intu and Fuz regulate spinal cord patterning through their roles in ciliogenesis. Furthermore, Fuz and Intu do not synergize with Vangl2, a core PCP protein, in either convergent extension or neural tube patterning. Finally, embryos doubly mutated for Intu and Fuz do not exhibit a more severe phenotype than either single mutant, suggesting the lack of functional redundancy between them and arguing against their roles in regulating convergent extension in mammals.
Intu and Fuz Regulate Mouse Morphogenesis and Spinal Cord Patterning Through Ciliogenesis
Homozygous mutant mouse embryos for an apparent Intu null allele die in mid-gestation with multiple patterning defects, including those in the CNS and limbs (Zeng et al.,2010). Ciliogenesis is partially disrupted in these mutant embryos, and our previous studies showed that Intu likely regulates ciliogenesis through an indirect mechanism. To understand the molecular mechanism underlying the roles of Intu in ciliogenesis, especially its relationship with the essential cilia-specific transport mechanism intraflagellar transport (IFT), and to address whether Intu has a cilia-independent role in mouse embryonic development and spinal cord patterning, we generated and characterized Intu; Ift88 double mutants.
Ift88 is one of the core-components of IFT complex B and is essential for ciliogenesis and mouse embryonic development (Murcia et al.,2000; Liu et al.,2005; Jia et al.,2009). Homozygous Ift88−/− null mutant embryos show a twisted body axis, frequent exencephaly and pericardial edema at embryonic day (E) 10.5 (Fig. 1B,C). On a C3H/HeN background, more than half of the Ift88−/− mutant embryos arrest at an earlier stage and appear significantly smaller than their littermates at E10.5 (compare Fig. 1C with A and B). In contrast, most Intu−/− mutants complete neurulation and exhibit a slightly tightened mesencephalic flexure at E10.5 (Fig. 1D). By crossing Ift88+/− and Intu+/− heterozygous mutants, we obtained double heterozygous carrier mice at the expected ratio (Table 1). Intu−/−; Ift88+/− mutant embryos exhibit the same morphological defects as Intu−/− single mutants, suggesting the lack of strong synergistic interaction between these two genes (data not shown). Finally, double homozygous Intu−/−; Ift88−/− mutants exhibit the same spectrum of morphological defects as Ift88−/− mutants at E10.5, suggesting that Intu regulates mouse embryonic development through its important role in cilia formation (Fig. 1E,F).
Table 1. All Double Heterozygous Mutant Pups Were Weaned at the Expected Ratio
# Pups at Weaning
# Pups Expected
Only pups with abnormal (looped or kinked) tails were genotyped.
We then investigated whether the roles of Intu in spinal cord D/V patterning is cilia-dependent. The complete loss of cilia in Ift88−/− mutants results in the loss of the floor plate (Foxa2+ at E10.5) and the progenitors of V3 interneurons (Nkx2.2+; Huangfu et al.,2003; Liu et al.,2005; Fig. 2A,B,E,F; Supp. Fig. S1A,B,E,F, which is available online). Our previous data indicated that motor neurons (Lhx3+ and Isl1+) are expanded to the ventral midline of the spinal cord in the anterior spinal cord and are greatly reduced in the posterior spinal cord in Ift88−/− mutants (Huangfu et al.,2003; Liu et al.,2005; Supp. Fig. S2A,B,E,F,I, J,M,N). This ventral expansion of mature motor neurons results from a ventral expansion of their progenitors expressing a transcription factor Olig2 (Fig. 2I,J; Supp. Fig. S1I,J). The reduction of motor neurons in the posterior spinal cord appears to be a defect in neurogenesis because Olig2-expressing cells are not greatly reduced in the posterior Ift88−/− mutant spinal cord (Fig. 2I,J; Supp. Fig. S1I,J). In contrast, the Intu−/− mutant spinal cord exhibits only mild and variable patterning defects. In the posterior spinal cord, floor plate cells are rarely present and progenitors of V3 interneurons and motor neurons occupy the ventral midline of the spinal cord (Fig. 2C,G,K). The D/V patterning is less disrupted in the anterior part of the Intu−/− mutant spinal cord, with only a slight reduction of the floor plate and occasional ventral expansion of V3 interneurons (Supp. Fig. S1C,G). Pax6, a marker for the progenitors of the dorsal and lateral cell types of the spinal cord including the motor neurons, is ventrally expanded at all axial levels in the Ift88−/− mutant spinal cord (Fig. 2Q,R; Supp. Fig. 1Q,R). In Intu−/− mutants, Pax6 is expanded in the posterior, but not in the anterior part of the spinal cord (Fig. 2S and Supp. Fig. S1S). As we reported previously, the patterning of the more dorsal regions of the spinal cord is not disrupted by the loss of Ift88 or Intu as indicated by the normal expression of Nkx6.1 (labels the floor plate and progenitors for V2, V3 interneurons and motor neurons) and Pax7 (labels progenitors for all dorsal neurons) (Fig. 2M–O,U–W; Supp. Fig. S1M–O,U–W). Intu−/−; Ift88−/− double mutants exhibit the same spinal cord D/V patterning defects as those in Ift88−/− single mutants, including the complete loss of the floor plate and V3 interneurons, as well as the ventral expansion of the motor neurons and their progenitors (Fig. 2D,H,L,T; Supp. Figs. S1D,H,L,T; 2D,H,L,P). The patterns of the more dorsal spinal cord regions remain unchanged (Fig. 2P,X; Supp. Fig. S1P,X). These results indicate that the D/V patterning defects of the Intu−/− mutant spinal cord is likely the consequence of the partial loss of cilia.
We previously reported that an apparent null mutation in Fuz leads to partial disruption of ciliogenesis and multiple patterning defects in the CNS and limbs (Heydeck et al.,2009). We found that Fuz−/−; Ift88−/− double homozygous mutants exhibit the same morphological and patterning defects as Ift88−/− single mutants, suggesting that the function of Fuz in mouse development is also dependent on cilia (data not shown).
Vangl2 Shows No Apparent Synergistic Interaction With Fuz and Intu in Neurulation
Fuz was originally identified in Drosophila as a PCP effector protein that plays essential roles in regulating the polarity of wing cells (Collier and Gubb,1997). Surprisingly, we have not been able to observe any convergent extension defects in Fuz−/− null mouse mutant embryos (Heydeck et al.,2009). The only defect that may be attributed to a weak convergent extension defect was the kinked tail in a late stage embryo carrying a likely hypomorphic allele of Fuz (Gray et al.,2009). We hypothesized that if Fuz indeed plays a minor role in convergent extension, then it may genetically interact with one of the core PCP proteins, Vangl2. Vangl2 is mutated in the semi-dominant mouse mutant Loop tail (Vangl2Lp). Heterozygous Vangl2+/Lp mutant mice exhibit looped tails and a head-bobbing phenotype with incomplete penetrance (Smith and Stein,1962). Homozygous Vangl2Lp/Lp mutants die before birth and exhibit multiple PCP defects including craniorachischisis and randomized hair cell polarity in the inner ears (Montcouquiol et al.,2003; Ybot-Gonzalez et al.,2007). Of interest, Vangl2 synergistically interacts with multiple PCP regulators, leading to much more severe PCP defects when one copy of Vangl2 is removed on a heterozygous or homozygous mutant background for another PCP gene (e.g., Lu et al.,2004; Wang et al.,2006; Qian et al.,2007). We therefore addressed whether removing Vangl2 would reveal any roles of Fuz in regulating convergent extension.
When we crossed mice heterozygous for Fuz and Vangl2Lp, we obtained Fuz+/−; Vangl2+/Lp double heterozygous mutants at the expected ratio (Table 1). Visual inspection of Fuz+/−; Vangl2+/Lp double heterozygotes did not reveal a more severe looped tail or head bobbing phenotype than Vangl2+/Lp heterozygotes alone (data not shown). We subsequently crossed the double heterozygotes to generate Fuz−/−; Vangl2Lp/Lp double homozygous mutant embryos. Vangl2+/Lp heterozygous mutant embryos do not show obvious morphological defects at E9.5 (Fig. 3A,B), but most of them start to show the looped-tail phenotype at E12.5 (Fig. 3F,F′,G,G′). Loss of Fuz does not exacerbate the Vangl2+/Lp heterozygote phenotype, except for one embryo at E12.5 that shows a clearly twisted body axis (Fig. 3C,H,H′). Removing one copy or both copies of Fuz in Vangl2Lp/Lp homozygous mutants results in neural tube defects resulting from an additive effect of simultaneous loss of Fuz and Vangl2 (Fig. 3D,E). Occasionally, the Fuz+/−; Vangl2Lp/Lp and Fuz−/−; Vangl2Lp/Lp double homozygous mutants exhibit a unique kink posterior to the forelimbs (Fig. 3D,E). This phenotype has not been reported in any existing mouse PCP mutant and may not solely result from defective convergent extension. The Fuz−/−; Vangl2Lp/Lp double homozygous mutant embryos occasionally exhibit exencephaly extended into the midbrain and polydactyly in all four limbs at E12.5, consistent with the loss of Fuz function (Fig. 3J,J′). In addition, they invariably exhibit craniorachischisis, a phenotype suggestive of loss of Vangl2 function (Fig. 3I,I′,J,J′). These results appear to suggest against a role of Fuz in mediating convergent extension in cooperation with Vangl2.
To further address the potential synergistic interaction between Vangl2 and Fuz, we examined the transverse sections of the spinal cord at E9.5. At this stage, the expression domains of Foxa2 and Nkx2.2 start to segregate but still partially overlap (Fig. 4A,E). There is a reduction of Foxa2-expressing cells and an increase of Nkx2.2-expressing cells in Fuz−/− homozygous mutants (Fig. 4B,F). Another indication of the ventral patterning defects is the sporadic Olig2-expressing cells in the ventral-most region in the Fuz−/− mutant, but not the wild-type, spinal cords (Fig. 4I,J). Consistent with a defect in convergent extension, the floor plate of the Vangl2Lp/Lp mutant spinal cord is expanded, which can be appreciated through both morphology and an expanded Foxa2 expression domain (Fig. 4C,G,K). However, the overall patterning along the D/V axis is not affected, as indicated by proper expression of Nkx2.2 and Olig2 (Fig. 4G,K). The floor plate in the Fuz−/−; Vangl2Lp/Lp double homozygous mutant spinal cord is significantly narrower than that of Vangl2Lp/Lp single mutants based on both morphology and reduced number of Foxa2-expressing cells (Fig. 4D). This is likely due to a defect in Hh signaling and failure to define the floor plate cell fate, rather than a rescue of the convergent extension defects in Vangl2Lp/Lp mutants. Consistent with this observation, Nkx2.2 and Olig2 expression domains are closer to the ventral midline than in the Vangl2Lp/Lp mutants (Fig. 4H,L). These findings suggest that Fuz does not play a significant role in the PCP pathway, which is essential for convergent extension movement during neurulation.
Vangl2 and Intu exhibit no apparent synergistic interaction in neurulation. Intu−/− homozygous mutants exhibit occasional midbrain exencephaly at E10.5, and removing one copy of Vangl2 does not exacerbate this phenotype (Fig. 5A,B). Removing one copy of Intu in Vangl2Lp/Lp homozygous mutants does not have an impact on the Vangl2Lp/Lp mutant morphology (Fig. 5C). Similar to Fuz−/−; Vangl2Lp/Lp mutants, Intu−/−; Vangl2Lp/Lp double homozygous mutants occasionally exhibit a twisted or kinked body axis, which may not solely result from defects in convergent extension (Fig. 5D). An examination of the transverse sections of the spinal cords reveals a simple additive effect between Vangl2 and Intu. At this stage, Foxa2 labels the floor plate and Nkx2.2 labels the V3 interneurons (Fig. 5E,I). The numbers of both Foxa2 and Nkx2.2-expressing cells are greatly reduced in the ventral midline of the Intu−/−; Vangl2Lp/Lp double homozygous mutant spinal cord (Fig. 5F,J). Olig2 expression domain is expanded to the ventral-most region in the Intu−/−; Vangl2Lp/Lp double homozygous mutant, but not wild-type, spinal cords (Fig. 5M,N). Consistent with a key role of Vangl2 in convergent extension, the floor plate of the Intu+/−; Vangl2Lp/Lp mutant spinal cord is expanded, reflected in both morphology and an expanded Foxa2 expression domain (Fig. 5G,K,O). However, the overall patterning along the D/V axis is not affected, as indicated by proper expression of Nkx2.2 and Olig2 (Fig. 5K,O). The floor plate in the Intu−/−; Vangl2Lp/Lp double homozygous mutant spinal cord is significantly narrower than that of Vangl2Lp/Lp single mutants based on both morphology and reduced number of Foxa2-expressing cells (Fig. 5H). This is likely due to a defect in Hh signaling and failure to define the floor plate cell fate, rather than a rescue of the convergent extension defects in Vangl2Lp/Lp mutants. Consistent with this observation, Nkx2.2 and Olig2 expression domains are present in the ventral midline of the Intu−/−; Vangl2Lp/Lp double homozygous mutant spinal cord (Fig. 5L,P). These findings suggest that Intu does not play a significant role in convergent extension.
Inturned and Fuzzy Do Not Play Redundant Roles in Convergent Extension and CNS Patterning
The lack of severe convergent extension defects and relatively mild ciliogenic defects in Intu−/− and Fuz−/− mutants raises the possibility that these two PCP effector proteins may have redundant functions (Heydeck et al.,2009; Zeng et al.,2010). To address this possibility, we generated Intu; Fuz double mutants. The double heterozygous mutant mice are healthy and fertile (Table 1). We subsequently characterized E10.5 Intu−/−; Fuz−/− double homozygous mutants and found that they are of normal size, with a slightly tight mesencephalic flexure, and morphologically indistinguishable from Intu−/− and Fuz−/− single mutants (Fig. 6). The apparent lack of craniorachischisis indicative of abnormal convergent extension movement in the Intu−/−; Fuz−/− double homozygous mutant embryos suggests that these PCP effector protein homologues do not play redundant roles in convergent extension.
We then investigated the relationship between Intu and Fuz in the D/V patterning of the spinal cord. In the posterior spinal cord, the floor plate is significantly reduced in both Intu−/− and Fuz−/− mutants, and V3 interneurons and motor neurons are expanded to the ventral midline (Fig. 7B,C,F,G,J,K). The more dorsal regions of the spinal cord are not altered in either mutant (Fig. 7N,O,R,S,V,W). The D/V patterning of the posterior spinal cord in Intu−/−; Fuz−/− double homozygous mutants is the same as that of either single mutant (Fig. 7D,H,L,P,T,X). In the anterior spinal cord, all ventral cell types are present in Intu−/− and Fuz−/− mutants, with only the floor plate slightly reduced in size (Supp. Fig. 3). The patterning defects in Intu−/−; Fuz−/− double homozygous mutants are similar to those of Intu−/− and Fuz−/− single mutants in the anterior spinal cord (Supp. Fig. S3). The spatial arrangement of postmitotic motor neurons and V2 interneurons is also similar in Intu−/− and Fuz−/− single mutants, as well as in Intu−/−; Fuz−/− double homozygous mutants (Supp. Fig. S4). The lack of more severe morphological and patterning defects in Intu−/−; Fuz−/− double homozygous mutants suggests that Intu and Fuz do not share redundant functions in either ciliogenesis or convergent extension during early neural tube development.
In Drosophila, the PCP effector genes were defined genetically as tissue polarity regulators downstream of the core-PCP genes (Simons and Mlodzik,2008). The molecular mechanisms underlying the function of these PCP effector proteins remain elusive and whether they are involved in cell polarity control in vertebrates has not been adequately addressed. Our previous work showed that mouse mutants for two PCP effector gene homologues, Intu and Fuz, exhibit defects in ciliogenesis, but not in convergent extension (Heydeck et al.,2009; Zeng et al.,2010). In the current study, we took a genetic approach to show that both Intu and Fuz regulate mouse embryonic development and spinal cord patterning through their important roles in cilia formation. Furthermore, we show that loss of mouse Fuz or Intu does not exacerbate the convergent extension defects in Vangl2Lp heterozygous and homozygous mutant embryos. Finally, the Intu−/−; Fuz−/− double homozygous mutant embryos exhibit the same spinal cord patterning defects, and lack of convergent extension defects, as Intu−/− and Fuz−/− single mutants, suggesting that these two proteins play nonredundant and nonsynergistic roles in mammalian development.
Both Intu−/− and Fuz−/− homozygous mutant embryos exhibit multiple developmental defects, including polydactyly and ventral spinal cord patterning defects (Gray et al.,2009; Heydeck et al.,2009; Zeng et al.,2010). The similarities between these mutants and previously reported cilia mutants, as well as the findings that cilia are partially lost in these mutants suggest that these developmental defects are likely secondary to abnormal cilia formation. However, Inturned and Fuzzy have been shown in Drosophila to regulate wing cell polarity, and studies in frogs suggested their involvement in convergent extension and vesicular trafficking (Adler et al.,1994; Collier and Gubb,1997; Park et al.,2006; Gray et al.,2009). Therefore, it is possible that the difference between the spinal cord patterning in these PCP effector mutants and mutants with complete loss of cilia, such as Ift88−/− mutants, could arise from nonciliary functions of Intu and Fuz. We analyzed Intu−/−; Ift88−/− and Fuz−/−; Ift88−/− double homozygous mutants and found that the double mutants exhibit the same phenotype as Ift88−/− single mutants, strongly suggesting that Intu and Fuz regulate mouse spinal cord patterning mainly, if not exclusively, through their roles in cilia formation.
Intu−/− and Fuz−/− mutants both have defects in cilia formation, but neither exhibits the complete loss of cilia (Heydeck et al.,2009; Zeng et al.,2010). The relatively mild ciliogenic defects in these mutants could be the result of residual protein function. However, our Intu mutant allele contains an N-terminal deletion including part of the PDZ domain and a frame shift, while our Fuz mutant allele contains a deletion including two-thirds of the Fuz locus. Therefore, both alleles appear to be null alleles. Another possibility is that Intu and Fuz play partially overlapping roles in cilia formation and can partially compensate for the loss of each other. Our characterization of Intu−/−; Fuz−/− double homozygous mutants shows that the simultaneous removal of both genes leads to the same degree of D/V patterning defects in the spinal cords as in Intu−/− and Fuz−/− single mutants, suggesting that these two proteins do not compensate for each other. One likely scenario is that these two proteins act at two distinct steps of the same cellular pathway such that the mutation of both genes leads to the same effect as the mutation of either gene individually. In support of this view, recent cell biological studies in frogs have indicated that Intu and Fuz regulate ciliogenesis at distinct steps (Park et al.,2008; Gray et al.,2009).
Finally, it was surprising that neither Intu−/− nor Fuz−/− mutant embryos exhibit craniorachischisis, a hallmark of defects in PCP-dependent convergent extension movement. It was reported that a likely hypomorphic mutant allele for Fuz exhibits a kinked tail, but it is not clear whether this is a result of defective convergent extension (Gray et al.,2009). We reasoned that, if mammalian Intu and Fuz do play a minor role in convergent extension, removing these genes together with one or both copies of Vangl2 would lead to more severe convergent extension defects than removing Vangl2 alone. In contrast, our analyses of Fuz−/−; Vangl2+/Lp, Intu−/−; Vangl2+/Lp, Fuz−/−; Vangl2Lp/Lp and Intu−/−; Vangl2Lp/Lp mutants, respectively, showed simple additive effects and did not reveal obvious signs of synergistic interactions between Vangl2 and these two PCP effector genes in convergent extension. We conclude that mammalian Fuz and Intu do not play a direct role in convergent extension. We cannot rule out a possible late, tissue-specific role for Fuz or Intu in cell polarity control, such as in the inner ears.
Our data also suggest that the lack of convergent extension defects in Intu−/− and Fuz−/− null mutants are not due to their redundant roles in PCP regulation. We never observed NTD in the posterior hindbrain and spinal cord of Intu−/−; Fuz−/− double homozygous mutants, and morphology of the brain and spinal cord shows no sign of convergent extension defects. These results further argue against a direct role of Intu and Fuz in convergent extension.
In conclusion, through a genetic approach we show that Intu and Fuz regulate mammalian development by regulating cilia formation. We also rule out the possibility of a redundant function between these two PCP effector proteins in either cilia formation or convergent extension cell movement. In addition, Fuz and Intu do not show obvious synergistic interaction with Vangl2. Our results strongly argue that Intu and Fuz play divergent roles in mammals compared with their invertebrate homologues.
IntuΔex2 (Zeng et al.,2010), FuzGT (Heydeck et al.,2009), and Ift88Δ2,3βGal (Murcia et al.,2000) mutants were genotyped as previously described. Vangl2Lp-m1Jus (Kibar et al.,2001) mice were obtained from MMRRC and genotyped using simple sequence length polymorphism markers D1mit105 and D1mit115. All mice were kept on a C3H/HeN (Charles River Lab) background.
E10.5 mouse embryos were fixed in fresh 4% paraformaldehyde for 1 hr and processed for cryosection at 10 μm. Sections were incubated with primary antibodies diluted in phosphate-buffered saline supplemented with 1% goat serum and 0.1% Triton X-100 overnight at 4°C, followed with 1-hr incubation with secondary antibodies diluted in the same solution at room temperature. Sections are visualized using a Nikon E600 fluorescence microscope and photos were taken with a QImaging Micropublisher digital camera. Antibodies against Shh (5E1, 1:100 dilution), Foxa2 (4C7, 1:100 dilution), Nkx2.2 (74.5A5, 1:40 dilution), Nkx6.1 (F55A12, 1:500 dilution), Pax6 (PAX6, 1:1,000 dilution), and Pax7 (PAX7, 1:1,000 dilution) were obtained from Developmental Studies Hybridoma Bank. The antibody against Olig2 (AB9610, 1:1,000 dilution) was purchased from Millipore. Cy3-conjugated goat anti-mouse IgG (115-166-003, used at 2.5 μg/ml) and goat anti-rabbit IgG (111-165-003, used at 2.5 μg/ml) were purchased from Jackson ImmunoResearch.
We thank Jinling Liu for critically reading the manuscript. We thank Dr. Noel Murcia for Ift88 mutant mice. The monoclonal antibodies against Shh, Nkx2.2, Isl1, Pax6, and Pax7 developed by Dr. Jessell, and the one against Nkx6.1 developed by Dr. Madsen, were obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by The University of Iowa, Department of Biological Sciences, Iowa City. A.L. is supported by a Young Investigator Developmental Award from PKD foundation and a Penn State new lab start-up fund.