A tissue-specific role for intraflagellar transport genes during craniofacial development

Primary cilia are nearly ubiquitous, cellular projections that function to transduce molecular signals during development. Loss of functional primary cilia has a particularly profound effect on the developing craniofacial complex, causing several anomalies including craniosynostosis, micrognathia, midfacial dysplasia, cleft lip/palate and oral/dental defects. Development of the craniofacial complex is an intricate process that requires interactions between several different tissues including neural crest cells, neuroectoderm and surface ectoderm. To understand the tissue-specific requirements for primary cilia during craniofacial development we conditionally deleted three separate intraflagellar transport genes, Kif3a, Ift88 and Ttc21b with three distinct drivers, Wnt1-Cre, Crect and AP2-Cre which drive recombination in neural crest, surface ectoderm alone, and neural crest, surface ectoderm and neuroectoderm, respectively. We found that tissue-specific conditional loss of ciliary genes with different functions produces profoundly different facial phenotypes. Furthermore, analysis of basic cellular behaviors in these mutants suggests that loss of primary cilia in a distinct tissue has unique effects on development of adjacent tissues. Together, these data suggest specific spatiotemporal roles for intraflagellar transport genes and the primary cilium during craniofacial development.


Introduction
Primary cilia are ubiquitous, microtubule-based extensions that protrude off a plethora of cell types throughout development. Interest in primary cilia biology has grown exponentially over the last decade, mostly due to the identification of ciliopathies, a growing class of human syndromes that occur as a result of aberrant cilia function [1]. Although there is no established phenotypic criterion for diagnosis of a ciliopathy, it has been hypothesized that a ciliopathy a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 recombines within the dorsal neural tube giving rise to neural crest cells (NCCs) and a portion of dorsal neuroectoderm in the developing midbrain (Fig 1A and 1B) [19][20][21]. Neural crest cells make up the majority of the cranial mesenchyme and make numerous contributions to the craniofacial complex, most notably to the facial mesenchyme and skeletal elements. Second, we utilized the Crect driver [22] to target recombination to cells within the surface ectoderm (Fig 1C and 1D). The developing surface ectoderm houses many signaling centers that are important for directing craniofacial development such as the frontonasal ectodermal zone (FEZ) and the nasal pits [23][24][25][26][27]. The Crect driver recombines in the surface ectoderm of the developing face (n = 17; 53%); however, we also observed a less defined recombination pattern (n = 15; 47%; S1A,B). Finally, we implemented the AP2-Cre driver [28]. AP2-Cre-mediated recombination occurs in NCCs, surface ectoderm and neuroectoderm (Fig 1E and 1F). The neuroectoderm, particularly the forebrain, serves as the scaffold upon which the face develops [29]. In addition to physically supporting facial development, the neuroectoderm also serves as an important signaling center, supplying the face with essential molecular inputs that help to guide midfacial development [23,30]. Temporal onset and spatial domains of recombination for all three drivers used are summarized in Table 1. To confirm the efficacy and specificity of all three drivers we carried out immunostaining at both e11.5 and e14.5 for the ciliary marker Arl13b on Kif3a f/f ;Wnt1-Cre, Kif3a f/f ;Crect and Kif3a f/f ;AP2-Cre embryos and found cilia were absent from neural crest alone, surface ectoderm alone and both neural crest and surface ectoderm, respectively (S2 Fig). To address the role of individual ciliary proteins and the cilia in each of these tissues, we next conditionally ablated three distinct IFT ciliary genes (Kif3a, Ift88 and Ttc21b) and analyzed the resulting facial phenotype.
Loss of Kif3a in tissue-specific domains of the craniofacial complex generates a range of phenotypes KIF3A is a member of the kinesin superfamily and functions as an anterograde IFT protein [31]. To examine the role of Kif3a in tissues contributing to the craniofacial complex we conditionally excised Kif3a using each of the three drivers detailed above and examined craniofacial domains frequently affected in ciliopathies. Our previous work identified a widened midline, as determined by an increased distance between the nasal pits (internasal distance), as the distinguishing feature of Kif3a f/f ;Wnt1-Cre embryos [6,17]. As expected, the first obvious phenotype in Kif3a f/f ;Wnt1-Cre embryos at e11.5 was a significant increase in the internasal distance (n = 5), relative to wild-type embryos (n = 28) (Fig 2A, 2B and 2Y). In contrast, Kif3a f/f ;Crect embryos, in which Kif3a was lost in the surface ectoderm, did not have a significant difference in internasal distance when compared to wild-type embryos (Fig 2C and 2Y, n = 10). In addition to the Crect recombination pattern reported in Fig 1, we also observed Kif3a f/f ;Crect mutants with the alternate, broader pattern of recombination (S1C and S1D Fig). Regardless of which recombination pattern was present, the craniofacial phenotypes generated were relatively similar (S1C-S1F Fig). In Kif3a f/f ;AP2-Cre embryos, in which Kif3a was lost in NCCs, surface ectoderm and some neuroectoderm, a significant increase in the internasal distance and medially rotated nasal pits were observed (Fig 2D and 2Y).
We continued our analysis of each mutant at e11.5 to determine if loss of Kif3a in different tissues and tissue combinations had an effect on cell differentiation, cell proliferation or cell death. We first examined the earliest stages of cell differentiation and formation of the skeletal condensations by performing peanut agglutinin (PNA) immunostaining. We observed domains of PNA staining in Kif3a f/f ;Wnt1-Cre mutants that were laterally shifted, relative to wild-type embryos (n = 3) (Fig 2E and 2F). Similar to the pattern observed in Kif3a f/f ;Wnt1-Cre mutants, a lateral displacement of the early condensations was observed in Kif3a f/f ;Crect (n = 3) and Kif3a f/f ;AP2-Cre (n = 3) mutants (Fig 2G and 2H). Thus, despite a shifted domain, the process of differentiation did not appear to be impaired in any of the ciliary mutants.
To determine if loss of Kif3a in various tissues of the craniofacial complex impacted cell proliferation and cell death, we performed immunostaining for phosphohistone H3 (PHH3) and cleaved caspase 3 (CC3) in the developing frontonasal prominence and palate at e11.5 (see areas analyzed for quantification in S3A and S3B Fig). We found that loss of Kif3a had tissue specific effects on cell proliferation and cell death ( Table 2, S1 Table). Loss of Kif3a within NCCs of the frontonasal prominence caused a significant decrease in cell proliferation, whereas loss of Kif3a in surface ectoderm or a combination of NCCs, surface ectoderm and neuroectoderm did not cause a significant change in proliferation relative to wild-type embryos (Fig 2I-2L and 2Z). Furthermore, loss of Kif3a in the surface ectoderm significantly increased the amount of cell death within the frontonasal prominence, whereas cell death was not significantly impacted when Kif3a was lost in NCCs alone (Wnt1-Cre), or a combination of tissues (AP2-Cre) (Fig 2M-2P and 2AA). In the developing palatal shelves, loss of Kif3a did not cause a statistically significant change in cell proliferation (Fig 2Q-2T and 2BB); however, a significant increase in cell death was observed within the palatal shelves of Kif3a f/f ;Crect and Kif3a f/f ;AP2-Cre mutants (Fig 2U-2X and 2CC). Thus, taken together, these results suggest a tissue specific function for Kif3a and the cilium in the developing craniofacial complex.
We continued our analysis of these tissue-specific mutants at e14.5. Again, the most striking phenotype of Kif3a f/f ;Wnt1-Cre embryos was the severe midline widening (Fig 3A, 3B and 3CC; n = 5). The midline of Kif3a f/f ;Crect embryos was also significantly wider, highly dysmorphic and featured numerous tissue nodules (Fig 3C and 3CC; n = 9). Similarly, Kif3a f/f ; AP2-Cre embryos had a significantly widened midline with a dysmorphic frontonasal prominence (Fig 3D and 3CC; n = 7). Despite an overall significant increase of internasal width, we observed some variability in the severity of the midfacial phenotype in Kif3a f/f ;AP2-Cre embryos (S4A and S4B Fig). All three mutants also presented with cleft palate (Fig 3E-3H). We also examined the development of the mandibular prominence within these three mutants. Consistent with our previous reports, Kif3a f/f ;Wnt1-Cre embryos had micrognathia (undersized jaw) and aglossia (no tongue) (S5A and S5B Fig). In contrast, the tongue was clearly present in Kif3a f/f ;Crect embryos, yet there were tissue hyperplasias similar to those seen on the developing frontonasal prominence (S5C Fig). The developing mandible of Kif3a f/f ;AP2-Cre embryos resembled that of the Kif3a f/f ;Wnt1-Cre embryos, presenting with aglossia and mild micrognathia (S5D Fig). Histologically, the midfacial widening in Kif3a f/f ;Wnt1-Cre embryos was apparent by the presence of a duplicated of the nasal septum (Fig 3I and 3J) [6]. The nasal septum is a cartilaginous structure derived from NCCs that occupy the frontonasal prominence. Both Kif3a f/f ;Crect and Kif3a f/f ;AP2-Cre embryos presented with a duplicated nasal septum (Fig 3K and 3L), yet with varying degrees of penetrance and severity (Table 3 and S4D and S4E Fig). We next analyzed each mutant to determine if cell proliferation and cell death were aberrant in either the developing frontonasal prominence or palate (see areas analyzed for quantification in S3C and S3D Fig). Cell proliferation was significantly increased in the developing frontonasal prominence in all three mutants (Fig 3M, 3O, 3Q, 3S and 3DD; Table 3); however, we also observed a significant increase in cell death in Kif3a f/f ;Crect mutants ( Fig 3N, 3P, 3R, 3T and 3EE). We next examined how loss of Kif3a affected palatal development at e14.5 in all three mutants. Neither cell proliferation nor cell death was significantly altered in the palate of Kif3a fl/fl ;  Wnt1-Cre mutants (Fig 3U, 3V, 3Y, 3Z, 3FF and 3GG). Cell proliferation and cell death were, however; significantly increased in Kif3a f/f ;Crect and Kif3a f/f ;AP2-Cre mutants (Fig 3W, 3X, 3AA, 3BB, 3FF and 3GG). Taken as a whole, these data again suggest that Kif3a and cilia in different craniofacial tissues have distinct roles in regulating craniofacial development, as each Cre-driver resulted in a unique phenotype and alterations in cellular behaviors. We next set out to determine if these phenotypes were specific to Kif3a.
Loss of Ift88 in tissue-specific domains of the craniofacial complex phenocopies Kif3a mutants IFT88 is another anterograde IFT protein essential for ciliogenesis [32]. To determine if the craniofacial phenotypes we observed with ablation of Kif3a were specific to Kif3a itself, or due to impaired ciliogenesis, we repeated our previous approach and conditionally knocked out Ift88 with the same drivers used to conditionally ablate Kif3a. As mentioned above, the first distinguishing feature of Kif3a f/f ;Wnt1-Cre embryos was a widened midline at e11.5, as determined by internasal distance [6,33]. Ift88 f/f ;Wnt1-Cre embryos at e11.5 also had a significant increase in the internasal distance ( Fig 4A, 4B and 4Y; n = 5), albeit slightly less severe than   While the overall morphology of the mutants appeared extremely similar, we again examined cell differentiation, cell proliferation and cell death of Ift88 tissue-specific mutants. We examined the earliest stages of cell differentiation and found that, similar to Kif3a mutants, Ift88 mutants contained altered domains of PNA positive cells (Fig 4E-4H). We next examined cell proliferation and cell death. Loss of Ift88 within NCCs (Wnt1-Cre) of the frontonasal prominence caused a significant increase in cell proliferation, whereas loss of Ift88 in surface ectoderm (Crect) or a combination of NCCs, surface ectoderm and neuroectoderm (AP2-Cre) did not cause a significant change in proliferation relative to wild-type embryos at e11.5 ( Fig  4I-4L and 4Z). Furthermore, the loss of Ift88 in surface ectoderm (Crect) significantly increased the amount of cell death within the frontonasal prominence, whereas cell death was     The characteristic duplication of the nasal septum was again present in the Ift88 f/f ;Wnt1-Cre embryos (Fig 5I and 5J). We did not observe a duplicated nasal septum in Ift88a f/f ;Crect embryos ( Fig 5K), whereas Ift88 f/f ;AP2-Cre embryos did have a duplicated septum (Fig 5L). We again examined cell proliferation and cell death within this area. Within the frontonasal prominence, cell proliferation was significantly increased in Ift88 f/f ;Wnt1-Cre embryos ( Fig  5M, 5O and 5DD; Table 5), similar to that observed in Kif3a f/f ;Wnt1-Cre embryos. Conversely, we observed Ift88 f/f ;Crect and Ift88 f/f ;AP2-Cre embryos had a slight, but significant decrease in proliferation within the frontonasal prominence (Fig 5Q, 5S and 5DD). With respect to cell death, we observed a significant increase in CC3-postive cells in both Ift88 f/f ;Wnt1-Cre and Ift88 fl/fl ;Crect embryos ( Fig 5N, 5P, 5R and 5EE). No significant change between the number of CC3-positive cells was detected in Ift88 fl/fl ;AP2-Cre embryos relative to control embryos ( Fig  5T and 5EE). Interestingly, there was no significant change in cell proliferation or cell death in any of the Ift88 mutants within the developing palate at e14.5 (Fig 5U-5BB, 5FF and 5GG). In sum, the gross craniofacial phenotypes between Kif3a and Ift88 mutants were relatively conserved, despite some differences in cell behaviors in affected areas. Together these data further supported the hypothesis that the cilium has distinct roles in individual tissues of the craniofacial complex, while additionally suggesting that ciliary proteins themselves may also have specific functions within each tissue.
Loss of Ttc21b in tissue-specific domains of the craniofacial complex does not phenocopy Kif3a or Ift88 mutants KIF3A and IFT88 are both anterograde intraflagellar transport proteins that function in the IFT-B complex to facilitate the transport of molecular cargo from the cell body to ciliary tip. TTC21B (also known as Ift139 and Thm1) is a retrograde intraflagellar transport protein that functions in the IFT-A complex in the retrograde transport of molecular cargo from the ciliary tip to the cell body [13]. To determine if ciliary proteins that function in distinct areas of the cilium affect craniofacial development differentially, we repeated our experimental strategy with Ttc21b f/aln mice, which have one floxed allele and one allele that contains the alien mutation, a null allele of Ttc21b [13].
Kif3a f/f ;Wnt1-Cre and Ift88 f/f ;Wnt1-Cre embryos at e11.5 have significant midfacial defects characterized by an increase in the internasal distance (Figs 2 & 4). In contrast, Ttc21b f/aln ; Wnt1-Cre embryos do not have a significant difference in internasal distance when compared to wild-type embryos (Fig 6A, 6B and 6Y; n = 4). Similar to Kif3a f/f ;Crect and Ift88 f/f ;Crect mutants, Ttc21b f/aln ;Crect embryos did not have a significantly wider internasal distance, yet their nasal pits were patent due to a failure of fusion between the frontonasal, lateral nasal and maxillary prominences (Fig 6C and 6Y; n = 5). Ttc21b f/aln ;AP2-Cre embryos, in which neural crest, surface ectoderm and some neuroectoderm were affected, also did not display midfacial widening (Fig 6D and 6Y; n = 4). These results again supported a hypothesis that the cilium plays tissue specific roles in craniofacial development. Furthermore, the observation of distinctly different phenotypes occurring with the same drivers suggested that individual ciliary genes have a unique function in each tissue.
We next analyzed each mutant to determine if loss of Ttc21b in different tissues and tissue combinations had an effect on cell differentiation, cell proliferation and cell death. Once again, differentiation of early skeletal condensations was examined via PNA staining. Despite the pattern of the early condensations being different, PNA positive domains were still detected in all mutants (Fig 6E-6H). To determine if cellular processes including cell proliferation and cell death were altered in these mutants, we examined PHH3 and CC3 staining in the developing frontonasal prominence and palate. Cell proliferation was significantly reduced in the frontonasal prominence of Ttc21b f/aln ;Wnt1-Cre mutants, yet there was no significant change in proliferation within Ttc21b f/aln ;Crect or Ttc21b f/aln ;AP2-Cre mutants (Fig 6I-6L and 6Z; Table 6). Cell death was also altered in a tissue-specific manner. There was a significant reduction in cell death within the frontonasal prominence of Ttc21b f/aln ;Wnt1-Cre mutants, a significant  increase in cell death in Ttc21b f/aln ;Crect mutants and no change cell death in Ttc21b f/aln ; AP2-Cre mutants relative to wild-type embryos (Fig 6M-6P and 6AA). There were also significant changes in cell proliferation and cell death within the developing palate. Loss of Ttc21b in NCCs (Wnt1-Cre) caused a significant reduction in cell proliferation, loss of Ttc21b in surface ectoderm (Crect) caused a significant increase in cell proliferation, and loss of Ttc21b in NCCs, surface ectoderm and neuroectoderm (AP2-Cre) caused a significant reduction in proliferation ( Fig 6Q-6T and 6BB). Finally, we examined how loss of Ttc21b in various tissues affected cell death within the developing palate. Whereas loss of Ttc21b in NCCs (Wnt1-Cre) or a combination of NCCs, surface ectoderm and neuroectoderm (AP2-Cre) had no effect on cell death in the developing palate, Ttc21b f/aln ;Crect embryos had a significant increase in cell death relative to wild-type controls (Fig 6U-6X and 6CC). We continued our analysis of these tissue-specific mutants at e14.5. In contrast to the striking midfacial phenotype of both Kif3a f/f ;Wnt1-Cre, and Ift88 f/f ;Wnt1-Cre, there was no measurable midfacial defect in Ttc21b f/aln ;Wnt1-Cre embryos (Fig 7A, 7B and 7CC; n = 4). The frontonasal prominence-derived midline of Ttc21b f/aln ;Crect embryos was dysmorphic, but not significantly wider than that of wild-type embryos (Fig 7C and 7CC; n = 6). The Ttc21b f/aln ;   AP2-Cre frontonasal prominence appeared morphologically normal, yet measured significantly wider than wild-types (Fig 7D and 7CC, n = 4). We next examined the development of the palate in these mutants. The palate of both Ttc21b f/aln ;Wnt1-Cre and Ttc21b f/aln ;Crect were cleft, however Ttc21b f/aln ;Wnt1-Cre palatal shelves appeared to be elevated and patent due to either developmental delay or palatal insufficiency, whereas the palatal shelves in the Ttc21b f/ aln ;Crect were hypoplastic and dysmorphic (Fig 7E-7G). The palate of Ttc21b f/aln ;AP2-Cre embryos appeared normal (Fig 7H).  (Fig 7I-7L). Thus, Ttc21b associated craniofacial phenotypes were less severe than those observed in Kif3a and Ift88 ablations.

Discussion
Ciliopathies are a broad class of diseases that affect various cells and tissues throughout the body. A review by Irigoin and Badano suggested that to fully understand both the biology of cilia and the pathology that arises when they are defective, the organelle must be examined both at different time-points during development and on different cell types [34]. Herein, we addressed this suggestion and evaluated the craniofacial (and neural; see accompanying manuscript by Snedeker et al.) phenotypes that arise when three different ciliary genes were conditionally deleted in various tissues of the craniofacial complex (Table 8). We observed that both tissue and gene identity contributed to the phenotypes produced in the developing face. These findings pose several interesting questions related to the role of cilia and ciliary proteins during development of the face and brain.

Severity of phenotype does not linearly correlate to the combination of tissues affected
We set up our experimental design to examine how loss of ciliary function would impact craniofacial development when it occurred in either NCCs (Wnt1-Cre), surface ectoderm (Crect), or a combination of NCCs, surface ectoderm and neuroectoderm (AP2-Cre). We had originally hypothesized that the resulting phenotype from AP2-Cre embryos would be the combination of those phenotypes observed with the Wnt1-Cre and Crect drivers. Interestingly, conditional ablation with AP2-Cre, did not consistently have a combinatorial or more severe phenotype than mutants created with Wnt1-Cre or Crect. There are several explanations for this finding. First, many events in craniofacial development occur as a result of sequential tissue-tissue interactions. For example, signaling centers in the surface ectoderm and neuroectoderm signal to adjacent NCCs during craniofacial development [25,27,35]. Thus, loss of cilia in the surface ectoderm or neuroectoderm could affect the recombined tissue itself (autonomous) by altering key signaling centers, and in turn, this could affect the adjacent NCCs (nonautonomous). However, when adjacent tissues lose cilia, as with the AP2-Cre driver, tissue-tissue signaling is globally disrupted. This could potentially alleviate some phenotypic presentations as aberrant signals are unable to be received by the adjacent tissue also lacking cilia. Second, the observation that loss of cilia in multiple tissues does not necessary correlate with a more severe phenotype could also be accounted for by the fact that some tissues may utilize cilia to a greater extent than others, thus generating epistatic and hypostatic tissues. Finally, it is possible that the timing of recombination in NCCs and surface ectoderm within the AP2-Cre driver occurs slightly later than that in the Wnt1-Cre or Crect, respectively, thus allowing for some important signaling to occur without incidence. Further molecular analyses must be performed to elucidate how each tissue interprets the loss of functional cilia. Conditional knockout of IFT-B genes results in more severe phenotypes than IFT-A genes Intraflagellar transport (IFT) is a cellular process in which molecular motors transport IFT particles (A and B) and cellular cargo along microtubules. In a ciliary context, IFT is essential for ciliogenesis as it transports tubulin subunits to the tip of growing cilia [36][37][38]. Within the cilium, IFT-B particles are moved from base-to-tip (anterograde transport) via kinesin-2 motors, whereas IFT-A particles are moved from tip-to-base (retrograde transport) via dynein motors [38][39][40][41]. KIF3A is a kinesin-2 motor protein that forms a heterotrimeric complex that is essential for anterograde transport. Furthermore, it is believed to be involved in other cellular processes, including neuronal transport, melanosome movement, and secretory pathway transport [42]. Loss of Kif3a results in the complete loss of the axoneme [43]. IFT88 is a member of the IFT-B complex, which also carries cargo in an anterograde direction. Loss of Ift88 produces truncated cilia in which the axoneme extends just beyond the transition zone [32]. Despite being separate proteins, both KIF3A and IFT88 are essential for ciliogenesis and anterograde IFT. The similarities in their role within the cilium likely account for the similar phenotypes generated when they are conditionally deleted out of various cell types and tissues. Our analyses herein and from our previous work [17]; however, consistently observed that Kif3a conditional mutants generated slightly more severe phenotypes (internasal width and degree of nasal septum separation) both phenotypically (Figs 3 and 5) and molecularly [33]. We surmise that the increased severity of phenotypes generated via the loss of Kif3a is due to a TTC21B (also known as Ift139 and Thm1) is an IFT-A protein that participates in retrograde transport. The aln mutation in Ttc21b, which produces a Ttc21b-null mutant, generates shorter, wider cilia that have a bulb-like structure at their distal tips [13]. Despite being structurally aberrant, these cilia are not as functionally compromised as those generated via the loss of IFT-B components Kif3a and Ift88, and thus are likely able to carry out more ciliary function. These findings are consistent with the less severe phenotypes generated in Ttc21b f/aln mutants (Figs 6 and 7). Currently, there is no 'characteristic' phenotype used to diagnose a craniofacial ciliopathy.
However, it is possible that differences in rate of protein degradation between KIF3A, IFT88 and TTC21B following Cre-recombination could also contribute to the variable phenotypes. To definitively test this hypothesis, reliable and robust antibodies for all three proteins would be necessary and protein turn-over assays would have to be performed in each individual tissue. Given that we have documented the loss of cilia after recombination (S2 Fig) [17], we speculate that the phenotypic difference observed are most likely due to the degree to which the cilium is compromised in each mutant. Determining if there is a characteristic phenotype generated depending upon which component of the cilium is compromised (e.g., basal bodies, transition zone or axoneme) could greatly assist in disease diagnosis and therapeutic approaches.
Loss of ciliary proteins affects various signaling pathways in distinct ways. Recently, determining the role for the cilium in coordinated signal transduction has dominated research within the field. A plethora of studies have examined how cilia contribute to the signaling of various molecular pathways including Hedgehog, Wnt, PDGF, etc., [3,5,41,44]. For some pathways, receptors are preferentially localized to the ciliary membrane [45][46][47]. For other pathways, loss of the cilium disrupts the transduction or activity of the pathway itself [13][14][15][48][49][50][51][52][53]. Several of the phenotypes observed in ciliary mutants resemble phenotypes generated when the above mentioned signaling pathways are impaired. A gain of Shh activity [54], loss of Wnt activity [55] or a loss of PDGF [56] activity all produce some degree of midfacial widening, similar to that observed in several of the mutants generated in this study. In light of the established role of cilia in the transduction of multiple signaling pathways as well as the similarities in the phenotypes produced when either the cilia or the signaling pathway is impaired, it is likely that the molecular basis for the phenotypes reported herein are due to a pleiotropic effect on several signaling pathways. Understanding precisely how the cilium transduces these signals, as well as the role of each signal in individual tissues, will be extremely valuable in assessing the basis for ciliopathic phenotypes. Furthermore, determining if specific ciliary proteins have a greater impact on the transduction of some signaling pathways versus others would be of great interest. If this hypothesis were proven true, then targeting the protein's function, independent of the role in ciliogenesis, would allow for defined manipulation of molecular signaling without impacting the cilium as an organelle. Together, studies such as these could provide new avenues of therapeutic intervention for ciliopathies.

Are all cilia created equal?
Cilia are frequently referred to as ubiquitous organelles and thought to be highly conserved throughout not only the embryo, but also among various species. Although there is a high degree of conservation, there are specialized cilia within the body including those within the inner ear, the olfactory epithelium, and the retina [57]. What makes these cilia specialized is the cadre of ciliary genes expressed within the cells and tissues they arise from. Thus, for all other cilia within the body, it would be expected that their conserved and ubiquitous nature would be accompanied by conserved and ubiquitous expression of the majority of ciliary genes. Despite this being the dogma of the ciliary field, a significant number of studies report distinct expression patterns for ciliary genes. Kif3a is predominantly expressed in brain, although trace amounts of Kif3a transcript are detected in various tissues [31]. Ift88 is most robustly expressed in testis, brain, kidney, lung and pancreas [58,59]. In contrast, in other vital organs, such as the heart, spleen, and liver, Ift88 expression is nearly undetectable [58]. In murine embryos, Ttc21b is broadly expressed at e6.5 and e7.5 [60]. At e8.5 it maintains a broad expression pattern with more robust levels of expression in the more posterior neural tube and somites. At e9.5-10.5 Ttc21b expression can be detected in a number of tissues, but most significantly in limbs, eyes and dorsal neural tube (See accompanying manuscript Snedeker et al.,). Thus, the expression patterns of these three ciliary genes clearly show that their expression is not ubiquitous. Thus, it is likely that cilia in certain regions of the embryo have a unique transcriptome ("ciliome") that could confer unique function to the cilium, providing an explanation as to why ciliopathies present with a variety of phenotypes.
Our findings within the craniofacial complex, as well as the developing brain (see accompanying manuscript by Snedeker et al.,) suggest that variable phenotypes in ciliopathies are due to unique spatiotemporal expression of ciliary genes or distinct roles for ciliary genes within tissues that contribute to the development of these organ systems. Furthermore, they present an opportunity to study the cilia not as static organelles, but as dynamic signaling hubs that determine how a cell responds to its molecular environment. Our ongoing studies use the mutants generated herein to address these possibilities and aim to determine if modulating expression of certain ciliary proteins can alter the functionality and sensitivity of the cilium to molecular stimuli.

Mouse strains and husbandry
All mouse alleles used in this study have been previously published: Ttc21b tm1c(KOMP)Wtsi-lacZ (Ttc21b flox ) allele [61]; Kif3a tm2Gsn (Kif3a flox ) [62]; Ift88 tm1Bky (Ift88 flox ) [63]. Timed matings were established and noon on the day of mating plug was designated embryonic day (e) 0.5. This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocol was approved by the Institutional Animal Care and Use Committee of the Cincinnati Children's Hospital Medical Center (protocol number IACUC2013-0113). Animals were housed in ventilated racks with automatic water and feeders providing Purina 5010 autoclavable rodent laboratory chow with a 12 hour light-dark cycles. Certified technical personnel and registered veterinary technicians provide daily observation and handling of lab animals. Signs of dehydration and pain as indicated by hunched and lethargic behavior were monitored to assess animal health. All euthanasia and embryo harvests were performed after isoflurane sedation to minimize animal suffering and discomfort. Animal euthanasia was via cervical dislocation.
Genotyping DNA was isolated from tissue samples of embryos. Genotype was determined by PCR using the primer sets listed below. Published protocols were used for all genotyping except for the Ttc21b aln allele where a custom Taqman assay was employed (Invitrogen; details available upon request). Expected products sizes are denoted in parentheses.

Embryo processing
Embryos were harvested at either e11.5 or e14.5, dissected, and imaged. All embryos were fixed in 4% PFA and paraffin embedded. Paraffin sections were cut to 10μm thickness.

Cell counts
Cell counts were performed using ImageJ software and the Cell Counter feature.

Imaging equipment
Whole mount images were taken using a Leica M165FC microscope. All other images were taken using a Leica DM5000 B microscope.

Safranin-O staining
Sections were de-paraffinized and rehydrated. Sections were then stained with Weigart's hematoxylin, rinsed in water and briefly stained with Fast Green (FCF) solution. Sections were rinsed briefly in 1% acetic acid then stained with Safranin-O. Sections were dehydrated and mounted with Permount (Fisher Scientific).

Statistics
Three embryos for each mutant genotype were collected and sectioned. Staining for cell proliferation and cell death were performed on serial sections. Counts were performed on each section and significance of cell proliferation and cell death were determined using the student's ttest. Boxplots were generated using BoxPlotR. (TIF) S1 Table. Cell counts for PHH3 and CC3 in all mutants at e11.5 and e14.5. Average cell counts, standard deviation and P-values for PHH3 and CC3 staining in epithelial or mesenchymal tissues of FNPs and palatal sections at e11.5 and e14.5. Green boxes indicate significance. N's refer to the number of sections counted across 3 separate embryos. (XLSX)