Tbx6 Regulates Left/Right Patterning in Mouse Embryos through Effects on Nodal Cilia and Perinodal Signaling

Background The determination of left/right body axis during early embryogenesis sets up a developmental cascade that coordinates the development of the viscera and is essential to the correct placement and alignment of organ systems and vasculature. Defective left-right patterning can lead to congenital cardiac malformations, vascular anomalies and other serious health problems. Here we describe a novel role for the T-box transcription factor gene Tbx6 in left/right body axis determination in the mouse. Results Embryos lacking Tbx6 show randomized embryo turning and heart looping. Our results point to multiple mechanisms for this effect. First, Dll1, a direct target of Tbx6, is down regulated around the node in Tbx6 mutants and there is a subsequent decrease in nodal signaling, which is required for laterality determination. Secondly, in spite of a lack of expression of Tbx6 in the node, we document a profound effect of the Tbx6 mutation on the morphology and motility of nodal cilia. This results in the loss of asymmetric calcium signaling at the periphery of the node, suggesting that unidirectional nodal flow is disrupted. To carry out these studies, we devised a novel method for direct labeling and live imaging cilia in vivo using a genetically-encoded fluorescent protein fusion that labels tubulin, combined with laser point scanning confocal microscopy for direct visualization of cilia movement. Conclusions We conclude that the transcription factor gene Tbx6 is essential for correct left/right axis determination in the mouse and acts through effects on notch signaling around the node as well as through an effect on the morphology and motility of the nodal cilia.


Introduction
The bilateral, mirror-image symmetry that characterizes the vertebrate body is little more than skin deep, as many internal organs such as the heart, gut, liver and lungs are asymmetric. The development of left/right (L/R) asymmetry involves the initial breaking of bilateral symmetry by the generation of an asymmetric signal in or near the embryonic node followed by the transfer of information from the node to the left lateral plate mesoderm (LPM). This induces a signaling cascade in the left LPM, and the transformation of L/R asymmetric signals into left-or rightspecific morphogenetic programs in the visceral organ primordia [1][2][3][4][5]. In the mouse, the first genes known to be asymmetrically expressed are Nodal around the node, followed by Nodal, Lefty2, and Pitx2 in the left LPM and Lefty1 in the left floor plate of the neural tube [6]. Morphological asymmetry appears first in the heart where a leftward displacement of the future atrioventricular canal presages the eventual dextral looping of the heart tube [7]. Shortly thereafter, the embryo undergoes axial rotation toward its right side. Additional morphological asymmetries appear during development of organs, culminating in a highly asymmetric visceral arrangement in the adult.
An early clue to the initial symmetry-breaking event came from observations that human patients with defects in dynein, a component of the ciliary motor, exhibit heterotaxia [8] as do mice carrying inversus viscerum (iv), a mutation in the gene encoding left-right dynein (lrd) [9]. Mutations in several other genes that affect cilia also cause laterality defects, notably, polycystin-2 (pkd2) [10], which codes for a calcium (Ca 2+ ) channel associated with cilia and thought to play a role as a mechanosensor, and polaris, which is involved in ciliary assembly and may have a role in intraflagellar transport [11]. Use of morpholinos in zebrafish provided direct evidence that pkd2 and polaris control L/R determination through effects on ciliated cells in the node [12].
Recent work has shown that the posteriorly tilted cilia in the ventral node beat with a clockwise rotational movement that sets up a leftward flow of extracellular fluid, known as the nodal flow. The reversal of nodal flow can reverse laterality [13] and an artificial fluid flow across the node can rescue the laterality defects in iv mutants, which have immotile cilia, as well as inversion of embryonic turning (inv) mutants, which have weakly motile cilia [13][14][15]. Thus nodal flow is the key symmetry-breaking event responsible for the establishment of laterality differences [14,[16][17][18].
There are two prevailing models regarding how nodal flow causes a break in symmetry and generates a laterality signal: the first, known as the two-cilia model, proposes that leftward nodal flow generated by motile cilia is sensed by nonmotile mechanosensory cilia at the periphery of the node, causing asymmetric Ca 2+ signaling that triggers a lateralized gene regulatory cascade [19,20]. The second model, known as the morphogen gradient model, proposes that an extracellular morphogen is transported leftward by nodal flow creating a gradient [17]. A variation on this model has recently been proposed [21] based on the observation of a novel mode of extracellular transport of morphogens via nodal vesicular parcels (NVPs). NVPs are released from microvilli on the surface of nodal cells and propelled toward the left wall of the node by the nodal flow where they fragment and release their contents. The vesicles have been suggested to transport sonic hedgehog (SHH) and retinoic acid (RA), thus potentially providing morphogens to the left side of the node and culminating in the asymmetric activation of Ca 2+ signaling [21]. However, unlike the situation in the chick, there is as yet no evidence for asymmetric hedgehog signaling in the node of the mouse [22] and putative morphogens responsible for symmetry breaking have not been identified.
The transforming growth factor-b (TGFb) family member Nodal is a key determinant of laterality and is involved in L/R determination at several levels [2]. In mice, Nodal is initially expressed symmetrically the periphery of the node, with a transient, higher level of expression developing on the left side [23,24]. By the 3-4-somite stage, it is also expressed in the left LPM in an asymmetric pattern that is conserved in all vertebrates. How Nodal expression in the node translates into asymmetric Nodal expression in the LPM is unknown, although recent work indicates that the signal may be relayed through the mesoderm [25]. Loss or reduction of perinodal Nodal results in lack of expression of Nodal, Lefty2, and Pitx2 in the LPM [26][27][28].
Several studies show that Notch signaling is directly upstream of Nodal expression [29][30][31][32]. Embryos mutant for the Notch ligand Delta-like 1 (Dll1), the primary mediator of Notch signaling RBPjk, or doubly mutant for the receptors Notch1 and Notch2 have randomized laterality. Furthermore, node-specific expression of Nodal can be eliminated by mutation of RBP-J binding sites in the Nodal promoter [29]. Recently, Wnt3a has also been implicated early in the genetic hierarchy of laterality determination. Mutants for Wnt3a show randomized heart looping and turning, and expression of Nodal, Lefty2, and Pitx2 are all delayed and then expressed bilaterally, whereas Lefty1 is not expressed. In addition, reduced expression of polycystin-1 (PC1) in nodal cilia, a protein that interacts with polycystin-2 (PC2), indicates a potential role for Wnt3a in monocilia. It was postulated that WNT signaling might play dual roles in laterality determination by regulating Notch signaling through Dll1, as well as affecting nodal cilia through PC1 [33]. However, another study reported a lack of PC1 in nodal cilia and the lack of a laterality phenotype in Pkd1 mutant mice [34].
Tbx6 is a T-box transcription factor gene that has important roles in the specification of presomitic mesoderm (PSM) and the formation of somite borders [35][36][37][38][39]. Studies with the hypomorphic Tbx6 rv allele have revealed an interaction between Tbx6 and Dll1 in the PSM [40,41]. Furthermore, Tbx6 acts synergistically with WNT signaling to regulate Dll1 expression in PSM [42], raising the possibility of involvement of Tbx6 in laterality determination through regulation of Dll1. We have investigated laterality in Tbx6 mutant embryos and have observed heterotaxia in the homozygous null mutants, which has not been previously reported. Here we explore the mechanism of this laterality defect and show that Tbx6 is upstream of the Notch signaling pathway in the perinodal region. In addition, Tbx6 has profound effects on the morphology and motility of nodal cilia that result in a disruption of asymmetric Ca 2+ signaling around the node.
Embryos were collected from timed matings; noon on the day of the plug was considered embryonic day (E) 0.5. Early embryos were staged morphologically [47]; in later embryos, somite numbers were used for staging wild type embryos and mutant embryos were stage matched by head or heart morphology. Genotypes of embryos carrying Tbx6 tm2Pa were ascertained on the basis of fluorescence intensity, which, as confirmed by PCR, correlates with the number of mutant alleles. E7.5-8.5 embryos from other crosses were genotyped by PCR using the following primers pairs (1) 59-GGGAGAATGAGGATCCAGG-39, (2) 59-TACCATCCACGAGAGTTGTAC-39 to obtain a 200 bp wild type allele fragment; and (3) 59-ATTGCACGCAGGTTC-TCCGG-39; (4) 59-GTCACGACGAGATCCTCGCC-39 to obtain a 550 bp mutant allele fragment. E9.5-10.5 null and compound heterozygous mutant embryos were recognized by the characteristic mutant phenotype [36,38].

ISH and b-galactosidase staining
Embryos for ISH were dissected in phosphate buffered saline (PBS) with 0.2% albumin bovine serum (Fraction V, Sigma), fixed overnight in 4% paraformaldehyde in PBS at 4uC, dehydrated in methanol and stored at 220uC. Whole mount ISH was performed using antisense riboprobes [48]. b-galactosidase staining was performed according to standard protocols [49].
Immunofluorescence. Embryos were dissected in PBS and fixed for 2-24 hours in 4% paraformaldehyde in PBS at 4uC, washed in PBS and stored in PBS + 0.1% azide at 4uC. The protocol described by Nakaya et al. [33] was used with minor modifications. Briefly, after incubation with the primary antibody, embryos were incubated in blocking solution (Vector Labs, Vectastain ABC kit) for 1 h at 4uC, followed by incubation with secondary antibody overnight. Embryos were extensively washed in TBST, incubated with streptavidin-conjugated substrates (e.g. Streptavidin AlexaFluor-543 or Streptavidin AlexaFluor-633, Molecular Probes) for 30 m-1 h at room temperature, and then washed in TBST. Embryos were counterstained with Hoechst (Molecular Probes) to label nuclei, mounted in SlowFade Antifade (Molecular Probes) and imaged using laser scanning confocal microscopy. Primary antibodies used were against acetylatedtubulin (1:400, Sigma) and Tbx6 (1:300, Santa Cruz).
Scanning Electron Microscopy (SEM). Embryos for SEM were dissected in PBS and fixed in 0.25% glutaraldehyde in PBS for at least 48 hours, dehydrated into 100% ethanol and criticalpoint dried prior to coating with gold/palladium in a Denton Vacuum Desk 1V sputter coater. Coated embryos were mounted onto double-sided adhesive tape on metal stubs and photographed with a Jeol JSM 35 Scanning Electron Microscope or a Zeiss Field Emission Scanning Electron Microscope Supra 25.
Analysis of Calcium Signaling. For live imaging of calcium signaling, embryos carrying the Tbx6 tm1Pa allele were dissected at E7.5-E8.0 in PB-1 medium [50], then placed in 50% DMEM/F12 and 50% rat serum supplemented with penicillin-streptomycin and glutamine for 30 m prior to incubation with the fluorescent calcium indicators Fluo-3 or Fluo-4 (Molecular Probes; 10 mm in DMEM/F12) with 25% rat serum for 20 m. Fluo-4 is an analog of Fluo-3 with increased fluorescence excitation at 488 nm, thus Fluo-4 was used for all mutant embryos analyzed. Embryos were then washed 3 times in DMEM/F12 with 25% rat serum and placed node-side down into individual wells made in an agarosecoated prewarmed coverslip-bottomed dish (MatTek). The dish was re-equilibriated in an incubator for 10-20 m and then placed in a heated, humidified and gassed chamber on an inverted microscope (Zeiss Axiovert 200M) attached to a Zeiss LSM510 META laser scanning confocal for live imaging. Fluo-3 and Fluo-4 were excited using a 488 nm Argon laser. z-stacks of xy images were acquired at 2 mm intervals using a 20x/NA0.75 Plan-Apochromat objective. All dissections, incubations and imaging were performed at 37uC. Raw data was processed using Zeiss AIM software (Carl Zeiss Microsystems at http:://www.zeiss.com/).
Imaging cilia movement. To generate plasmid pCX::tau-GFP, a green fluorescent protein (GFP) fusion to the bovine microtubule binding protein tau with a rabbit b-globin polyadenlyation sequence was placed under the regulation of the CAGGS promoter [51], which drives widespread transgene expression. The CAG::tau-GFP strain of mice was generated through ES cell mediated transgenesis [52]. Mice having widespread expression are viable and fertile and indistinguishable from non-transgenic littermates (S. Nowotschin and A.-K.H. unpublished observations).
Mice carrying the CAG::tau-GFP transgene were crossed with mice carrying the Tbx6 tm2pa allele. Embryos for live imaging of nodal cilia movement were treated as described for Ca 2+ signaling.
GFP was excited using a 488 nm Argon laser. Where necessary linear unmixing [53] was used to separate GFP from yellow fluorescent protein (YFP) as described previously [54]. z-stacks of xy images were acquired at 1 mm intervals using either a 20x/ NA0.75 Plan-Apochromat or a 40x/NA1.3 Plan-Neofluar objective. Scan speeds were set at approximately 3 seconds per xy frame, with each frame comprising 5126512 pixels. z-stacks were rendered and time-lapse movies were generated using the Zeiss LSM Image Browser. Cilia were analyzed with the investigator blind to the genotype of the embryos.

Heterotaxia of Tbx6 null mutant embryos
The dextral looping of the embryonic heart is one of the earliest morphological signs of bilateral asymmetry. Normally, the heart tube loops horizontally from left (inflow) to right (outflow) in a Cshaped loop when viewed from the ventral aspect. In Tbx6 null mutants, 40% of E9.5-E10.5 embryos (22/55) showed abnormal heart looping that was either completely reversed (J-shaped) or looped ventrally rather than horizontally (Table 1 and Figure 1A-D) whereas all heterozygous and wild type embryos (n = 231) had normal heart looping. Rotation of the mouse embryo, during turning to a fetal position, is normally in an anticlockwise direction (towards the embryo's right side) resulting in the embryo becoming enveloped by its membranes with the umbilical vessels, placenta, and tail on the right side of the embryo ( Figure 1E). The vitelline vessels enter and leave the yolk sac on the left side ( Figure 1F). While the pattern is rarely disrupted in Tbx6 heterozygous and wild type mice (4/151; 3%), it is completely reversed in 18% of Tbx6 homozygous null mutant embryos (7/38), indicating that turning took place with a clockwise rotation ( Figure 1E&F). An additional 4 embryos (4/38) showed abnormal laterality of one or more of these features ( Table 1).
Reversal of heart looping did not always correlate with the direction of turning, as two mutant embryos exhibited reversed turning but normal heart looping and 8 embryos had reversed heart looping but normal turning. This result indicates heterotaxia, the independent randomization of situs in different features. Overall, 50% (19/38) of mutant embryos scored for heart looping and turning displayed some form of laterality defect. Null/ hypomorphic (Tbx6 tm2Pa /Tbx6 rv ) compound heterozygous mutants, Table 1. Laterality defects in E9.5-E10.5 embryos from crosses between Tbx6 tm2Pa /+ and Tbx6 tm1Pa /+ heterozygous mice and between Tbx6 tm2Pa /+ and Tbx6 rv /+ heterozygous mice.  Two had tails twisted 180 0 so that the umbilicus looped around the dorsal side of the tail; this appeared to be due to greater flexibility of the tail of mutants because of the lack of somites, rather than to an altered direction of turning 3 Five had ventrally looped hearts 4 One had a tail twisted 180 0

5
Wild type includes all morphologically normal littermates, which include +/+ and heterozygous embryo doi:10.1371/journal.pone.0002511.t001 on the other hand, did not display any abnormalities in laterality (n = 22, Table 1). These results indicate that the loss, but not the reduction, of Tbx6 randomizes laterality determination in early embryos.

Molecular correlates of laterality defects in Tbx6 mutant embryos
Nodal expression is normally seen around the node at early gastrulation, becoming stronger on the left side. At 2-6 somite stages, it appears in the left LPM. In Tbx6 null embryos (n = 6), perinodal expression was reduced at E7.5 in early head fold (EHF) to late head fold (LHF) stages compared with wild type embryos (Tbx6 +/+ and +/2; n = 15) (Figure 2A,B). Decreased expression was also observed in E8.0-8.5 mutant embryos at 3-7 somite stages. Furthermore, none of these mutants showed expression in the LPM (n = 5) whereas robust expression around the node and in the left LPM was observed in wild type embryos (n = 7) ( Figure 2C,D). To extend the narrow time window of Nodal detection by ISH, we made use of a Nodal-lacZ reporter allele [23]. In Nodal-lacZ heterozygous embryos of 5-12 somites, b galactosidase staining reported robust Nodal expression around the node and in the left LPM of wild type embryos (n = 32) whereas compound Nodal-lacZ heterozygous, Tbx6 mutant embryos equivalent to 3-10 somite stages as judged by head morphology showed only modest expression around the node and no expression in the LPM (n = 8)( Figure 2E,F). This lack of expression, however, could be compounded by the fact that the Nodal-lacZ allele is a null allele.
Lefty1, which is transiently expressed in the left floor plate of the neural tube at early somite stages, was detected in 1/16 wild type and 3/18 mutant embryos ( Figure 2G,H). Two of these mutants had only a few scattered cells in the midline ( Figure 2H), whereas the other had expression throughout the length of the floor plate similar to the wild type embryo. Lefty2 and Nodal are normally coexpressed in the left LPM. We found no expression of Lefty2 in the LPM in homozygous mutant embryos (n = 9) whereas expression was detected in the left LPM or heart of 8/10 stagematched wild type embryos ( Figure 2I,J).
Pitx2 is a global executor of L/R patterning. It is a target of the nodal and Lefty pathways in the LPM and is expressed shortly after Nodal in a broader domain on the left side. At 3-6 somite stages, wild type embryos showed Pitx2 expression in the left LPM, which became limited to the left inflow tract of the heart during 7-10 somite stages (n = 33). At stages where LPM expression of Pitx2 would be expected, the majority of homozygous mutants had no LPM expression (22/24)( Figure 2L-O), but a single mutant embryo of ,3 somites had expression in the right LPM and another of ,8 somites had expression in the left LPM in a more restricted domain than wild type ( Figure 2K,P). At later stages, no mutant embryos had LPM expression (n = 14). Only 4/14 mutant embryos had expression in the left atrium of the heart, 2 of which had dextrally looped hearts. The mutant embryos lacking expression either had hearts that had not yet looped (n = 6) or had ventrally looped hearts (n = 4). Taken together, these results indicate that in the absence of Tbx6, Nodal expression is greatly diminished around the node and is missing from the left LPM. The absence of either Nodal or Lefty2 in the left or right LPM, the lack of Pitx2 expression in most mutant embryos, and inappropriate right-sided expression in one mutant indicate that the transfer of laterality information from the node to the left lateral plate is disrupted.

Alterations in Notch signaling in Tbx6 mutants
Notch signaling is known to be upstream of Nodal expression in the node. The decreased Nodal expression observed around the node implicates the Notch signaling pathway in the disruption of laterality determination in Tbx6 mutant embryos. We examined the expression of Notch signaling pathway components in Tbx6 mutant embryos. The Notch receptors Notch1 and Notch2 which are both required for the establishment of asymmetry were expressed normally in all E7.5 embryos (n = 17 and 13, respectively) from Tbx6 heterozygous matings ( Figure 3A-D) indicating that there is likely no major disruption of expression of these receptors in mutant embryos. Expression of the Notch ligand, Dll1, however, is greatly diminished in the primitive streak and PSM of Tbx6 mutant embryos at E7.5 (n = 6)( Figure 3E-G, I-K).
Dll1 is a known downstream target of Tbx6 in the PSM at E8-8.5 [42]. We previously showed that Tbx6 is expressed in the primitive streak and PSM from E7.0, but is excluded from the node [35]. To compare with Dll1 expression at E7.5, we examined wild type embryos (n = 8) and found Tbx6 expression extending halfway around the node in EHF to LHF stage embryos in an area overlapping with Dll1 expression (Figure 3H,L). Furthermore, we have shown that Tbx6 protein is present in the PSM but not in the node (Figure 4). This suggests that Tbx6 expression overlaps with Dll1 in the PSM and perinodal region at the time L/R asymmetry is established and that the reduction in Dll1 expression in Tbx6 mutant embryos could be a direct effect.

WNT signaling is marginally affected in Tbx6 mutant embryos
Because WNT signaling in the node regulates nodal signaling through the Notch pathway [33], we assessed WNT signaling in Tbx6 mutant embryos. In early bud to LHF stage embryos, Wnt3a expression in the primitive streak was similar between wild type (n = 7) and 4/5 Tbx6 mutants; the exceptional mutant embryo showed less intense expression ( Figure 3M,N). In embryos ranging from 1-12 somites, Wnt3a expression in PSM, neural plate and the base of the allantois was similar in wild type (n = 6) and mutant (n = 17) embryos, although 2 mutants of approximately 3-6 somites had reduced expression compared with stage matched controls (Figure 3 O,P).
To determine sites of active WNT signaling, we used two transgenic lines, TCF/Lef-lacZ and TOPgal that report Wnt/bcatenin activity with b-galactosidase readout. With TOPgal, bgalactosidase activity was detected in both Tbx6 wild type (n = 24) and homozygous mutant (n = 14) embryos, although 3 mutants had reduced levels in the PSM (Figure 3Q,R). With TCF/Lef-lacZ, 1/5 mutant embryos had reduced PSM expression compared with wild type (n = 8; Figure 3S,T). Thus most mutant embryos are indistinguishable from wildtype, although a small proportion (7/ 41; 17%) had reduced Wnt3a expression or WNT signaling.
To determine whether Tbx6 and Wnt3a interact genetically, we intercrossed Tbx6 tm2Pa and Wnt3a heterozygotes and examined the direction of heart looping and somite morphology of double heterozygous offspring. At E10.5, all Tbx6; Wnt3a double heterozygotes (n = 18) and all Tbx6 single heterozygotes (n = 19) displayed normal heart looping and somite formation, indicating that there is no genetic interaction between Wnt3a and Tbx6 with respect to direction of heart looping or somite segmentation. We conclude that although some Tbx6 homozygous mutants display reduced Wnt signaling, this may not play a significant role in the disruption of laterality determination in Tbx6 mutant embryos.

Abnormalities of the node and nodal cilia
The integrity of the midline, node and nodal cilia was examined using additional molecular markers and by direct observation of the nodal cilia. We saw no indication of disruption of the notochord in embryos of 2-8 somites using a Brachyury probe (n = 5 wildtype; 8 mutants) or of the node and anterior midline of E7.5 embryos using chordin (n = 17 wild type; 6 mutants) or Shh probes (n = 7 wild type, 5 mutants) ( Figure 5).
Antibodies against acetylated a-tubulin, which marks stable microtubules, were used to detect the presence of cilia in the node of embryos at E7.5-8.0. Wild type embryos (n = 4) had long, filamentous nodal cilia ( Figure 6A-C) whereas homozygous mutant embryos (n = 6) showed evidence of shorter, thicker cilia, some with terminal knobs (Figure 6D-F). SEM revealed gross abnormalities in the monocilia of the node in Tbx6 mutant embryos (n = 15) compared with wild type embryos (n = 6). Wild type cilia were long slender projections ( Figure 6G,J) whereas most mutant cilia had terminal bulges or balloon-like structures ( Figure 6H,K). A single mutant embryo had straight cilia but these were shorter than wild type ( Figure 6I,L). The node of this embryo was also misshapen and contained cells with the   appearance of the surrounding visceral endoderm. ( Figure 6I). Clearly, the lack of Tbx6 has a profound effect on the morphology of monocilia in the cells of the node, even though Tbx6 is not expressed in these cells at this time.

Live imaging of nodal cilia motility
To determine if the cilia of Tbx6 mutant embryos are motile, we devised a method to directly image cilia in living embryos that relies on 1) a strain of mice expressing a fluorescent protein fusion that labels cilia, and 2) a mode of image acquisition that provides information on the dynamics of GFP distribution thereby providing a readout of cilia movement. We reasoned that we could noninvasively label and subsequently live image cilia by imaging tubulin, a central structural component of cilia, using geneticallyencoded fluorescent protein fusions that would incorporate into microtubules. We generated a fusion of GFP to the bovine microtubule binding protein tau and demonstrated that it labeled and permitted the dynamic visualization of microtubules in embryonic stem (ES) cells and mouse embryos, most easily observed through stereotypical rearrangements in the cytoskeleton that accompany mitosis ( Figure 7D-G and supplementary Video S1).
Laser point scanning confocal optical sections in the vicinity of the node reveal fluorescence in basal bodies of nodal cells connected to long, narrow projections. These are easily seen in single (xy) optical sections taken from a z-stack of the entire node ( Figure 7A-C and supplementary Video S1). To confirm that we were visualizing cilia and to document their movement, highspeed acquisition of z-stacks of xy sections were taken at timed intervals (3D time-lapse). Subsequent 3D rendering of xyz data revealed GFP-labeled projections that changed their orientation between successive time points ( Figure 7H-K), suggesting movement and supporting the idea that the observed projections are cilia. However, since it has been determined that nodal cilia rotate at approximately 600 rpm [17], the rate of image acquisition possible with either a spinning disc or laser scanning confocal is too slow to permit the continuous imaging required to visualize rapid rotation.
To circumvent this problem, we acquired z-stacks of optical slices (i.e. xy images) of the node with a laser point scanning confocal microscope at a defined speed (determined empirically to be between 3 and15 s) to generate kymographs that depict cilia movement. In such optical sections, the y-axis represents a time axis ( Figure 8A) and the xy image serves as a kymograph documenting the spatial displacement of GFP revealed by the traces of GFP fluorescence reflecting cilia movement ( Figure 8B). Because we acquired z-stacks of xy data the majority of cilia were imaged on consecutive optical sections representing independent measurements and verification of their movement. We observed a variety of types of GFP traces which we have grouped into 5 categories ( Table 2): 1) regular beat: sinusoidal traces or a series of parallel lines were interpreted as motile cilia rapidly rotating with constant periodicity, the shape of the trace depending on the angle of observation, (Figure 8B,C,D,F, red arrowheads); 2) straight: nonmotile cilia were observed as straight lines that did not move or moved only slightly between consecutive sections ( Figure 8C,E,F, blue arrowheads); 3) wavy: wavy traces that moved slightly between consecutive sections suggested slowly swaying cilia that either could be weakly motile cilia or nonmotile cilia being passively moved by fluid currents (Figure 8C, green arrowhead; 4) erratic: erratically moving, zigzag traces indicated motile cilia with uncoordinated, irregular movements ( Figure 8E,F, yellow arrowheads); 5) spots: stationery or moving spots of fluorescence could either indicate morphologically abnormal short cilia or immotile cilia viewed along their length.

Abnormal cilia motility in Tbx6 mutant embryos
Using this new method of visualizing cilia motility in living embryos, we observed that all wild type embryos (n = 13) displayed mainly two types of traces, 1) regular beat, corresponding to motile cilia ( Figure 8C,D,D9, red arrowheads) and 2) straight, corresponding to nonmotile, presumably mechanosensory cilia ( Figure 8C,C9, blue arrowhead), with minor populations of cilia in the other categories described in the previous section. The proportion of each type of cilia at the periphery and in the pit of the node was quantified for 8 wild type embryos for which the z stacks encompassed the node from crown to pit. At the periphery of the node, nonmotile cilia predominated (57%) whereas regularly beating cilia predominated in the pit of the node (65%) but were also fairly common at the periphery (18%, Table 2). Cilia categorized as wavy were more common at the periphery than in the pit of the node, compatible with these being nonmotile, mechanosensory cilia. Tbx6 mutant embryos (n = 6) displayed the same types of cilia but the proportion of each type and the distribution between the periphery and the pit of the node was significantly different from wild type in the 4 embryos with images spanning from the crown to the pit of the node ( Table 2) (X 2 = 21.8, 4df, p,001 for distribution at the periphery; X 2 = 59.4, 4df, p,0.001 for distribution in the pit). At the periphery of the node in mutants, the proportion of straight cilia ( Figure 8E,E9, blue arrowheads) was the same as wild type (57%) but there was a lower proportion of waving and regularly beating cilia and higher proportions of erratically moving cilia ( Figure 8E,E9 yellow arrowheads) and spots, which most likely represent the short, globular cilia observed with SEM and anti-tubulin ( Figure 6). In the pit of the node in mutants there were higher proportions of nonmotile cilia ( Figure 8F, blue arrowhead), erratically moving cilia ( Figure 8E,E9,F,F9; yellow arrowheads) and spots ( Figure 8F, purple arrowheads) compared with controls, and a decrease in the proportion of regularly rotating cilia ( Figure 8F,F9, red arrowheads). These results indicate that Tbx6 mutant embryos display severe ciliary defects throughout the node but particularly in the pit of the node with the proportion of regularly rotating cilia decreased at the expense of erratically moving or stationary cilia, many of which have abnormal morphology.
Furthermore, the tau-GFP transgene, which highlights microtubules, revealed cellular irregularities within and around the node of mutant embryos ( Figure 8E,F). The movement of cilia and the differences between the cellular structure of Tbx6 mutant and wild type nodes can be visualized in the movies (z stacks) of successive images (Supplementary Videos S2 and S3).

Loss of Tbx6 results in loss of asymmetric intracellular Ca 2+ at the periphery of the node
Asymmetric Ca 2+ has been detected on the left side of the node both extracellularly in chick [31] and intracellularly in mouse [19] and is dependent on a leftward nodal flow. In order to determine whether the abnormalities in the node and nodal cilia in Tbx6 mutant embryos affects Ca 2+ signaling, we incubated whole embryos (yolk sac and placenta removed) in the presence of either the Fluo-3 or Fluo-4 ionophores, and imaged whole, living embryos through the node and perinodal region using confocal microscopy. In wild type embryos, we did not detect asymmetric Ca 2+ levels within cells of the node itself. However in 0-4 somite stage wild type embryos, we observed a robust asymmetric increase of intracellular Ca 2+ on the left side of the node, which is not present in embryos of 8 somites or greater. In contrast, Tbx6 homozygous mutant embryos showed no asymmetries in intracellular Ca 2+ in the region around the node (Figure 9 and Table 3). As functional cilia and a leftward nodal flow are necessary to generate asymmetric Ca 2+ around the node [19], these results indicate that the abnormally shaped cilia in Tbx6 mutant embryos are unable to generate a unidirectional nodal flow. Table 2. Classification of cilia at the periphery and in the pit of the node from 8 wild type and 4 mutant embryos for which z-stack images encompassed the entire node.

Discussion
Tbx6 affects Notch signaling around the node Our study has revealed a novel role for the T-box transcription factor Tbx6 in determination of the L/R body axis. Homozygous mutants, but not compound null/hypomorphic heterozygotes, show randomization of the direction of heart looping and independent alterations in the direction of embryo turning. The molecular phenotype of the Tbx6 mutant embryos indicates a disruption of nodal signaling and is similar to the phenotype of embryos with decreased levels of Nodal expression around the node [26]. We observed no Nodal expression in the LPM mesoderm of mutant embryos, however, the downstream target, Pitx2, was expressed in the LPM in a minority of mutant embryos. This is most likely due to low residual levels of Nodal expression in the mutants as several other mutations with laterality defects show a similar expression of Pitx2 in the absence of detectable LPM Nodal expression [55,56]. Another possibility is that the LPM itself is missing or defective in Tbx6 mutant embryos although morphological and molecular evidence argues against this possibility [36](D. Concepcion and V. E. Papaioannou, unpublished observations).
It is known that perinodal nodal signaling is controlled by the Notch signaling pathway and that disruption of Dll1 or other components of the Notch pathway result in abnormal L/R determination [29][30][31][32]. Furthermore, as Dll1 is a direct downstream target of Tbx6 in the paraxial mesoderm [40,41], our observation that Dll1 is down regulated around the node at E7.5 in an area coincident with Tbx6 expression indicates that Tbx6 has a direct effect on Dll1 expression and suggests one possible mechanism for the disruption of L/R determination in Tbx6 null mutants. The asymmetric expression of Nodal is thought to be a response to preexisting asymmetry around the node [31], although no gene with earlier asymmetric expression has yet been found in mice. We do not think that Tbx6 plays this role but rather that the lack of Tbx6 results in reduction of Dll1 and, consequently, Nodal expression below a critical threshold for nodal signaling to the left LPM.
The effect of Tbx6 on Dll1 expression and the subsequent reduction in Nodal could account for the observed disruptions in L/R determination. Three studies of Dll1 null mutants all report randomized laterality although they disagree as to whether LPM expression of genes such as Nodal and Lefty2 are randomly expressed [32] or absent from the LPM [29,30,32], as seen in Tbx6 mutants. A difference between Dll1 and Tbx6 mutants is the expression of Lefty1 in the latter. Although we saw Lefty1 in only a single wild type embryo, possibly due to the transient nature of expression, we nonetheless observed expression in three mutants, whereas Dll1 null mutants have no Lefty1. This Lefty1 expression in Tbx6 mutants may be due to residual Dll1 and Nodal expression.
Wnt3a, which is expressed in the node and PSM, has been implicated in L/R patterning through an effect on the Notch/ Delta pathway [33]. Furthermore, there is evidence that Dll1 is synergistically regulated by Tbx6 and WNT signaling, and it has been suggested that Wnt3a is upstream of Tbx6 [42]. However, Tbx6 is normally expressed in Wnt3a mutants at early stages (0-2 somites) and is only moderately down regulated later [57]. Thus, Tbx6 does not appear to be regulated by Wnt3a at the time of laterality determination. Because Tbx6 and Wnt3a are expressed in overlapping domains in the PSM, we examined WNT signaling in Tbx6 mutant embryos and although a few mutants had decreased levels of WNT signaling, the significance of this observation is not clear. We also investigated whether Tbx6 and Wnt3a genetically interact but found no disruption of laterality in double heterozy-gotes. Thus, the inactivation of one allele of each gene is not sufficient to disturb laterality determination. It is possible that a more sensitized screen using hypomorphic and null alleles might reveal a genetic interaction.

New method of live imaging cilia and cilia movement
We have developed a new method of visualizing the movement of cilia of the node in living mouse embryos. The tau-GFP transgene produces a fusion protein that marks tubulin and allows the visualization of microtubules, including basal bodies and cilia. Using this transgene with scanning confocal microscopy at a reduced scan speed, we have produced kymographs of cilia, essentially tracing their movement in real time. We have visualized the relatively stationary mechanosensory cilia around the periphery of the node, as well as the regular, rapid rotational movement of nodal cilia. Moreover, since the parameters (e.g. the number of pixels per frame and the scan rate) of image acquisition are known, the rotation frequency can be calculated. Interestingly, live imaging of basal bodies in early somite stage (E8.0) wild type embryos did not reveal a caudal bias in their position as has been suggested [5]. Additional experiments will be required to determine if asymmetric positioning of the basal body within nodal cells occurs at earlier or later stages. With this method, we have been able to detect distinct differences in cilia morphology and motility associated with the Tbx6 mutant phenotype as well as detecting differences in the microtubule structure of the nodal cells themselves. This method will be valuable in making detailed assessments of cilia function in other mutants as well as quantifying the kinetics of movement.

Effect of Tbx6 on the node and nodal cilia
Although a decrease in Dll1 signaling would be sufficient to result in disruptions in L/R axis determination, we discovered that the Tbx6 mutants also have grossly malformed nodal cilia, a striking departure from the Dll1 phenotype. In Dll1 mutants, nodal cilia were normal although disruptions in the structure of the node were noted in a few cases [29,32]. The cilia morphology of Tbx6 mutants is similar to that seen in cilia of Chlamydomonas mutants associated with intraflagellar transport (IFT) defects [58,59]. Moreover, mice with a mutation in Dnchc2, encoding a subunit of the retrograde IFT motor [60] show a single bulge along their length. Using a novel method of visualizing cilia movement, we have determined that the morphological abnormalities in the Tbx6 mutant cilia are accompanied by disruptions in cilia motility. Although we cannot judge whether the short cilia which appear as fluorescent spots are rotating regularly, there are clearly more cilia with erratic movements and more nonmotile cilia in the pit of the node of mutant embryos,. Furthermore, we show that the asymmetric Ca 2+ flux on the left side of the node is abolished in mutants. It has been demonstrated that asymmetric Ca 2+ signaling is dependent on a unidirectional nodal flow, as embryos lacking motile cilia do not show elevated Ca 2+ levels [19]. Thus, using this functional readout of the nodal flow, we favor the interpretation that the cilia of Tbx6 mutant embryos, some of which are capable of movement, cannot generate a robust unidirectional nodal flow, although we cannot rule out the possibility that the mechanosensory cilia are defective in responding to a flow.
Although it is not known precisely how the nodal flow is translated, one model proposes that mechanosensory cilia at the periphery of the node transduce the nodal flow leading to asymmetric Ca 2+ signal on the left side of the node [19,20]. A model has been proposed from evidence in chick that Ca 2+ modulates Notch signaling activity [31]. Thus, Tbx6 appears to have effects on at least two separate aspects of laterality determination which could both impinge on Notch signaling: one a direct effect on expression of Dll1 in PSM and the second, an effect on the morphology and motility of nodal cilia which interferes with the nodal flow ( Figure 10).
There appears to be no ortholog of mammalian Tbx6 in zebrafish [61,62], but a gene in the same subfamily, tbx16, which is mutated in the zebrafish spadetail (spt) mutation, also has an effect on laterality determination, possibly through multiple pathways similar to what we are postulating for Tbx6. tbx16 regulates the expression of pkd2, the gene coding for polycystin-2 (PC-2), the proposed Ca 2+ ion channel associated with nodal cilia [12] and it regulates nodal activity and notch signaling through regulation of the nodal antagonist, charon [63].
Tbx6 and tbx16 show similarities in expression in PSM and both have effects on somitogenesis. However, Tbx6 is not expressed in the mouse node whereas tbx16 is expressed in the precursor cells of Kupffer's vesicle, a ciliated epithelial structure analogous to the mouse node, and is necessary for its morphogenesis [64]. Another T-box gene, Brachyury, which is expressed in the node, affects laterality determination in mice [65], as does the ortholog in zebrafish [66]. The brachyury mutant notail (ntl) results in down regulation of the IFT gene, polaris [12], and along with tbx16, brachyury regulates nodal activity and the morphogenesis of Kupffer's vesicle. Thus, while several T-box genes play roles in L/R patterning, Tbx6 appears to be unique in doing so through an effect on the morphology and function of the nodal cilia in the absence of node expression. It is not known at present how Tbx6 affects the nodal cilia but it is interesting to note that a peripheral effect on asymmetric gene expression at the node in chick embryos was postulated when it was demonstrated that the induction of asymmetric Nodal expression requires secondary signals produced in the PSM [67]. Alternatively, the affected cells of the node might be the progeny of cells that once expressed Tbx6 and then downregulated Tbx6 as they migrated into the node. If the node is derived from cells that previously expressed Tbx6, the effect on cilia might be through direct transcriptional regulation of genes required for ciliogenesis. Future studies will be needed to trace the lineage of Tbx6 expressing cells and to identify targets of Tbx6 in the PSM to make the link between expression of Tbx6 and the abnormal cilia in the node.

Supporting Information
Video S3 z-stack of xy images in the vicinity of the node of an E8.25 (6 somite stage) CAG::tau-GFP hemizygous embryo that is homozygous null mutant for Tbx6. Starting ventrally at a level around the crown of the node, successive xy optical slices are dorsalward and depict regions deeper into the node ending in the pit. The movie contains 34 individual xy slices taken at 1 mm intervals. Found at: doi:10.1371/journal.pone.0002511.s001 (0.66 MB MPG) Video S1 3D time-lapse sequence of a ventral region encompassing the node of an E8.25 (6 somite stage) CAG::tau-GFP hemizygous mouse embryo. Dynamics of cell movement and microtubule organization are evident. During the course of the movie several cells located at the periphery (crown) of the node are seen to divide. The movie covers a 4 hr 10 min period. It comprises 50 z,-stacks acquired at 5 min intervals. Each frame represents a 3D reconstruction of a 14 mm z-stack with each xy slice taken at a 2 mm interval. Video S2 z-stack of xy images in the vicinity of the node of an E8.25 (6 somite stage) CAG::tau-GFP hemizygous mouse embryo, wild type for Tbx6. Starting ventrally at a level around the crown of the node, successive xy optical slices are dorsalward and depict regions deeper into the node ending in the pit. The movie contains 25 individual xy slices taken at 1 mm intervals. Found at: doi:10.1371/journal.pone.0002511.s003 (0.35 MB MPG) Figure 10. Model of two routes by which Tbx6 could affect left/ right axis determination. Symmetric Tbx6 expression in the presomitic mesoderm and primitive streak (blue area) directly regulates Dll1 expression, which in turn regulates Nodal through notch signaling. In addition, through an unknown mechanism, Tbx6 affects the morphology and motility of the cilia in the node. The model proposes that in the absence of Tbx6, Dll1 and Nodal expression are reduced below a threshold level to initiate the cascade of asymmetric gene expression, and in addition, the motility of the cilia is compromised to the extent that the nodal flow and/or the mechanosensory cilia are disrupted resulting in the lack of Ca +2 signaling at the periphery of the node. LPM = lateral plate mesoderm, noto = notochord. doi:10.1371/journal.pone.0002511.g010