|Home | About | Journals | Submit | Contact Us | Français|
One of the most highly conserved steps in left-right patterning is asymmetric gene expression in lateral plate mesoderm (LPM). Here, we quantitatively describe the timing of the posterior to anterior (PA) wave-like propagation of zebrafish southpaw (Nodal) and pitx2 in LPM and lefty1 in the midline. By altering the timing of the PA wave, we provide evidence that the PA wave in the LPM instructs brain asymmetry. We find that initiation of pitx2 in LPM and lefty1 in midline depends on Southpaw, and that casanova (sox32) and two Nodal inhibitors, lefty1 and charon, have distinct roles upstream of PA wave initiation. Surprisingly, Casanova, endoderm and Kupffer’s Vesicle are not required for normal timing of southpaw initiation and PA propagation. In contrast, lefty1 morphants display precocious asymmetric initiation of southpaw with an intrinsic left-hand orientation, whereas charon morphants have premature initiation without LR orientation, indicating distinct roles for these Nodal antagonists.
The establishment of left-right (LR) asymmetries in vertebrates is a highly conserved process. Before the morphogenesis of asymmetries in the brain, heart and gut, members of the nodal family of TGFβ cell-cell signaling factors, such as mouse nodal and zebrafish southpaw (spaw), are expressed in the left lateral plate mesoderm (LPM), but not the right LPM. Nodal signaling induces expression of lefty family members and the transcription factor pitx2 in the left LPM and organ primordia. Downstream asymmetrically expressed genes in organ primordia then drive asymmetric organ morphogenesis, through predominantly unknown mechanisms (Ramsdell and Yost, 1998; Hamada et al., 2002; Levin, 2005; Shiratori and Hamada, 2006).
Both left and right LPM have the ability to express nodal family members in vertebrates. For example, ectopic placement of Nodal protein in the right LPM induces expression of nodal and perturbs organ asymmetry in chick and mouse embryos (Levin et al., 1997; Yamamoto et al., 2003). Mutations or embryological manipulations that disrupt earlier steps in LR patterning often result in expression of nodal in both the left and right LPM (Lohr et al., 1998; Bisgrove et al., 2000). In zebrafish, at least three steps are required to ensure correct asymmetric expression of spaw (zebrafish nodal family member) and subsequent LR patterning. First, cilia-dependent asymmetric fluid flow in Kupffer’s vesicle (KV), located at the tail bud region, is required for the initiation of asymmetric spaw expression in LPM, as demonstrated by knockdown of genes exclusively in KV cells, not LPM (Amack and Yost, 2004; Essner et al., 2005; Amack et al., 2007). In these experiments, spaw expression is initiated in the LPM, but the orientation is “randomized”, such that individual embryos have expression in right, left, both or neither sides. This suggests that the asymmetric fluid flow biases spaw expression in the left LPM. However, because the timing of spaw expression has not been quantified, it is unknown whether laterality mutants or morphants initiate spaw earlier or later than wildtype.
The second step is the propagation of spaw expression from the site of initiation in the LPM at the anterior-posterior (AP) level of the KV, through the LPM toward the head. Since Nodal can induce the expression of itself, it is likely that nodal expression propagates from posterior to anterior via an autoregulatory mechanism. In mice, specific asymmetric cis-regulatory elements (ASE) have been identified in Nodal and Lefty genes and are responsive to the Nodal signaling pathway (Yamamoto et al., 2003). Because the rates of movement of the posterior-to-anterior (PA) wave of spaw expression have not been quantified, it is unknown whether various mutants or morphants have effects on LR patterning by altering the PA wave.
The third step is an inhibitory mechanism by which asymmetric gene expression in the left LPM is prevented from crossing over into the right LPM. The embryonic midline (notochord and/or floorplate of the neural tube) was first proposed to serve this role based on manipulations in Xenopus and zebrafish mutants (Danos and Yost, 1996). Analysis in mice indicates that lefty1 (lft1) is required in midline (Meno et al., 1998) and is induced by Nodal, probably from left LPM. Since Lft1 is a diffusible Nodal inhibitor, midline Lft1 prevents Nodal signaling in the left LPM from activating Nodal signaling in the right LPM (Nakamura et al., 2006).
To analyze the three steps described above, we have profiled the posterior to anterior temporal progression of spaw and pitx2 expression in the left LPM and lft1 in the midline. We tested three genes known to have left-right phenotypes, casanova (cas), lft1 and charon (cha). Cas is a Sox32 family transcription factor required for zebrafish endoderm development and KV formation, Lft1 is a TGFβ family Nodal signaling inhibitor expressed in the midline, and Cha is a Cerberus family Nodal signaling inhibitor expressed in peri-KV region ( Kikuchi et al., 2001; Sakaguchi et al., 2001; Feldman et al., 2002; Hashimoto et al., 2004). We found although all three morphants have the same spaw P-A propagation rates in the LPM, they have distinct molecular phenotypes that can be discerned by this quantitative method. The PA wave of spaw is initiated in cas morphants at the same time as wild type, but is initiated earlier in lft1 and cha morphants. In addition, lft1 morphants retain an intrinsic LR orientation, predominantly displaying initiation of spaw on the left side before the right side, whereas cas and cha mophants lose this orientation. Using these observations, we provide evidence that the PA wave of spaw in LPM is an instructive signal for a brain asymmetry.
In order to obtain a detailed description of the temporal and spatial expression patterns of spaw, pitx2 and lft1, zebrafish embryos were collected at different stages from 13–18 somite stages (SS) and analyzed by whole mount in situ hybridization with a mixture of spaw/lft1/lefty2/myoD (Fig. 1A and D), pitx2/myoD (Fig. 1B) or spaw/lft1 (Fig. 1C) probes. The developmental stage of each embryo was recorded as the total number of somite pairs labeled by the myoD probe. The position of the anterior boundary of each gene expression pattern was recorded as the somite number at the same position along the anterior-posterior (AP) axis. For example, the embryo in Fig. 1A had 15 total somite pairs and the anterior expression boundary of spaw was at the 7th somite. To quantitatively analyze the dynamic PA propagation of spaw, pitx2 and lft1, the average somite position of anterior gene expression boundary was calculated and plotted for each gene at each somite stage (SS) from 13 to 17 SS (Fig. 1E). The slope of this plot reflects the rate of PA propagation for each gene. For example, the PA progression of spaw was 2.3 somite lengths per somite generation time, during the period from 13–17 SS.
LPM spaw expression was detectable at approximately 12 SS (Long et al., 2003) at the AP level of the tail bud. Most embryos displayed strong expression of spaw in the left LPM by 13 SS. At 15 SS, the boundary was at the 7th somite (Fig. 1A, 1E). By 18 SS, the anterior boundary of spaw expression was anterior to the level of the first somite, making measurement of anterior boundary more difficult. The expression of spaw continued its PA propagation until it reached the anterior end of LPM at approximately 22 SS stage (data not shown).
The expression of pitx2 was detectable at 13 SS. At 15 SS, the anterior expression boundary of pitx2 was at the 10th somite (Fig. 1B). At 17SS, the anterior expression boundary of pitx2 in LPM was adjacent to the 2nd somite (Fig. 1E). Subsequent expression of pitx2 in LPM more anterior than the first somite was very weak and difficult to reliably detect by in situ analysis (data not shown). Thus, pitx2 had an expression pattern in LPM similar to spaw, propagating anteriorly, but with a lag time of 2 somite stages (Fig. 1E), consistent with the suggestion that Spaw induces pitx2 expression in LPM.
The expression of lft1 in the midline was bi-phasic. In the first phase, lft1 was expressed in the posterior midline at 10 SS, before detectable spaw expression in LPM (Fig. 1C). This posterior expression domain was static and did not propagate anteriorly. In the second phase, beginning at 15 SS, the expression of midline lft1 propagated anteriorly (Fig. 1D, 1E), and the anterior boundary of the lft1 PA wave in the midline was always more posterior than the anterior boundary of spaw and pitx2 in LPM (Fig. 1D, 1E). This suggests that the PA propagation of lft1 in the midline is subordinate to the PA propagation of spaw and/or pitx2 in LPM. The expression of lft1 reached the anterior end of the midline at about 20–22 SS (data not shown).
There was approximately a 2 somite length difference between the anterior boundary of spaw and that of pitx2, indicating that the initiation of pitx2 occurred approximately one hour (2 somite stages) after spaw initiation (Fig. 1E). Similar, lft1 PA wave in the midline occurred approximately 1 hour behind pitx2, although the first phase of posterior midline lft1 was expressed before the LPM spaw expression (Fig. 1E). Thus, the timing of initiation of these three genes was different, with spaw first, followed by pitx2, then the second phase of lft1 in the midline. However, once the PA propagations for each of these genes were initiated, the propagation rates of all three genes were approximately equivalent, 2.3 somite lengths per somite generation time, as reflected in the comparable slopes in Fig. 1E.
Both pitx2 and lft1 can be induced by Nodal signaling pathway. To test whether the expression patterns of these two genes depends on Spaw, we injected a spaw translation-blocking MO (Long et al., 2003) into zebrafish embryos at one cell stage and examined the expression of pitx2 and lft1 in morphants at several stages of development. Knockdown of Spaw resulted in the absence of detectable spaw (n=25/25) and pitx2 (n=22/22) mRNA in LPM, throughout the stages of initiation and PA propagation (data not shown, Fig. 2A and B). In contrast, the symmetrical peri-KV domains of spaw, analogous to “perinodal” expression of nodal in mouse (Lowe et al., 1996), were expressed in spaw morphants (n=25/25) at 16 SS (Fig. 2C and D). This indicates that the initiation and maintenance of peri-KV spaw expression is not dependent on Spaw function.
In the midline, both phases of lft1 expression were absent in spaw morphants (Fig. 2C and D, data not shown) (n=25). The early expression of lft1 in the posterior midline is likely induced by Spaw from the peri-KV region, whereas Spaw from the LPM is not involved because it is not expressed at this stage (green arrows, Fig 2C). This is consistent with a current model that proposes the midline lft1 is initially induced by a nodal family member emanating from the node (Nakamura et al., 2006).
These results indicate that Spaw is necessary for the expression of pitx2 in LPM and for both phases of lft1 expression in the midline. We propose that the PA propagation of spaw in LPM controls the propagation of pitx2 in LPM and the second phase of lft1 in the midline. This explanation is consistent with our spatio-temporal measurements (Figure 1E) that indicate that the initiation of PA propagation of spaw occurs first, followed by pitx2 in LPM then lft1 in the midline. In addition, the propagation speeds of pitx2 and lft1 are similar to the speed of spaw propagation in LPM (slopes in Fig 1E), suggesting LPM spaw is the pace maker for both pitx2 and lft1 propagation.
As reported earlier, spaw MO injection blocked LPM spaw expression (Long et al., 2003). The most parsimonious explanation for the absence of asymmetric gene expression in LPM is that the initiation step is dependent on the function of Spaw protein expressed in the peri-KV domains. However, it is also possible that transcription of spaw is transiently initiated in LPM but requires a Spaw-dependent autoregulatory mechanism to maintain expression. In contrast, expression and maintenance of spaw in bilaterally symmetric peri-KV domains is independent of Spaw function, suggesting that some other signaling pathway, distinct from Spaw, is required for induction and maintenance of this spaw domain.
Several genes are known to be upstream of asymmetric LPM gene expression, but it is not known whether they regulate the timing of initiation or the propagation rate of spaw in LPM. We chose well characterized morpholinos to knock down three genes, cas, lft1 and cha. Current standard analysis of morphants or mutants uses four categories of gene expression in LPM, left-sided (normal) expression, right-sided (reversed) expression, bilateral (expression in both sides) expression and absent expression (Bisgrove et al., 2000). The injection of any of above MOs led to predominantly bilateral spaw expression according to this analysis (Fig. 3A). However, since expression patterns are typically assayed late in somitogenesis, it is unknown whether the primary alterations in LPM expression are due to perturbations in the timing of initiation, PA propagation rates, or both, and whether these processes occur independently in the left and right LPM. Therefore, using our spatio-temporal analysis, we assessed stage-specific LPM expression patterns in embryos injected with these three morpholinos. Our quantitative spatio-temporal analyses revealed three categorical distinctions among bilateral morphants.
First, it is striking that “bilateral expression” is not truly symmetrical in any of these morphants: the anterior spaw expression boundaries in the left or right LPM were often not at the same position along the anterior-posterior axis at a given stage, suggesting that PA wave is not parallel in the left and right sides. We classified bilateral spaw expression into three groups: left side leading (Fig. 3B), equivalent (Fig. 3C) and right side leading (Fig. 3D). Although each morphant resulted in predominantly bilateral expression, the proportions of embryos with left side leading, equivalent or right side leading LPM spaw expression were strikingly different when comparing cas, cha, and lft1 morphants. A good proportion of cas and cha morphants, 48% and 30% respectively, have equivalent anterior boundaries of the left and right side PA waves, as well as 21% and 17% in which the right-side PA wave precedes that on the left-side. This suggests that the side on which spaw is first expressed is random in cas and cha morphants. In contrast, in most lft1 morphants (85%) the PA wave on the left precedes the PA wave on the right, suggesting lft1 morphants retain a left side bias.
Second, by plotting the PA propagation of the leading side, we can estimate the timing of earliest initiation, regardless of side of initiation, and the rate of propagation. We found the rates at which the leading side of spaw expression was propagated from posterior to anterior were similar in each morphant (Fig. 4). The spaw expression profile of Control MO (Fig. 4A) was indistinguishable from wild type, indicating MO injection itself does not alter the spaw expression profile in LPM. For cas MO, lft1 MO and cha MO, once spaw was initiated, either in the left LPM or right LPM, the rate of PA propagation was similar to wildtype controls. To further confirm these results, we calculated the slopes of each curve (Fig. 4B–D). For cas MO, the slope of the curve indicated that the PA propagation rate was a distance of 2.5 somite lengths per somite generation. The WT control for these morphants was also 2.5. Lefty MO (2.5), WT controls (2.3), cha MO (2.6) and corresponding WT (2.3) had similar slopes. The curves for the trailing side (data not shown) had comparable rates of PA propagation, with initiation either earlier or similar to wildtype. From this we conclude that knockdown of Cas, Cha or Lft1 does not alter the mechanisms by which spaw expression is propagated from posterior to anterior.
In contrast to PA propagation rates, there were striking differences in the apparent timing of spaw initiation in each morphant. In cas morphants, the leading edge of PA wave at specific stages was identical to that of wildtype controls (Fig. 4B), suggesting spaw was initiated at the same time as wildtype. For example, in cas morphants at 13 SS, the leading edge was at the PA position of the 11th somite, as in wildtype. Since cas morphants lack endoderm (Kikuchi et al., 2001; Sakaguchi et al., 2001), it can be concluded that endoderm is not required for the normal timing of spaw initiation or for the PA propagation of spaw in LPM. Cas morphants also lack a functional KV (Essner et al., 2005), and have only remnants of the KV precursor cells called dorsal forerunner cells (Essner et al., 2002; Essner et al., 2005). Thus, our results suggest that the timing of spaw initiation and propagation in LPM does not require a functional KV, and that other mechanisms control these processes. Strikingly, the similar proportions of left-leading and right-leading phenotypes suggest that spaw is initiated stochastically and the mechanisms that establish laterality are obliterated in cas morphants (Fig. 3E). This suggests that the primary function of the KV is to superimpose a left-right asymmetry onto an underlying, but independent, mechanism that initiates and propagates spaw PA wave in LPM.
In contrast, the leading edge of the PA wave in lft1 and cha morphants was significantly ahead of the leading edge in control embryos (Fig. 4C, D), suggesting the PA wave is initiated earlier in these morpholino injected embryos compared to wildtype. lft1 morphants serve as an extreme example. At 13 SS, when the PA wave had reached only the level of the 11th somite in control embryos, it was already at the 5th somite in lft1 morphant. In wildtype embryos the PA wave did not reach this anterior level until 16 SS, at which stage the PA wave was significantly anterior to the first somite in lft1 morphants. Given that the rates of PA propagation were similar in morphants and wildtype embryos (similar slopes, Fig. 4C), this suggests that spaw expression is initiated at least four to five somite stages approximately two hours earlier in lft1 morphants compared to wildtype embryos.
Both Lft1 and Cha are Spaw inhibitors in zebrafish (Feldman et al., 2002, Hashimoto et al., 2004). From the observation that both lft1 and cha morphants have earlier spaw initiation in LPM (Fig. 4C, D), we speculate that removal of these inhibitors from the peri-KV region allows signals from the KV (perhaps peri-KV Spaw protein) to reach the LPM at a faster rate, or in higher effective doses. However, there is an important distinction in the spaw expression profiles in lft1 and cha morphants. 48% of cha morphants had either bilateral equivalent or right leading spaw expression (Fig. 3E) in contrast to lft1 morphants, which had only 14% bilateral equivalent or right leading (Fig. 3E). Thus, cha morphants lost a left-side biased spaw expression, while lft1 morphants still retain the bias. These results and the bilateral expression of cha in peri-KV domain (Hashimoto et al., 2004) suggest that Cha may be important for modulating LR asymmetry signals, emanating from KV. In contrast, Lft1 does not appear to be involved in establishing LR asymmetry signals.
The results of lft1 morphants are consistent with the observations in lft1 mouse mutants, in which most embryos show the left LPM nodal expression before the right side nodal expression (Meno et al., 1998). The most likely explanation for the eventual bilateral symmetry of spaw is that lft1 serves as an inhibitor or barrier in the midline. We postulated that if spaw expression in lft1 morphants was initiated earlier in the left side than that in the right side, we should be able to find embryos with spaw expression only in one side, predominantly the left side. In embryos collected at 9–11 SS, 28 embryos had spaw expression in the left side only (Fig. 3F), vs just 5 embryos (Fig. 3G) with right sided only spaw expression. The rest of the embryos had either no spaw LPM expression at all (n=82) or spaw expression on both sides (n=9). The relative large number of embryos with no spaw expression and relative small number of embryos with bilateral spaw expression were consistent with the proposal that lft1 morphants at 9–11 SS are on the cusp of initiating spaw in LPM. Wildtype embryos (98%, n=88) didn’t express spaw at this stage (Fig. 3H), consistent with the notion that lft1 morphants initiate spaw expression earlier. At late stages (15–17 SS), 100% (n=99) of the lft1 morphants had bilateral spaw expression, compared to 98% (n=104) of the wildetype had left only spaw expression. These observations suggest that at spaw LPM initiation stage, most lft1 morphants retained a normal LR axis and were able to initiate spaw expression in the LPM earlier than the right LPM, and that midline lft1 per se does not affect the laterality of spaw initiation in LPM, but it serves as an inhibitor to prevent subsequent right side spaw initiation.
A previous study indicated that spaw MO injected embryos do not express brain left-right asymmetric marker, lft1 (Long et al., 2003), suggesting Spaw influences brain LR asymmetry. However, it is not clear whether spaw expression in LPM affects gene expression in the brain. It is also not clear if the timing of the PA wave influences organ asymmetry. One advantage of the lft1 morphants described here is that the PA wave of spaw is delivered to the anterior end of the LPM at significantly earlier stages than wildtype. We postulate that if spaw in LPM controls brain asymmetry, early arrival of Spaw in anterior LPM might alter the timing of LR patterning in the brain. To test this, lft1 morphants at 17–19 SS stages were assessed by whole mount in situ hybridization with spaw/lft1/lefty2/myoD probes. Besides expression in the midline, lft1 also serves as a left side brain asymmetry marker (Bisgrove et al., 2000). At this stage, no uninjected embryos expressed lft1 in the brain (0%, n=52). However, a substantial portion of lft1 morphants (36%, n=85) had lft1 expression in the brain, suggesting the early arrival of spaw in anterior LPM induces the precocious expression of this brain asymmetry marker. In the morphants expressing brain lft1, most had bilaterally equivalent spaw expression in the LPM, with a corresponding bilaterally symmetric expression of lft1 in the brain (n=16) (Fig. 5A). This correlation of spaw expression in LPM and lft1 in the brain is strengthened by examining embryos with asymmetric spaw expression in the anterior LPM. In morphants with stronger spaw expression on the left side, suggesting spaw on the left side reached anterior part earlier than right side, lft1 expression was always in the left brain only (n=12/12, Fig. 5B). In morphants with stronger spaw expression in the right LPM, lft1 expression was only in the right brain (n=3/3, Fig. 5C). Thus, the side that has spaw LPM expression earlier also induces asymmetric lft1 brain earlier. These examples of early asymmetric lft1 brain expression stand in striking contrast to lft1 morphants examined at the normal time of lft1 brain expression (20–24 SS) in which almost all of the morphants display bilateral lft1 expression, with no examples of reversed expression (Amack and Yost, 2004). Together, these suggests that lft1 morphants that have precocious asymmetric lft1 expression in the brain (38% left sided and 10% reversed) at earlier stages would later resolve into bilateral lft1 expression, and that lft1 expressing cells have a developmental competence to receive inducing signals from LPM for an extended period, starting at least at 17 SS. While these data are only correlative, they are consistent with the proposal that Spaw in the anterior LPM is an instructive signal for lft1 expression in brain.
Zebrafish embryos collected at different stages were fixed in 4% Formaldehyde. Whole mount in situ was carried out as described (Bisgrove et al., 2000). Following probes were used: spaw (Long et al., 2003), cha (Hashimoto et al., 2004), myoD (Weinberg et al., 1996), lft1 (Bisgrove et al., 1999), lefty2 (lft2) (Bisgrove et al., 1999) and pitx2 (Essner et al., 2000). Probe for lft2 was included in some cases, but at the stages analyzed, lft2 expression either does not occur or is occluded by spaw expression.
Morpholinos used in this study are: spaw MO (5′-GCACGCTATGACTGGCTGCATTGCG-3′) (Long et al., 2003); cha MO (5′-CAAAAAAGCCGACCTGAAAAGTCAT-3′) (Hashimoto et al., 2004); lft1 MO (5′-GGCGCGGACTGAAGTCATCTTTTCA-3′) (Feldman et al., 2002) and casanova MO (5′-CAGGGAGCATCCGGTCGAGATACAT) (Sakaguchi et al., 2001). MOs were injected at the 1–4 cell stages. Roughly 1 nl of MO was injected into the yolk just beneath the cell. The MO concentration injected for each MO is: spaw MO 1mM, cha MO 0.5 mM, lft1 MO 0.4 mM and cas MO 0.2mM.
We thank M. Karthikeyan and J. Shen for technical support and BW. Bisgrove for comments. This work was supported by a grant from NIHLB to HJY.