Pitx controls amphioxus asymmetric morphogenesis by promoting left-side development and repressing right-side formation

Left-right (LR) asymmetry is an essential feature of bilateral animals. Studies in vertebrates show that LR asymmetry formation comprises three major steps: symmetry breaking, asymmetric gene expression, and LR morphogenesis. Although much progress has been made in the first two events, mechanisms underlying asymmetric morphogenesis remain largely unknown due to the complex developmental processes deployed by vertebrate organs. We here addressed this question by studying Pitx gene function in the basal chordate amphioxus whose asymmetric organogenesis, unlike that in vertebrates, occurs essentially in situ and does not rely on cell migration. Pitx null mutation in amphioxus causes loss of all left-sided organs and incomplete ectopic formation of all right-sided organs on the left side, whereas Pitx partial loss-of-function leads to milder phenotypes with only some LR organs lost or ectopically formed. At the N1 to N3 stages, Pitx expression is gradually expanded from the dorsal anterior domain to surrounding regions. This leads to activation of genes like Lhx3 and/or Prop1 and Pit, which are essential for left-side organs, and downregulation of genes like Hex and/or Nkx2.1 and FoxE4, which are required for right-side organs to form ectopically on the left side. In Pitx mutants, the left-side expressed genes are not activated, while the right-side genes fail to decrease expression on the left side. In contrast, in embryos overexpressing Pitx genes, the left-side genes are induced ectopically on the right side, and the right-side genes are inhibited. Several Pitx binding sites are identified in the upstream sequences of the left-side and right-side genes which are essential for activation of the former and repression of the latter by Pitx. Our results demonstrate that (1) Pitx is a major (although not the only) determinant of asymmetric morphogenesis in amphioxus, (2) the development of different LR organs have distinct requirements for Pitx activity, and (3) Pitx controls amphioxus LR morphogenesis probably through inducing left-side organs and inhibiting right-side organs directly. These findings show much more dependence of LR organogenesis on Pitx in amphioxus than in vertebrates. They also provide insight into the molecular developmental mechanism of some vertebrate LR organs like the lungs and atria, since they show a right-isomerism phenotype in Pitx2 knockout mice like right-sided organs in Pitx mutant amphioxus. Our results also explain why some organs like the adenohypophysis are asymmetrically located in amphioxus but symmetrically positioned in vertebrates.

Conclusions: Our results demonstrate that (1) Pitx is a major (although not the only) determinant of asymmetric morphogenesis in amphioxus, (2) the development of different LR organs have distinct requirements for Pitx activity, and (3) Pitx controls amphioxus LR morphogenesis probably through inducing left-side organs and inhibiting right-side organs directly. These findings show much more dependence of LR organogenesis on Pitx in amphioxus than in vertebrates. They also provide insight into the molecular developmental mechanism of some vertebrate LR organs like the lungs and atria, since they show a right-isomerism phenotype in Pitx2 knockout mice like right-sided organs in Pitx mutant amphioxus. Our results also explain why some organs like the adenohypophysis are asymmetrically located in amphioxus but symmetrically positioned in vertebrates.

Background
While vertebrates exhibit external left-right (LR) symmetry, their visceral organs are LR asymmetric in terms of shape, size, position, and rotation direction [1]. For example, the human heart, spleen, and stomach are offset to the left while the gall bladder and liver sit to the right, and the left and right lung and kidney are of different size and shape [2]. Establishment of LR asymmetry is highly conserved among vertebrates and comprises three major steps: symmetry breaking, asymmetric gene expression, and LR morphogenesis [1,3]. Symmetry breaking in most vertebrate groups is initiated by a cilia-driven fluid flow present in the LR organizer of early somite stage embryos [4]. Influenced by this flow, the Dand5 gene first exhibits an asymmetric expression pattern (R > L) through unilateral mRNA decay in cells around the LR organizer. The Dand5 protein is a Nodal signaling inhibitor, thus its right side-biased expression leads to R < L asymmetric Nodal signaling at the LR organizer. The Nodal signaling pathway is transferred further to the left lateral plate mesoderm (LPM) from the organizer and induces asymmetric expression of Nodal itself, followed by Lefty and Pitx2 [5].
Although progress has been made in elucidating the upstream patterning events, mechanisms controlling LR morphogenesis of visceral organs remain largely unknown [3]. This has been complicated by the complex developmental process of vertebrate LR organs, which involves many asymmetric cell behaviors and coordination of various organs within the cavity [3]. Among genes induced by the Nodal signaling pathway, Pitx2 encodes a homeodomain-containing transcription factor and its asymmetric expression persists in developing visceral organs after asymmetric Nodal expression disappears [6]. Therefore, Pitx2 was thought to play an essential role in asymmetric organogenesis. Consistently, knockout of Pitx2 in mice or knockdown in chicken and frogs cause laterality defects in many visceral organs, including randomized positioning of several visceral organs and right isomerism of the lungs and atrium [7][8][9][10]. In mice, development of different LR organs requires distinct dosages of Pitx2 activity [11][12][13]. During mouse and chicken gut development, Pitx2 induces asymmetric cell condensation of dorsal mesentery (DM) attached to the gut by targeting the non-canonical Wnt signaling mediator Daam2, resulting in the tilting of the gut tube toward the left [14]. However, Pitx2 appears to play a very limited role in zebrafish LR development, since its mutation or knockdown causes either no or very weak LR defects [15][16][17]. This indicates different requirements of Pitx2 for LR development among vertebrates. While the development of most LR asymmetric organs relies on both Nodal and Pitx2, some asymmetric morphogenetic events depend on Nodal but not Pitx2. For example, in mice, Nodal deficiency randomizes directions of heart looping, axial rotation, and stomach sidedness [18], but none of these events are affected in Pitx2 null mutants [11,12]. This indicates genes other than Pitx2 downstream of the Nodal signaling pathway also play certain roles in LR morphogenesis in vertebrates [19]. Moreover, a recent study suggests that heart looping in zebrafish, chick, and mouse is controlled by a BMP-Prrx1/Snail1 cascade on the right side [20,21]. However, this suggestion has been questioned by a later study, which shows that neither single nor double mutants of Prrx1a and Prrx1b genes cause LR defects in zebrafish embryos [22]. This leaves it uncertain as to whether vertebrates have a right-side determinant signal.
Orthologues of the Nodal and Pitx2 gene have been identified in several diverse groups of invertebrate bilaterians [23]. In most of these animals, the two genes show asymmetric expression along the LR axis. In addition, inhibiting the Nodal signaling pathway also abolishes Pitx expression and leads to defects in LR establishment [24][25][26][27][28]. This demonstrates a certain level of conservation for the genetic control of LR development between invertebrates and vertebrates. However, as in vertebrates, molecular mechanisms underlying LR morphogenesis in invertebrates are largely unknown. In addition, whether Pitx is required for LR asymmetry in invertebrates has not been functionally tested. To address these questions, we here analyzed Pitx gene function in the LR morphogenesis of the cephalochordate amphioxus. Amphioxus shows pronounced LR asymmetries in its pharynx at larval stages, with the mouth, preoral pit and the duct of the club-shaped gland (hereafter refer to as CSG) forming on the left side, and the gill slits, endostyle, and the glandular region of the CSG on the right side [28,29]. The arrangement of the somites and peripheral nerves is also asymmetric in amphioxus beginning from the N3 stage. Importantly, unlike vertebrates, LR organogenesis of amphioxus occurs essentially in situ and does not rely on asymmetric cell behaviors, and knockout of genes essential for early LR patterning in amphioxus leads to simply either two-right-side (right isomerism) or two-left-side (left isomerism) phenotypes [24,28,[30][31][32][33]. Despite these differences, the molecular mechanisms underlying the early development of LR asymmetry in amphioxus are very similar to that in vertebrates, as they also require cilia movement and Dand5, Nodal, and Lefty genes [24,28,[30][31][32]. This, together with its position among chordates, indicates that amphioxus is a promising organism for understanding mechanisms and evolution of vertebrate LR morphogenesis.
We here show that Pitx is a major (although not the only) determinant in amphioxus LR morphogenesis, with its mutation causing loss of all left-side organs and ectopic formation of all right-side organs on the left side. We further demonstrate that different LR organs of amphioxus embryos have distinct requirements for Pitx dosage. Finally, we show that Pitx executes its functions in amphioxus LR morphogenesis probably through directly inducing left-side organs and inhibiting right-side structures. These results demonstrate a conserved role for Pitx in LR development among chordates, and also shed important insights into the molecular mechanisms underlying animal LR morphogenesis.

Pitx encodes two isoforms with different expression patterns in amphioxus embryos
Vertebrates have three highly conserved Pitx2 mRNA isoforms: Pitx2a, Pitx2b, and Pitx2c. Among them, Pitx2c is asymmetrically expressed, while the other two, transcribed from an alternative promoter, are not [34]. Two corresponding isoforms (Pitxa/b and Pitxc) have also been reported in tunicate embryos [35]. Our analysis revealed two isoforms for the amphioxus Pitx gene that were transcribed from different transcription initiation sites and were interrupted by two and three introns respectively (Fig. 1a). As they share similar exon/ intron structures to that of the two tunicate Pitx isoforms, we named them Pitxa/b and Pitxc respectively.
The expression patterns of Pitxa/b and Pitxc in amphioxus embryos was analyzed using in situ hybridization with probes corresponding to their specific exons (Fig. 1b-g'). Pitxa/b expression was first detected in two patches of cells in the neural ectoderm of embryos at N4 stage (Fig. 1c, c'). Double in situ analysis revealed that these Pitxa/b-positive cells were located more posteriorly than the dorsal Hesse organ expressing Mop (Additional file 1: Fig. S1). At T1 stage, a new Pitxa/b expression domain was observed in the posterior half of the preoral pit (Fig. 1d, d'). Pitxc expression was detected in the left dorsal side of embryos at N1 stage (Fig. 1e, e'), and at N4 and T1 stages, Pitxc was strongly expressed in the anterior left pharyngeal region and weakly in the left side of the tailbud ( Fig. 1f-g'). These results Fig. 1 Expression patterns of amphioxus Pitx isoforms and TALEN target sites of Pitx gene. a Amphioxus Pitx isoforms. White boxes represent untranslated regions (UTR) and black and red boxes (Pitx homeodomain) represent coding sequences. Broken lines indicate introns. We obtained the partial sequence of Pitxa/b according to available amphioxus EST data and the predicted remaining sequence is supplemented as a dotted slashed box (coding sequence) and dotted white box (UTR). b-b' Expression of Pitxa/b in left lateral (b-d) and dorsal (b'-d') views. e-g' Expression of Pitxc in left lateral (e-g) and dorsal (e'-g') views. Blue lines in a show the location of Pitxc or Pitxa/b probes and green lines in a show the location of Pitx probe used for whole-mount in situ hybridization (WISH). h Numbers 1, 2, and 3 in a show the location of TALEN target sites in Pitxc and Pitxa/b. Binding sites for TALEN pairs [forward (Fw) and reverse (Rv)] used in this study are highlighted in gray. The SacI, AatII, and TatI restriction site in the spacer is underlined respectively. WT, wild type sequence. A, anterior; P, posterior; D, dorsal; V, ventral; L, left side; R; right side. Scale bars in (b-g') are 50 μm demonstrated that Pitx in amphioxus also encodes different isoforms with distinct expression patterns, similar to Pitx in vertebrates and tunicates.
Pitx mutation affects amphioxus development while Pitxc mutation does not To study the function of Pitx in amphioxus development, we generated Pitx mutants using the TALENmediated genome editing method [36,37]. Three pairs of TALEN constructs (TALEN1-3) were assembled, which targeted the first coding exon of Pitxc isoform and two other coding exons shared by Pitxc and Pitxa/b (Fig. 1h). Injection of TALEN mRNAs into amphioxus embryos revealed efficient mutation (30-50%) at each of the target sites (Additional file 1: Fig. S2). Using these TALENs, we obtained four Pitx mutant lines (19-bp and 1-bp deletion for TALEN1, 13-bp deletion for TALEN2, and 18-bp insertion plus 7-bp deletion for TALEN3) (Fig. 1h). We further crossed these heterozygous animals (for TALEN1, females carrying the 19-bp deletion and males carrying the 1-bp deletion were used) to examine if the mutations affected embryonic development. Unexpectedly, we observed no LR or other defects in embryos derived from crosses between animals carrying mutations at the TALEN1 site ( Fig. 2a-b'). Genotyping of 12 randomly selected 3-gill slit larvae (L3 stage) from the cross-identified 4 homozygous and 8 heterozygous/WT (wild type) individuals (Additional file 1: Fig. S3). The homozygous mutants can normally survive to adulthood. In contrast, we observed around 25% (136/542 for the TALEN2 site and 141/550 for the TALEN3 site) of embryos showing specific and identical phenotypes in the anterior pharyngeal region at L3 stage, from crosses of animals with mutations at the TALEN2 or TALEN3 site. Genotyping analysis confirmed that all individuals showing these phenotypes were homozygous mutants, while individuals showing normal morphology were either Pitx heterozygotes or wild type (Additional file 1: Fig. S4, S5). Homozygous Pitx TALEN2 or TALEN3 mutants died at L4 stage. Compared to WT/heterozygotes larvae (Fig. 2c, c', e, e'), homozygous Pitx TALEN2 or TALEN3 mutants lacked a mouth opening, preoral pit, and CSG duct on the left side, and developed an endostyle and the glandular region of the CSG on each side (Fig. 2d, d', f, f'). The gill slits, which first opened at the ventral midline and then extended up to the right side of the head in wild type larvae (Fig. 2c', e'), were also affected in the two Pitx mutants: they appeared at the ventral side but failed to open and extended up to the right side ( Fig. 2d', f'). Notably, these phenotypes were very similar to that observed in amphioxus larvae lacking the Nodal signaling pathway activity [28]. Since Pitx TALEN2 and TALEN3 mutants showed identical phenotype, Pitx TALEN2 mutants were used for further analysis. To simplify wording, we refer to Pitx TALEN2 mutants as Pitx mutants and those carrying mutations at the TALEN1 site as Pitxc mutants hereafter.
Pitx mutants lack left-side organs but form incomplete right-side organs on the left side The finding that Pitx mutants in amphioxus showed a symmetrical pharyngeal phenotype as observed in embryos deficient of the Nodal signaling pathway is unexpected, as Pitx is generally thought as a determinant of left-side development but not a LR patterning gene. To verify this, we examined expression of marker genes of pharyngeal organs including the preoral pit, mouth, CSG, and endostyle in N5 and T1 embryos. In WT and Pitx +/− embryos, Lhx3, Prop1, and Pit were expressed in the left anterior endoderm where the preoral pit will later form ( Fig. 3a-f, arrowheads). Consistent with loss of the preoral pit in Pitx −/− larvae, expression of these genes in the prospective preoral pit region were lost in Pitx −/− embryos ( Fig. 3a'-f'). Likewise, Pou4, which is normally expressed in the developing mouth on the left side (Fig. 3g, arrow), also lost its expression in Pitx −/− mutants (Fig. 3g'). Nkx2.1 and Hex which are expressed in the endostyle are asymmetrically expressed in WT and Pitx +/− embryos in a L < R manner ( Fig. 3h-k). In Pitx −/− mutants, Nkx2.1 and Hex expression became symmetric, with the expression pattern on both sides being similar to that in the right side of wild type embryos ( Fig. 3h'-k'). However, careful examination revealed that Nkx2.1 expression on the right side ( Fig. 3i', black arrowhead) was not identical to its expression on the left side ( Fig. 3i', red purple arrowhead): the former was extended more posteriorly than the latter (Fig. 3i'). This indicated that the endostyle region was almost but not fully duplicated on the left side of Pitx −/− mutants. A similar case was also observed in the CSG of Pitx −/− mutants. In WT and Pitx +/− embryos, FoxE4 was expressed in the whole CSG including the right-sided glandular region and the left-sided duct (Fig. 3l, m), while Krox was transcribed only in the glandular region (Fig. 3n, o). In Pitx −/− embryos, although FoxE4 expression on the right side seemed to be fully duplicated on the left side ( Fig. 3l', m'), Krox expression was not since its expression domain on the left side was slightly smaller than that on the right side (Fig. 3n', o'). These results demonstrated that the pharyngeal structure of Pitx −/− mutants was not fully symmetrical unlike that observed in embryos lacking Nodal signaling pathway [28]. Consistent with this, asymmetric expression patterns of the LR patterning genes, like Dand5, Nodal, Lefty, and Pitx were not altered in Pitx −/− embryos at N0 (Additional file 1: Fig Fig. S6d', g', j') and Nodal, Lefty, and Pitx expression in the forming preoral pit was lost at T1 stage (Additional file 1: Fig. S6h', i', j').
In addition to the LR organs examined above, Hatschek's nephridium, hematopoietic domains, and Hatschek's right diverticulum (short for HRD) are also positioned asymmetrically on one side of amphioxus embryos. Among them, the HRD is derived from the right anterior-dorsal endoderm, and Hatschek's nephridium and hematopoietic domains are developed from the left first and right first somites, respectively [38,39]. To evaluate if the development or positioning of these organs were disrupted due to Pitx loss-offunction, we analyzed genes specifically expressed in them in Pitx −/− mutants at N5 stage. In WT and Pitx +/− embryos, Gata1/2/3, Pdvegfr, and Scl exhibited two bilateral and slightly asymmetrical expression domains, which respectively denoted left-side Hatschek's nephridium (Fig. 4c, e, g, blue arrowhead) and the right-side hematopoietic domain (Fig.  4c, e, g, black arrowhead) [38]. In Pitx −/− mutant embryos, however, their expression was bilaterally symmetric (Fig. 4d, f, h, black arrowhead). To determine the identity of the Fig. 2 Pitx mutants phenotype. In Pitx TALEN1 +/+; +/− larvae (a, a'), Pitx TALEN2 +/+; +/− larvae (c, c'), and Pitx TALEN3 +/+; +/− larvae (e, e'), the preoral pit and mouth are formed on the left side, with the endostyle, club-shaped gland (CSG) and first gill slit developing on the right side. Pitx and Pitx TALEN3 −/− mutants (f, f') lack a mouth opening and preoral pit on the left side, forming an endostyle and CSG on each side. The larvae at L3 (3 gill slits larva) stage are observed in lateral views in (a, b, c, d, e, f) and dorsal views in (a', b', c', e', f'). m, mouth; pp: preoral pit; en, endostyle; csg, club-shaped gland; g: gill slit. A, anterior; P, posterior; D, dorsal; V, ventral; L, left side; R; right side. Scale bars:50 μm expression domains, we analyzed Pax2/5/8 gene, which is expressed in Hatschek's nephridium but not in the hematopoietic center [40]. We found that Pax2/5/8 expression in the prospective nephridium is absent in Pitx −/− mutants (Additional file 1: Fig. S7c, d). This result indicated that Pitx −/− mutants lacked Hatschek's nephridium but formed a hematopoietic domain on each side. We also noticed a Gata1/2/3 expression domain in the right-side HRD in normal embryos (Fig. 4c, arrow) and found that this domain was present on both sides of Pitx −/− mutants (Fig. 4d, arrow). Moreover, asymmetric expression (R > L) of the Hand gene in the lateral/ventral mesoderm derived from amphioxus somites [38] became bilaterally expressed in Pitx −/− mutants although still in a R > L pattern (Fig. 4b).
Together, the above results indicate that in Pitx mutants, all left-side organs are lost while the rightside organs form normally on the right side and ectopically (albeit not incompletely) on the left side.
Pitx acts downstream of Nodal signaling to control asymmetric arrangement of amphioxus somites and peripheral nerves Different from vertebrates, amphioxus somites and peripheral nerves (aligned with the somites) are asymmetrically arranged along the LR axis, with the left ones being offset about half a segment anterior to the right ones [41]. These asymmetries become first recognizable at mid-neurula (with 8-10 somites) when somitogenesis shifts from a symmetrical way by enterocoelous outpocketing of paraxial mesoderm, to a left-to-right alternate fashion by pinching off from the tailbud [42]. To assay if the arrangement of the somites and peripheral nerves were affected in Pitx −/− mutants, we examined expression of the m-actin gene, a marker of the somites at stage N5, and acetylated tubulin, a marker of the peripheral nerves at L3 stage [28]. We found both markers were asymmetrically expressed in the mutants, similar to that observed in WT and heterozygous embryos (Fig. 4i, k). However, compared to WT and heterozygous  Significance determined by t-test is indicated by asterisks: **P < 0.01. p-s The arrangement of posterior somites is marked by m-actin. Dashed lines mark outlines of newly budded somites. o The yellow ellipses indicate somites and red lines indicate peripheral nerves. no, notochord, nt, neural tube. All the embryos in above images are dorsal views. Numbers in the bottom right corner of a panel show the number of times the phenotype was seen, out of the total number of embryos from that genotype analyzed (a-l), or the phenotype was observed in the total number of embryos examined (p-s). A, anterior; P, posterior; L, left side; R; right side. Scale bars: 50 μm embryos, the mutants exhibited significantly shorter distances between the left somites or peripheral nerves and their corresponding ones on the right side (Fig. 4j, l, m, n). These results showed that the asymmetric arrangement of the somites and peripheral nerves was weakened but not abolished in Pitx −/− mutants.
Between N5 and early larvae stages, both Nodal and Pitx are constantly expressed in the left side of the amphioxus tailbud (Additional file 1: Fig. S6e, g, h, j, arrowheads). We speculated that Nodal signaling pathway acts upstream of Pitx to determine somite asymmetry. To test this, we treated embryos with the Nodal signaling inhibitor SB505124 in different developmental windows and examined Pitx expression as well as the asymmetry of newly formed somites. To exclude the influence of Nodal function in early LR patterning on the analysis, we treated embryos with the drug after the N4 stage when the LR axis is thought to have been fixed [28,33], namely from 8S (neurula with 8 somites) to 11S (neurula with 11 somites), and from 10S (neurula with 10 somites) to around T0. As expected, both treatments eliminated Pitx expression in the tailbud of the treated embryos (Additional file 1: Fig. S8a'-d') and abolished the asymmetry of the newly budded somites (Fig. 4p-s). These results indicated that the Nodal signaling pathway regulated the asymmetric generation of somites from the tailbud, and that besides Pitx, other factors (X) downstream of Nodal signal also played vital roles in the process (Fig. 4o). We also conducted treatments on later developmental stages, from T0 to T1, and from T1 to L0 stage. Similarly, both treatments could effectively eliminate Pitx expression from the tailbud (Additional file 1: Fig. S8i'-l'). We did not assess if the treated embryos had symmetric somites, since somites at these stages of embryos were not easy to recognize.
Together, the above results indicated that the mutation at the TALEN1 site actually affected Pitxc function, albeit very mildly, and that the development of various LR organs are of different sensibility to Pitx activity reduction. Among them, the mouth, endostyle, CSG, posterior part of the preoral pit, and the asymmetry of the somites and peripheral nerves were sensitive to an intermediate level of Pitx activity reduction (Pitxa/c homozygotes), while the lateral/ventral mesoderm, Hatschek's nephridium, hematopoietic domain, HBD, and the anterior part of the preoral pit were sensitive to total loss of Pitx activity (Pitx −/− ), with none of them sensitive to a low level of Pitx activity reduction (Pitxc −/− ).
Lhx3 and Hex are upstream transcription factors for the development of vertebrate pituitary and thyroid, respectively [49,50], and are among the earliest genes activated in the amphioxus preoral pit and endostyle, respectively [45,47]. We therefore speculated that Pitx regulated the development of the preoral pit and endostyle through directly regulating the expression of Lhx3 and Hex. To test this, we analyzed upstream sequences of the two genes and identified three Pitx binding sites for each gene (Fig. 9a, d, Additional file 1: Table S1). Luciferase assay showed that both sequences were able to drive significantly higher levels of luciferase activity than the control (Fig. 9b, c, e, f). Co-injection of Pitxc mRNA with the constructs increased luciferase activity of Lhx3 construct (Fig. 9b), but decreased luciferase activity of Hex construct (Fig. 9e). Moreover, mutation of the Pitx binding sites abolished activity of Lhx3 sequence (Fig.  9c), but increased activity of Hex sequence (Fig. 9f). These results suggested that Lhx3 and Hex were likely direct transcriptional targets of Pitx in amphioxus, and Pitx binding activated Lhx3 expression, while Pitx binding repressed Hex expression.

Discussion
Requirement of Pitx2 for LR morphogenesis has been demonstrated in mice by gene knockout studies [7][8][9][10] and in chick and frogs using gene knockdown strategies [7][8][9][10]. However, its role in zebrafish LR morphogenesis is still under debate. Using a morpholino knockdown method, Liu and Semina found no LR defects in zebrafish Pitx2 morphants [16], while Garric et al. reported a weak phenotype in habenular asymmetry [17]. In a later study, Ji et al. confirmed the finding of Liu and Semina by examining several lines of Pitx2 mutants and revealed that these mutants are able to survive to adults without any LR defects [15,16]. Outside vertebrates, Pitx  (BS1, BS2, BS3). Numbers show the distance before the start codon. Two luciferase reporter gene constructs are tested in amphioxus embryos with 0 or 3 Pitx binding sites mutated in Lhx3 (a) or Hex (d) promoter. b, e Histogram shows relative levels of luciferase expression for a construct including 0 Pitx binding sites mutated, a group co-injection of Pitxc mRNA with the construct, a control luciferase vector pGL3 without Lhx3 or Hex sequence. c, f Histogram shows relative levels of luciferase expression for each construct (0 or 3 Pitx binding sites mutated), a control luciferase vector pGL3. Significance determined by t-test is indicated by asterisks: *P < 0.05, **P < 0.01, ***P < 0.001 function has not been examined, although its asymmetric expression and regulation by Nodal signaling pathways have been reported in several invertebrate animals [24][25][26][27][28]. We here demonstrate that Pitx is required for LR morphogenesis in the chordate amphioxus. This indicates that involvement of Pitx in LR development predates chordate diversification, and also strengthens the importance of the gene in this process.
Although Pitx function in asymmetric development is generally conserved among chordates, the dependence of the gene in their LR morphogenesis is apparently different. In amphioxus, Pitx mutation affects asymmetries of most if not all LR organs, and causes loss of all left-side organs and ectopic formation of right-side structures on the left side. Whereas Pitx2 deficiency in zebrafish leads to no (or very weak) LR phenotype as mentioned above, and Pitx2 loss-of-function in mice causes right isomerism of the lungs and atrium, and defects in other visceral organs, but the position of the stomach and heart looping remain unaffected [8,[51][52][53][54][55]. These differences reflect that LR organogenesis in amphioxus relies more heavily on Pitx than that in vertebrates, and that LR organogenesis in vertebrates may involve more regulatory factors.
Pitx2c is the only isoform of Pitx2 showing an asymmetric expression pattern in vertebrates [34]. Mice specially lacking asymmetric Pitx2c expression manifest LR defects comparable to Pitx knockout mice [8,11], indicating that Pitx2c is a major if not the only player out of the three Pitx2 isoforms in mouse LR development. Similarly, we identified two Pitx isoforms in amphioxus: Pitxa/b and Pitxc, among which the amphioxus Pitxc showed asymmetric expression in embryos while Pitxa/b did not. However, unlike Pitx2c in mice, Pitxc mutation in amphioxus does not cause any LR defects. One possible interpretation of this discrepancy is that the mutation we introduced is hypomorphic. Indeed, Pitxa/c homozygotes from a cross between Pitxc heterozygotes and Pitx heterozygotes manifested LR defects at the larval stages, and importantly these defects were much milder compared to that observed in Pitx −/− mutants. Furthermore, there are two alternative in-frame start codons immediately after the mutation site we introduced in the Pitxc locus (Additional file 1: Table S2). These findings suggest that the Pitxc −/− mutants may use one of the two alternative start codons to make a dampened version of Pitxc protein. Interestingly, Pitxc expression was slightly upregulated in Pitxc −/− mutants (Additional file 1: Fig. S14d) compared to their WT/Pitx −/− sibling (Additional file 1: Fig. S14c), while Pitxa/b expression remained unchanged (Additional file 1: Fig. S14a, b). We suggest this upregulation can, to some extent, compensate dampened Pitxc function and contribute toward normal development of Pitxc −/− embryos.
Our study provides several lines of evidence indicating that Pitx controls amphioxus LR morphogenesis by promoting development of organs on the left side and repressing formation of organs on the right side. First, Pitx mutation in amphioxus resulted in loss of left-side organs and ectopic formation of right-side organs on the left side. Second, genes essential for left-side organs like Lhx3 and Prop1 were activated after Pitx expression and their expression was observed in the region where Pitx was highly expressed, while genes required for the formation of organs on the right side, at least for the endostyle and CSG, were initiated either earlier (like Hex) or later (like FoxE4) than Pitx, and their expression was observed in the region where Pitx was not expressed. Importantly, the right-side genes were initially symmetrically expressed, and they became asymmetrically expressed (decreased expression on the left side) only when Pitx expression expanded into their expression domain. Third, while Pitx −/− mutants showed no expression of left-side genes and nearly bilaterally symmetric expression of right-side genes, Pitx overexpression induced expression of the left-side genes and inhibited that of the right-side genes. Fourth, Hex and Lhx3 genes, upstream regulators of the endostyle and preoral pit, respectively, had Pitx binding sites in their regulatory sequences, and mutation of these sites increased the activity of the Hex sequence, but decreased that of the Lhx3 sequence. Lastly, although BMP signaling pathway was thought to be a determinant of right-side development in vertebrates [20], our results demonstrated that this was not the case in amphioxus, as inhibition of BMP signaling after the N1 stage did not affect the asymmetrical expression pattern of Pitx, but reduced its expression on the left side (Additional file 1: Fig. S15d-i, d'-i'). However, it should be noted that blocking the BMP signal from late gastrula stage in amphioxus embryos eliminated Pitx expression on both sides (Additional file 1: Fig. S15b, b', c, c'), which is probably caused by ectopic induction of Nodal inhibitor Dand5 on the left side [56]. We are uncertain if vertebrates use a similar mechanism to pattern their LR organs in general, but we speculate that this may be the case in lungs and atria since Pitx2 is expressed in the left side of these organs in mice, and its deficiency results in right isomerism of these organs at the expense of their left-sided structures [8,[51][52][53][54][55]. Consistent with this, a previous study showed that Pitx2c inhibited the development of sinoatrial node (a structure derived from the right atrium) by directly repressing Shox2, a transcriptional regulator of sinoatrial node gene program [57].
We showed that, as in vertebrates [3], LR morphogenesis in amphioxus required Pitx-independent signals (factor X) downstream of Nodal signaling. Bmp2/4 is recently shown to be asymmetrically expressed in amphioxus embryos from N1 to N2 stage in a L > R manner, and inhibition of Nodal signaling at the gastrula stage eliminates Bmp2/4 expression [56]. This suggests a possibility for Bmp2/4 to be the X factor. To test this, we treated Pitx mutants with BMP inhibitor DMH1 from N0 to N5 stage or N0 to T1 stage and assayed the expression of Krox (marker of the glandular region of the CSG), Nkx2.1 (marker of the endostyle), and m-actin (marker of the somites). The treatment abolished the expression of the Krox gene in Pitx −/− mutants and their siblings at the N5 stage (Additional file 1: Fig. S16m, n), indicating that BMP signaling pathway was required for Krox expression. However, the treatment did not affect the expression of Nkx2.1 and m-actin in either Pitx −/− mutants (Additional file 1: Fig. S16p, t) or their siblings (Additional file 1: Fig. S16o, s). As a positive control, the treatment of mutants with the Nodal signaling inhibitor SB505124 at the same stages caused symmetric expression of Krox and Nkx2.1 in all examined embryos (Pitx −/− mutants and their sibling) (Additional file 1: Fig.  S16e-h). NOTCH, FGF, and RA play essential roles in the generation of somites from the tailbud of vertebrate embryos [33,58]. Specifically, inhibition of RA signaling activity in zebrafish, mouse, and chicken led to asymmetric somites [59,60]. Different from vertebrates, RA and FGF signaling are not required for the formation and asymmetric patterning of somites from the tailbud in amphioxus [33]. However, this does not exclude the possibility of the signals as the X factor. To test this, we conducted similar experiments as described above and found that none of them appeared to be the X factor as treatment of Pitx mutants with inhibitors of NOTCH (DAPT), FGF (SU5402), and RA (BMS493) signaling pathways did not affect the expression patterns of mactin (Additional file 1: Fig. S17k-p) and Nkx2.1 (Additional file 1: Fig. S16c-h) genes. Further studies are required to identify the nature of the X signal in the future.
The homologous relationship between the amphioxus preoral pit and vertebrate adenohypophysis was first indicated by anatomical studies [61], and later strengthened by gene expression data [43,62]. We here demonstrate that similar to vertebrates [63], Pitx is able to direct the development of amphioxus preoral pit through controlling expression of Lhx3 and/or Prop1 genes. This further confirms the homologous relationship of the two organs. However, it is striking that the preoral pit is located on the left side of amphioxus while the adenohypophysis is formed at the anterior midline region of vertebrate embryos [43,64]. As Pitx is a master regulator of the two organs, we speculate that this difference is probably related to changes in Pitx expression between amphioxus and vertebrates. Previous studies showed that inhibition of Nodal signaling before 8S stages abolished Pitx expression in amphioxus embryos [24]. We here further showed that Pitx expression relies on the Nodal signaling pathway at least until stage T1. This demonstrates that Nodal signaling is absolutely required for Pitx expression in amphioxus not only in the LR patterning phase, but also in early stage of LR morphogenesis. Vertebrates have evolved three Pitx genes (Pitx1, Pitx2 and Pitx3) by two rounds of whole genome duplications after their divergence from cephalochordates [65]. Among them, Pitx1 and Pitx2 are expressed in the adenohypophysis and are both required for its development [66,67]. However, their expressions in the adenohypophysis are regulated by signaling pathways other than Nodal [68]. Acquirement of Nodalindependent regulation of Pitx genes during vertebrate evolution is therefore likely a key event enabling the shift of the adenohypophysis to the midline in vertebrates.
Somites are unique to amphioxus and vertebrates within the chordates. Although vertebrate somites and the posterior somites of amphioxus are both generated from the tailbud, their formation mode and the patterning of them along the LR axis are different. In vertebrates, somites are formed symmetrically under a clock and wavefront mechanism [33,58], whereas amphioxus lacks the mechanism and develops the posterior somites asymmetrically [69]. RA signal is required for symmetric somitogenesis of vertebrates through buffering the lateralizing influence of the LR machinery [59,60]. Application of RA or BMS009 (an antagonist of the RA receptor) does not affect the asymmetric patterning of amphioxus somites, indicating RA signaling pathway is not involved in the process [33]. However, different from vertebrates, Pitx is continuously expressed in the left side of tailbud from the 8-somite stage during amphioxus development, and this expression relies strictly on the Nodal activity there. Moreover, inhibition of the Nodal activity after 8-somite stages or knockout of Pitx affects the asymmetry of amphioxus somites. Therefore, Nodal signal and Pitx, but not RA signal, are responsible for the asymmetric generation of somites in amphioxus. Considering that vertebrates have evolved a much more sophisticated process in somitogenesis than amphioxus, we speculate that the acquisition of RA signal, together with the loss of Nodal-Pitx cascade, in vertebrates might be the cause for their symmetric somitogenesis.

Conclusions
In summary, our study reveals that Pitx is a dominant (but not the only) factor downstream of Nodal signaling, in the LR morphogenesis of amphioxus (Fig. 10). During the development of amphioxus pharyngeal organs, Pitx promotes the development of left-side organs by directly activating expression of genes like Lhx3 (and/or Prop1 and Pit) (Fig. 10). Simultaneously, it acts together with an unidentified factor X, also downstream of Nodal signaling, to inhibit formation of right-side organs by directly repressing expression of genes like Hex (and/or Nkx2.1 and FoxE4) in a dosage manner (Fig. 10). This provides essential clues for understanding the mechanisms of Pitx2 in the regulation of asymmetric organogenesis in vertebrates, and also sheds important insights on why some organs like adenohypophysis are asymmetrically positioned in the amphioxus, but are symmetrically placed in vertebrates.

Animals and embryo cultivation
Amphioxus Branchiostoma floridae were obtained from Dr. Jr-Kai Yu's laboratory at the Institute of Cellular and Organismic Biology, Academia Sinica, Taiwan. They were maintained and induced to spawn following the protocol as we described and used in B. belcheri [70,71], which is slightly different from the protocol recently developed for B. floridae [37]. Egg fertilization and embryo culture were carried out according to our previous report [72]. Embryos are staged according to a recent study unless otherwise stated [73,74].

Isolation of Pitx isoforms in amphioxus
Pitxc isoform sequence was assembled from in-house generated transcriptome data [75], and Pitxa/b isoform (accession number: FE578121) was acquired by searching publicly available EST data in the NCBI database using Blast with the Pitxc sequence as the query. Three primers, two located in the first exons of Pitxc and Pitxa/b and one located in their common region (2nd exon of Pitxc and 3rd exon of Pitxa/b) were designed to verify their existence in amphioxus embryos. Exon/intron structures of the two isoforms were determined by aligning their mRNA sequences to the corresponding genome sequence on UCSC (genome-asia.ucsc.edu).

Mutant generation and genotyping
Generation and detection of Pitx mutants was conducted using the TALEN method as previously described [24,36,37,75]. TALEN target sites of Pitx and mutation types used in the study are shown in Fig. 1. Primers for amplification of the flanking regions of the Pitx target sites and their amplicon sequences and sizes are listed in Additional file 2: Table  S1. Genotyping of live embryos and embryos examined by whole mount in situ hybridization was performed as previously described [32,37].
Embryos at desired stages were fixed overnight with 4% (wt/vol) PFA-MOPS-EGTA (pH 7.5) at 4°C, and then stored at − 20°C in 70% ethanol (vol/vol) for use. Whole-mount in situ hybridization (WISH) was performed following the description by Yu and Holland [78]. For the three Pitx gene probes, the one (called Pitx probe) transcribed from Pitxc isoform (indicated by the green line in Fig. 1a) was used for most experiments unless otherwise stated. Stained embryos were photographed using an inverted microscope (Olympus, IX71). After imaging, embryos were embedded in agarose- paraffin, sectioned at 5 μm with a Leica RM2016 microtome, and then photographed using a brightfield microscope (Zeiss, AX10).
Acetylated α-tubulin immunofluorescence staining for detection of peripheral nerves was performed as previously described [32]. Before staining, embryos and larvae were de-ciliated by a short incubation in 0.83 M NaCl seawater and then fixed in 4% (wt/vol) PFA-MOPS-EGTA (pH 7.5) at 4°C overnight [32]. DAPI (Invitrogen, 1 μg/ml in PBST) was used for nuclear staining. Images of stained embryos were acquired using a Zeiss LSM 780 confocal microscope.

Quantification and statistical analysis
The distance between somites or peripheral axons was measured using ImageJ software. Scatter boxplots were drawn using OriginPro (OriginLab Corporation). Statistical significance was tested by independent-samples ttests with SPSS Statistics version 19 (IBM Corporation).

Luciferase activity assays
Regulatory sequences of Lhx3 (a 500-bp fragment upstream of its transcriptional start) and Hex (a 554-bp fragment upstream of the transcriptional start) genes were determined according to the ATAC-seq data of B. lanceolatum [79]. They were respectively inserted upstream of the firefly luciferase gene in pGL3 (Promega) and modified by PCR to generate mutant versions (Additional file 2: Table S3). Pitx binding sites in the regulatory sequences of Lhx3 and Hex were predicted using JASPAR profiles at http://jaspar.genereg.net/. Injection solutions were prepared to contain 60 ng/μL of Lhx3 or Hex constructs, or Renilla luciferase vector pRL-SV40 (control), and 3 mg/ml fluorescein-labeled dextran (10,000 MW, Thermo Fisher Scientific), 12.5% glycerol, with or without 100 ng/μL Pitxc mRNA. Embryos were injected at unfertilized egg stage [72] and assayed at N4 stage (eight somites, 60 embryos for each experiment); uninjected embryos from the same batch were used as the negative control. Firefly luciferase expression levels from pGL3 and Renilla luciferase from pRL-SV40 were detected with the Dual Luciferase Kit (Promega) using a GloMax luminometer with an integration of 15 s, and the level of firefly luciferase was normalized to Renilla luciferase activity for each experiment.
Pharmacological treatments SB505124 (Sigma), dorsomorphin homolog-1 (DMH1, Sigma), DAPT (Calbiochem), SU5402 (Calbiochem), and BMS593 (Sigma) were used respectively to inhibit Nodal, BMP, NOTCH, FGF, and RA signaling pathway. They were dissolved in DMSO following standard protocols and used at the following concentrations: 50 μM for SB505124, 40 μM for DMH1, 150 μM for DAPT, 50 μM for SU5402, and 1 μM for BMS593. SB505124 treatment was conducted at N0 (neurula with zero somite), 8S, 10S, T0, and T1 stages; DMH1, DAPT, SU5402, and BMS593 treatments were all conducted at N0 stage. For each treatment, an equal volume of DMSO was applied to the same batch of embryos and used as a negative control. All treatments were performed in six-well plates containing 3 mL of filtered seawater and 200-300 embryos. The treated embryos were fixed at required stages for in situ hybridization or cultured continuously for morphological observation.