ABSTRACT
The origin of the vertebrate head is one of the great unresolved issues in vertebrate evolutionary developmental biology. Although many of the novelties in the vertebrate head and pharynx derive from the neural crest, it is still unknown how early vertebrates patterned the neural crest within the ancestral body plan they inherited from invertebrate chordates. Here, using a basal vertebrate, the sea lamprey, we show that homologs of Semaphorin3F (Sema3F) ligand and its Neuropilin (Nrp) receptors show complementary and dynamic patterns of expression that correlate with key periods of neural crest development (migration and patterning of cranial neural crest-derived structures). Using CRISPR/Cas9-mediated mutagenesis, we demonstrate that lamprey Sema3F is essential for patterning of neural crest-derived melanocytes, cranial ganglia and the head skeleton, but is not required for neural crest migration or patterning of trunk neural crest derivatives. Based on comparisons with jawed vertebrates, our results suggest that the deployment of Nrp-Sema3F signaling, along with other intercellular guidance cues, was pivotal in allowing early vertebrates to organize and pattern cranial neural crest cells into many of the hallmark structures that define the vertebrate head.
INTRODUCTION
A key event in early vertebrate evolution was the transition from a sessile, filter-feeding lifestyle to one of active predation (Gans and Northcutt, 1983). This event was driven, in large part, by a transformation of the ancestral chordate pharynx, resulting in the vertebrate ‘new head’ (Forey and Janvier, 1994; Gans and Northcutt, 1983; Northcutt, 2005; Northcutt and Gans, 1983). The new vertebrate pharynx was muscularized and buttressed by a robust endoskeleton made of cellular cartilage, which in turn provided support and protection for a complex central nervous system and paired sensory organs, thereby facilitating a more active, predatory lifestyle (Donoghue and Keating, 2014; Graham, 2001; Square et al., 2016b). It was the integration and coordination of these traits that distinguished the first vertebrates morphologically and behaviorally from their closest relatives: the invertebrate chordates (Donoghue and Keating, 2014; Gans and Northcutt, 1983).
Much of the head skeleton and sensory organ systems of vertebrates are formed during embryonic development from the neural crest, a migratory and multipotent cell population unique to the vertebrate lineage (Donoghue et al., 2008; Hall, 2008; His, 1868; Le Douarin, 1999; Muñoz and Trainor, 2015; Santagati and Rijli, 2003). At the molecular level, neural crest cell development is controlled by a complex, integrated gene regulatory network (GRN) that progressively refines the developmental state of this cell type from early induction and specification to terminal differentiation (Betancur et al., 2010; Meulemans and Bronner-Fraser, 2004; Sauka-Spengler and Bronner-Fraser, 2008; Simões-Costa and Bronner, 2015). Under the control of this GRN, neural crest cells are specified in the dorsal-most part of the embryonic central nervous system, from which they detach and then migrate throughout the head and trunk (Bronner, 2012; Clay and Halloran, 2010; Duband et al., 1995). After arriving at their destinations, neural crest cells differentiate into an array of cell types, including melanocytes, smooth muscle cells, neurons and glia of the peripheral sensory nervous system, as well as cartilage and bone that comprise the head and pharyngeal skeleton (Green et al., 2015).
Although neural crest cells are a vertebrate innovation, the pharyngeal apparatus, where the neural crest builds much of the head skeleton and sensory systems, is not. In fact, the pharynx is a general feature of deuterostome embryos, as is the gene regulatory network that orchestrates pharyngeal development (Gillis et al., 2012; Ou et al., 2012; Rychel et al., 2005; Veitch et al., 1999). Thus, the formation of the new vertebrate head required not only the origin of new cell types and gene regulatory networks, but also the integration and coordination of ancestral (pharynx development) and derived (neural crest development) developmental-genetic programs (Graham and Richardson, 2012; Veitch et al., 1999). Despite the significance of this event in early vertebrate evolution, the molecular mechanisms that coupled these two developmental processes are unknown.
In jawed (gnathostome) vertebrates, the migratory routes of neural crest cells and patterning of neural crest-derived structures in the head and pharynx are controlled, in part, by signaling interactions between receptors on neural crest cells and their corresponding ligands secreted from other cells into the extracellular environment (Gammill et al., 2007, 2006; Krull et al., 1997; Minoux and Rijli, 2010). Examples of such signaling systems include Robo-Slit, Eph-Ephrin, CXCR4-Sdf, and Neuropilin (Nrp)/Plexin-Semaphorin (Sema), each of which patterns neural crest cells by attraction, repulsion, or a combination thereof (Theveneau and Mayor, 2012, 2014). Although many of these signaling systems are evolutionarily conserved between vertebrates and invertebrates, one particular group, the class III family of Sema ligands (Sema3) and their Nrp receptors (Nrp1 and Nrp2), emerged and were duplicated within the vertebrate lineage (Yazdani and Terman, 2006). Thus, the deployment of Nrp/Sema3 signaling in vertebrates correlates with the appearance of neural crest cells and the new vertebrate head, suggesting a link between Nrp/Sema3-mediated neural crest patterning and the origin of vertebrate novelties, such as the head skeleton and sensory organ systems. This hypothesis is supported by the fact that, in jawed vertebrates, Sema3 (Sema3D and Sema3F) and Nrp (Nrp1 and Nrp2) protein activity is necessary to organize migratory neural crest and pattern neural crest-derived cranial sensory ganglia and elements of the head and pharyngeal skeleton (Berndt and Halloran, 2006; Gammill et al., 2007, 2006; Kulesa and Gammill, 2010; Yu and Moens, 2005). However, it is unknown whether a key Nrp/Sema3 patterning function for neural crest is unique to the jawed vertebrate clade, or is instead a deeply conserved feature of neural crest biology that was also present in the first jawless (agnathan) vertebrates over 500 million years ago.
To distinguish between these possibilities, we examined the expression patterns of the Nrp/Sema3F signaling system, and the functional roles of Sema3F protein during neural crest development in a basal vertebrate, the sea lamprey (Petromyzon marinus). Lampreys are a group of jawless vertebrates that, along with hagfish, constitute the only extant members of the cyclostome (‘agnathan’) clade, which includes diverse fish-like forms that first appeared during the Paleozoic era (McCauley et al., 2015) and are the sister taxa of all other living vertebrates. Given that they occupy the most basal phylogenetic position among extant vertebrates, and readily produce embryos that are amenable to experimental analysis, lampreys are ideal evolutionary and developmental models to study the origin of vertebrate-specific traits (Green and Bronner, 2014; McCauley et al., 2015) through comparisons with developmental mechanisms in jawed vertebrates. Lampreys, similar to their jawed vertebrate relatives, have neural crest cells that migrate into the head and pharynx, following stereotyped routes (McCauley and Bronner-Fraser, 2003; York et al., 2017). Once at their targeted destinations, these neural crest cells are patterned into a pharyngeal and head skeleton made of cellular cartilage, as well as cranial sensory ganglia (Jandzik et al., 2014; Lakiza et al., 2011; McCauley and Bronner-Fraser, 2003, 2006; Square et al., 2016b). Although these structures are presumed to be homologous to cranial cartilage and ganglia in jawed vertebrates (McCauley and Bronner-Fraser, 2006; Modrell et al., 2014), the molecular, cellular and genetic mechanisms responsible for their patterning in jawless vertebrates are unknown.
Our results, together with the prior identification of lamprey Nrp and Sema3 genes (Sauka-Spengler et al., 2007; Shifman and Selzer, 2006, 2007) reveal that the lamprey genome, similar to that of many other vertebrates, encodes homologs of both Nrp1/2 and Sema3. Focusing on Nrp/Sema3F signaling, we show that Sema3F and its Nrp receptors (lamprey NrpA, NrpB and NrpC) show dynamic and complementary patterns of expression that correlate with key steps of neural crest development (migration; early and late patterning of pigment, cranial sensory neurons and head skeleton), similar to jawed vertebrates. Using clustered regularly interspaced short palindromic repeat (CRISPR)/Cas9-mediated genome editing, we demonstrate that Sema3F signaling is not required for the segregation of migratory neural crest streams, but is essential for patterning of pigment, cranial sensory neurons and elements of the head and pharyngeal skeleton at multiple stages of development. Taken together, our results suggest that Nrp receptors and Sema3, as a vertebrate-specific signaling system, act as cellular guidance cues to pattern the cranial neural crest into vertebrate-specific novelties, which allowed stem vertebrates to coordinate neural crest migration and differentiation programs.
RESULTS
Molecular phylogenetics of vertebrate class III Semaphorins and Neuropilins
Previous research identified a Class III (Sema3) and a Class IV (Sema4) Semaphorin, as well as Sema receptors (Nrps) in lamprey (Shifman and Selzer, 2006, 2007). To investigate Nrp/Sema3 signaling activity during neural crest development in lamprey, we PCR amplified and cloned a 553-bp sequence of the gene previously identified as encoding Sema3. Our analysis of the sea lamprey genome uncovered additional Sema3s (Fig. S1A). Phylogenetic analysis confirmed with strong support that the Sema3 sequence first identified (Shifman and Selzer, 2006, 2007) is a member of the Sema3F clade (Fig. S1A) and also confirmed that another previously identified Sema3 in lamprey is likely a member of the Sema3D clade (Fig. S1A) (Sauka-Spengler et al., 2007). Our genomic searches also uncovered three lamprey homologs of vertebrate Nrp1 and Nrp2 receptors. Our phylogenetic analysis suggested that, similar to jawed vertebrates, lampreys have two Nrp paralogy groups (Fig. S1B). However, our analysis was unable to resolve strict paralogy of lamprey Nrps with either jawed vertebrate Nrp1 or Nrp2. Therefore, we named these groups NrpA and NrpB (Fig. S1B), with the two NrpA copies (NrpA1 and NrpA2) likely originating from a lamprey-specific gene duplication event (Fig. S1B).
Lamprey Nrp-Sema3F expression correlates with patterning of cranial neural crest
As a first step toward understanding the contribution of Nrp/Sema3F signaling to neural crest development in basal vertebrates, we characterized their expression patterns throughout lamprey embryogenesis. In jawed vertebrates, early Sema3F expression occurs during neural crest migration, often localizing to neural crest-free zones of the forebrain, hindbrain and pharynx, with migratory neural crest showing complementary expression of Nrp1/2 receptors (Gammill et al., 2007). We found comparable expression patterns of Sema3F/Nrps during lamprey neural crest development (Figs 1-7). In Tahara stage (T) 22 lamprey embryos, neural crest cells are migrating (Sauka-Spengler et al., 2007; Square et al., 2016a; Tahara, 1988; York et al., 2017). At this time, Sema3F transcripts were enriched in forebrain and hindbrain, with expression appearing in the ectoderm and evaginating endodermal pouch near pharyngeal arch (pa) 3 (Fig. 1A,B). By T23, Sema3F expression weakens in the forebrain concomitant with expansion into pa3-4 (Fig. 1C), with expression in the nascent endodermal pouch of pa4 (Fig. 1C,D). At early T24, postmigratory crest cells have colonized the pharynx and, by late T24, gradually become restricted within each of the differentiating pharyngeal arches (McCauley and Bronner-Fraser, 2003). Our expression analysis at these stages showed that expanded Sema3F expression accompanied caudal differentiation of pa3-6, with expression gradually increasing within the pharyngeal ectoderm along the anteroposterior axis (pa3-5 in Fig. 1E,F, and pa3-6 in Fig. 1G,H).
Early cranial expression of lamprey Sema3F. (A,B) At T22, Sema3F transcripts localize to the forebrain (fb), hindbrain (hb) and pa3 in the evaginating endoderm (arrowhead, B) and ectoderm (ect, B). (C,D) This endodermal expression (arrowheads, D) pattern appears again at T23 in pa3-4, as forebrain expression weakens (asterisk, C). (E-H), Throughout T24, as its expression still weakens in the forebrain (compare fb in A with asterisks in C,E,G), Sema3F expression expands throughout the pharyngeal ectoderm (ect, F,H) to more posterior pharyngeal arches (pa3-6 in E,G). The anterior side is facing left in all panels. Dashed lines indicate plane of section shown in the indicated panels. Scale bars: 100 μm.
Early cranial expression of lamprey Sema3F. (A,B) At T22, Sema3F transcripts localize to the forebrain (fb), hindbrain (hb) and pa3 in the evaginating endoderm (arrowhead, B) and ectoderm (ect, B). (C,D) This endodermal expression (arrowheads, D) pattern appears again at T23 in pa3-4, as forebrain expression weakens (asterisk, C). (E-H), Throughout T24, as its expression still weakens in the forebrain (compare fb in A with asterisks in C,E,G), Sema3F expression expands throughout the pharyngeal ectoderm (ect, F,H) to more posterior pharyngeal arches (pa3-6 in E,G). The anterior side is facing left in all panels. Dashed lines indicate plane of section shown in the indicated panels. Scale bars: 100 μm.
During T25/26, the lamprey pharynx elongates and neural crest cells within each pa gradually coalesce into dorsal-ventral stacks of prechondrocytes that prefigure the larval cartilage bars within pa3-9 (Martin et al., 2009; McCauley and Bronner-Fraser, 2003, 2006). Sema3F showed dynamic expression during these early patterning events of the head skeleton (Fig. 2). Continuing from pharyngeal ectodermal expression at late T24 (Fig. 1G,H), by T25, Sema3F mRNA expands into pa1-7 (Fig. 2A), with expression in the pharyngeal ectoderm (Fig. 2B) and epithelium lining the oral cavity (Fig. 2C). The relatively broad ectodermal Sema3F expression accompanying the pharyngeal arches at T25 gave way to sharp expression boundaries within pa1-9 by T26 (Fig. 2D). Sectioning of the T26 embryo in Fig. 2D revealed upregulated Sema3F expression in the rostral endoderm of each pharyngeal pouch, with expression weakening in the ectoderm (Fig. 2E, arrowheads and asterisk). This Sema3F expression pattern occurs just as neural crest cells in pa3-9 are patterned into SoxE1-positive dorsoventrally stacked rods of pharyngeal prechondrocytes (Cattell et al., 2011; McCauley and Bronner-Fraser, 2006). Double in situ hybridization for Sema3F and SoxE1 revealed Sema3F mRNA adjacent to SoxE1-positive prechondrocytes in the pharyngeal arches, suggesting that Sema3F expression patterns the early lamprey head skeleton (Fig. 2F).
Cranial expression of Sema3F during head skeleton patterning. (A-C) At T25, Sema3F expression is segmental in the ectoderm of pa1-7 (A; ect, B) and in the stomadeal epithelium (st in C). (D-F) By T26, Sema3F expression in pa1-9 (D) resolves sharply to the anterior pharyngeal endoderm (arrowheads, E,F), with weakened ectodermal expression (asterisk, E), a pattern adjacent to neural crest-derived pharyngeal prechondrocytes expressing SoxE1 (arrows, F). (G) By T27, Sema3F expression is upregulated ventrally in mucocartilage (mc, arrow). (H-N) At T28, Sema3F expression remains in mucocartilage (mc, H,J), is lost from the pharyngeal cartilage bars (cb, J) and appears in mesenchyme around mucocartilage of pa1-2 and upper (ul) and lower lips (ll), a pattern continuing through T30 (K-N). Except for cross-sections in I,J,M,N, all panels are oriented with the anterior facing left. da, dorsal aorta. Dashed lines in A,D,H,L indicate plane of section shown in the indicated panels. Boxed areas in I and M are shown at higher magnification in J and N, respectively. Scale bars: 100 μm.
Cranial expression of Sema3F during head skeleton patterning. (A-C) At T25, Sema3F expression is segmental in the ectoderm of pa1-7 (A; ect, B) and in the stomadeal epithelium (st in C). (D-F) By T26, Sema3F expression in pa1-9 (D) resolves sharply to the anterior pharyngeal endoderm (arrowheads, E,F), with weakened ectodermal expression (asterisk, E), a pattern adjacent to neural crest-derived pharyngeal prechondrocytes expressing SoxE1 (arrows, F). (G) By T27, Sema3F expression is upregulated ventrally in mucocartilage (mc, arrow). (H-N) At T28, Sema3F expression remains in mucocartilage (mc, H,J), is lost from the pharyngeal cartilage bars (cb, J) and appears in mesenchyme around mucocartilage of pa1-2 and upper (ul) and lower lips (ll), a pattern continuing through T30 (K-N). Except for cross-sections in I,J,M,N, all panels are oriented with the anterior facing left. da, dorsal aorta. Dashed lines in A,D,H,L indicate plane of section shown in the indicated panels. Boxed areas in I and M are shown at higher magnification in J and N, respectively. Scale bars: 100 μm.
Downregulation of Sema3F in pa3-9 by T27 (Fig. 2G) was followed by expression around the ventral mucocartilage, an elastin-like cartilage made of mesenchymal chondrocytes embedded in a loose extracellular matrix (mc, Fig. 2G). Mucocartilage is specific to jawless vertebrates and is the primary cartilage type in the velum (pa1-2), upper and lower lips, and floor of the pharynx (Johnels, 1948; Martin et al., 2009; Yao et al., 2011). From T28 to late T30 Sema3F expression expanded throughout the mucocartilage of the pharyngeal floor, pa1-2 and upper and lower lips (Fig. 2H-N).
After determining Sema3F expression in the lamprey head, we next characterized expression of the corresponding Nrp receptors to identify cell types that might respond to Sema3F signaling. Our expression analysis revealed that lamprey, similar to jawed vertebrates, expresses Nrp orthologs in cranial neural crest throughout head development in a pattern complementary to that of Sema3F expression (Figs 3-7). At T22, NrpA1 and NrpA2 paralogs were expressed in migratory cranial neural crest colonizing the mouth and pa1-3, and in the cranial ectoderm (Fig. 3A-D). Similarly, from T23 to late T24, late-migrating and postmigratory neural crest cells expressing NrpA1 and NrpA2 filled the lateral margin within the mesenchymal core of each of the differentiating pharyngeal arches (Fig. 3E-P). NrpB expression was similar to that of NrpA1 and NrpA2 in migratory cranial neural crest entering the anterior oropharynx and more posterior pharyngeal arches from T22 to early T24 (Fig. 4A-F), eventually occupying the mesenchymal core of each arch by late T24 (Fig. 4G,H).
Early cranial expression of lamprey NrpA1 and NrpA2. At T22 (A-D), NrpA1 and NrpA2 are expressed in pharyngeal ectoderm (ect) and migratory cranial neural crest colonizing pa1-3 (neural crest indicated by arrowheads), a pattern that continues into late T24 (E-P) as NrpA1-positive and NrpA2-positive neural crest cells colonize pa1-6. By late T24 (M-P), both NrpA1 and NrpA2 expression occupies pharyngeal ectoderm (ect) and the mesenchymal core of each arch. All panels are oriented with the anterior facing left. Dashed lines indicate plane of section shown in the indicated panels. Scale bars: 100 μm.
Early cranial expression of lamprey NrpA1 and NrpA2. At T22 (A-D), NrpA1 and NrpA2 are expressed in pharyngeal ectoderm (ect) and migratory cranial neural crest colonizing pa1-3 (neural crest indicated by arrowheads), a pattern that continues into late T24 (E-P) as NrpA1-positive and NrpA2-positive neural crest cells colonize pa1-6. By late T24 (M-P), both NrpA1 and NrpA2 expression occupies pharyngeal ectoderm (ect) and the mesenchymal core of each arch. All panels are oriented with the anterior facing left. Dashed lines indicate plane of section shown in the indicated panels. Scale bars: 100 μm.
Early cranial expression of lamprey NrpB. (A,B) Migratory cranial neural crest expressing NrpB enters the pharynx in pa1-3 (A), colonizing the mesenchymal core of nascent pharyngeal arches (arrowheads, B). NrpB expression also occurs in pharyngeal ectoderm (ect, B). (C-H) This pattern of NrpB-positive neural crest migration continues in pa1-6 from T23 to late T24 (neural crest indicated by arrowheads), but with weakening ectodermal expression (ect). All panels oriented with the anterior facing left. Dashed lines indicate plane of section shown in the indicated panels. Scale bars: 100 μm.
Early cranial expression of lamprey NrpB. (A,B) Migratory cranial neural crest expressing NrpB enters the pharynx in pa1-3 (A), colonizing the mesenchymal core of nascent pharyngeal arches (arrowheads, B). NrpB expression also occurs in pharyngeal ectoderm (ect, B). (C-H) This pattern of NrpB-positive neural crest migration continues in pa1-6 from T23 to late T24 (neural crest indicated by arrowheads), but with weakening ectodermal expression (ect). All panels oriented with the anterior facing left. Dashed lines indicate plane of section shown in the indicated panels. Scale bars: 100 μm.
Continuing from late T24, NrpA1-positive neural crest at T25 still occupied the lateral mesenchymal core of each of pharyngeal arch (Fig. 5A,B) and the oral mesenchyme (Fig. 5C). By T26, some NrpA1-expressing neural crest in the lateral mesenchymal core began to coalesce into the characteristic circular shape of pharyngeal prechondrocytes (Fig. 5D,E; compare Fig. 2F), and was expressed strongly in presumptive mucocartilage occupying pa1 and the lateral velar skeleton (Fig. 5F). From T27 to T30, NrpA1 expression in the head skeleton was weakened, with upregulation in other pharyngeal structures, including gill epithelium (Fig. 5G-N). Compared with NrpA1, the NrpA2 paralog showed comparable patterns of expression in postmigratory neural crest throughout the head and pharynx at T25 and T26 (Fig. 6A-F), but was also expressed in the mesoderm within each arch (Fig. 6B,E). From T27 to T30, NrpA2 mRNA was lost from pharyngeal prechondrocytes (Fig. 6G-J), concomitant with upregulation in the ventral somitic and epibranchial mesoderm (Fig. 6H-J), as well as the neural crest-derived hypobranchial bars (Fig. 6I,J) and larval gills (Fig. 6K-N). Lamprey NrpB expression during later stages of head development was similar to that of NrpA1, with transcripts localizing to postmigratory crest cells in the mouth and lateral margin of the core of each pharyngeal arch at T25 (Fig. 7A-C), with a subset of these cells in each arch contributing to nascent pharyngeal prechondrocytes at T26 (Fig. 7D,E), as well as elements of the velar skeleton, pa1 and pa2 (Fig. 7F). In late embryos and early larvae (T27-late T30), NrpB expression was maintained in the oropharynx and upper lip (Fig. 7G,H,K,N), whereas expression in the neural crest-derived pharyngeal cartilage bars in the posterior pharynx from earlier stages was replaced by mesodermal expression in the ventral somites (Fig. 7G-J) as well as epibranchial mesodermal mesenchyme that contributes to pharyngeal muscle fibers (Fig. 7K-N).
Cranial expression of NrpA1 during head skeleton patterning. (A-C) At T25, NrpA1 expression is in pa1-7 (A) in the lateral mesenchymal cores (arrowhead, B), and in upper lip (ul) mesenchyme around the stomodeum (st, C). (D-F) T26 NrpA1 expression remains in the lateral mesenchymal cores of pa1-9 (D; arrowhead, E), with some cells coalescing into pharyngeal prechondrocyte bars (arrows, E), as well as the skeleton of pa1 (F) and lateral velar skeleton (vs, F). (G-N) At T27, T28 NrpA1 pharyngeal expression (arrowheads) weakens in the cartilage bars (cb, outlined in J), and is upregulated in the gills (g in J), a pattern maintained into late T30 (K-N). Except for cross-sections in I,J,M,N, all panels are oriented with the anterior facing left, and the dorsal side oriented up. Dashed lines indicate plane of section shown in the indicated panels. Boxed areas in I and M are shown at higher magnification in J and N, respectively. nt, neural tube; s, somite. Scale bars: 100 μm.
Cranial expression of NrpA1 during head skeleton patterning. (A-C) At T25, NrpA1 expression is in pa1-7 (A) in the lateral mesenchymal cores (arrowhead, B), and in upper lip (ul) mesenchyme around the stomodeum (st, C). (D-F) T26 NrpA1 expression remains in the lateral mesenchymal cores of pa1-9 (D; arrowhead, E), with some cells coalescing into pharyngeal prechondrocyte bars (arrows, E), as well as the skeleton of pa1 (F) and lateral velar skeleton (vs, F). (G-N) At T27, T28 NrpA1 pharyngeal expression (arrowheads) weakens in the cartilage bars (cb, outlined in J), and is upregulated in the gills (g in J), a pattern maintained into late T30 (K-N). Except for cross-sections in I,J,M,N, all panels are oriented with the anterior facing left, and the dorsal side oriented up. Dashed lines indicate plane of section shown in the indicated panels. Boxed areas in I and M are shown at higher magnification in J and N, respectively. nt, neural tube; s, somite. Scale bars: 100 μm.
Cranial expression of NrpA2 during head skeleton patterning. (A-C) At T25, NrpA2 is expressed in postmigratory neural crest in pa1-7 (A) in the lateral mesenchymal cores (arrowhead, B) and mesoderm (mes in B), and also in upper lip (ul) mesenchyme around the stomodeum (st, C). (D-F) T26 NrpA2 transcripts occur throughout the mesenchymal cores of pa1-9 (D) in the neural crest (arrowhead, E) and mesoderm (mes, E), and in prechondrocytes in pa1-2 (F), but is absent from the velar skeleton (vs, F). (G-J) At T27, T28 NrpA2 expression is in the upper lip (ul), but in the pharynx (arrowheads), expression weakens in the pharyngeal cartilage bars (cb, outlined in J), and is upregulated in the somitic and epibranchial mesoderm, and the hypobranchial bar (ebm, hbb in J). (K-N) Throughout T30, upper lip (ul) expression is maintained, with pharyngeal expression in the gills (g in N). Except for cross-sections in I,J,M,N, all panels are oriented with the anterior facing left and dorsal side oriented up. Dashed lines in A,D,H,L indicate plane of section shown in the indicated panels. Boxed areas in I and M are shown at higher magnification in J and N, respectively. Scale bars: 100 μm.
Cranial expression of NrpA2 during head skeleton patterning. (A-C) At T25, NrpA2 is expressed in postmigratory neural crest in pa1-7 (A) in the lateral mesenchymal cores (arrowhead, B) and mesoderm (mes in B), and also in upper lip (ul) mesenchyme around the stomodeum (st, C). (D-F) T26 NrpA2 transcripts occur throughout the mesenchymal cores of pa1-9 (D) in the neural crest (arrowhead, E) and mesoderm (mes, E), and in prechondrocytes in pa1-2 (F), but is absent from the velar skeleton (vs, F). (G-J) At T27, T28 NrpA2 expression is in the upper lip (ul), but in the pharynx (arrowheads), expression weakens in the pharyngeal cartilage bars (cb, outlined in J), and is upregulated in the somitic and epibranchial mesoderm, and the hypobranchial bar (ebm, hbb in J). (K-N) Throughout T30, upper lip (ul) expression is maintained, with pharyngeal expression in the gills (g in N). Except for cross-sections in I,J,M,N, all panels are oriented with the anterior facing left and dorsal side oriented up. Dashed lines in A,D,H,L indicate plane of section shown in the indicated panels. Boxed areas in I and M are shown at higher magnification in J and N, respectively. Scale bars: 100 μm.
Cranial expression of NrpB during head skeleton patterning. (A-C) NrpB at T25 is expressed in postmigratory neural crest in pa1-7 (A) within the lateral compartment of the mesenchymal cores (arrowhead, B) and in the upper lip (ul) mesenchyme circumscribing the stomodeum (st in C). (D-F) T26 NrpB expression remains in the lateral mesenchymal cores of pa1-9 (D; arrowhead, E), with some cells coalescing into the circular shape of pharyngeal prechondrocyte bars (arrows, E), skeleton of pa1-2 (F), and the velar skeleton (vs, F). (G-J) T27 and T28 NrpB expression remains in the upper lip (ul), is gradually lost in the pharynx (arrowheads), from pharyngeal cartilage bars (cb in J), but is upregulated in ventral somites (s in J), epibranchial mesoderm (ebm in J) and pharyngeal arch muscle (pam in J). (K-N) Throughout T30, NrpB expression is in the upper lip (ul) and ventral somite-derived epibranchial and pharyngeal arch muscle (ebm and pam in M). nt, neural tube. Except for cross-sections in I,J,L and M, all panels are with the anterior facing left and the dorsal side oriented up. Dashed lines indicate plane of section shown in the indicated panels. Boxed areas in I and L are shown at higher magnification in J and M, respectively. Scale bars: 100 μm.
Cranial expression of NrpB during head skeleton patterning. (A-C) NrpB at T25 is expressed in postmigratory neural crest in pa1-7 (A) within the lateral compartment of the mesenchymal cores (arrowhead, B) and in the upper lip (ul) mesenchyme circumscribing the stomodeum (st in C). (D-F) T26 NrpB expression remains in the lateral mesenchymal cores of pa1-9 (D; arrowhead, E), with some cells coalescing into the circular shape of pharyngeal prechondrocyte bars (arrows, E), skeleton of pa1-2 (F), and the velar skeleton (vs, F). (G-J) T27 and T28 NrpB expression remains in the upper lip (ul), is gradually lost in the pharynx (arrowheads), from pharyngeal cartilage bars (cb in J), but is upregulated in ventral somites (s in J), epibranchial mesoderm (ebm in J) and pharyngeal arch muscle (pam in J). (K-N) Throughout T30, NrpB expression is in the upper lip (ul) and ventral somite-derived epibranchial and pharyngeal arch muscle (ebm and pam in M). nt, neural tube. Except for cross-sections in I,J,L and M, all panels are with the anterior facing left and the dorsal side oriented up. Dashed lines indicate plane of section shown in the indicated panels. Boxed areas in I and L are shown at higher magnification in J and M, respectively. Scale bars: 100 μm.
In summary, our expression analysis of lamprey Sema3F and Nrps showed that lamprey Nrp receptors occur on migratory cranial neural crest cells colonizing the early embryonic head (T22-T24) and maintain expression in postmigratory neural crest that will be patterned into elements of the head skeleton (T25-T26). At the same time, complementary expression of Sema3F ligand emanates from the adjacent pharyngeal ectoderm and endoderm, in patterns that are spatially and temporally dynamic. From late embryonic stages (T27-T28) into early larval development (T28-T30), Nrp/Sema3F expression is gradually lost from much of the neural crest-derived pharyngeal cartilage bars and is upregulated in cranial mesoderm and mucocartilage.
Lamprey Sema3F is not required for migration of cranial neural crest
Our results showed that pharyngeal expression of Nrps and Sema3F started during neural crest migration and early colonization of the pharynx (∼T23-T24, Figs 1, 3, 4). This was followed by pharyngeal arch expression that is suggestive of a role in patterning cartilage bars of the head skeleton in pa3-9 (∼T25-T26, Figs 2A-F, 5A-F, 6A-F and 7A-F). Finally, Sema3F and Nrp expression occurs in mucocartilage, (i.e. pa1-2, floor of the pharynx) and the hypobranchial bars of the pharynx (∼T27-30, Figs 2G-N and 6H-J). Given that spatiotemporal differences in Sema3F expression parallel early colonization of the pharynx by Nrp-positive neural crest, formation of cartilage bars in pa3-9, and formation of mucocartilage in pa1-2, we asked whether Sema3F signaling is required for each of these processes. To test the functional role of Sema3F during neural crest development, we used CRISPR/Cas9-mediated genome editing, as described in lamprey (Square et al., 2015; York et al., 2017; Zu et al., 2016).
In jawed vertebrates, migratory neural crest cells express transcription factors, such as Sox10 and n-Myc, among others, and are segregated into three migratory streams (Sauka-Spengler and Bronner-Fraser, 2006, 2008; Wakamatsu et al., 1997). The division of these streams is enforced in part by repellent Sema3F signaling in neural crest-free zones of the head, and functional loss of Sema3F activity results in their inappropriate mixing, which can lead to abnormal patterning of neural crest-derived structures (Gammill et al., 2006, 2007; Kulesa et al., 2010; Kulesa and Gammill, 2010). Similar to jawed vertebrates, lamprey cranial neural crest cells migrate in three streams and express homologs of Sox10 (lamprey SoxE2) and nMyc (Lakiza et al., 2011; McCauley and Bronner-Fraser, 2003; Sauka-Spengler et al., 2007), suggesting that Sema3F also functions in lamprey to segregate migratory cranial neural crest. However, in contrast to jawed vertebrates, we found that, in Sema3F mutant embryos (n=20/20), nMyc+ and SoxE2+ neural crest cells still migrated in three distinct streams (Fig. 8B,D). These migratory patterns were similar to those in negative control embryos (Fig. 8A,C), suggesting the lack of a prominent role for Sema3F in regulating neural crest migration during early development. See Fig. S3 for individual Sema3F mutant genotype sequences.
CRISPR/Cas9 knockout of Sema3F does not impair neural crest migration. (A,C) Control CRISPR embryos (ContCR) showing SoxE2 (A, T22) and nMyc (C, T23) expression in migratory neural crest streams (black arrowheads) separated by thin, crest-free zones (black asterisks). (B,D) Sema3F CRISPR mutants showing migratory neural crest streams (black arrowheads) expressing SoxE2 (B, T22) and nMyc (D, T23) segregated (black asterisks) similar to that of controls. The anterior side is facing left and the dorsal side is oriented up in all panels. Scale bars: 100 μm.
CRISPR/Cas9 knockout of Sema3F does not impair neural crest migration. (A,C) Control CRISPR embryos (ContCR) showing SoxE2 (A, T22) and nMyc (C, T23) expression in migratory neural crest streams (black arrowheads) separated by thin, crest-free zones (black asterisks). (B,D) Sema3F CRISPR mutants showing migratory neural crest streams (black arrowheads) expressing SoxE2 (B, T22) and nMyc (D, T23) segregated (black asterisks) similar to that of controls. The anterior side is facing left and the dorsal side is oriented up in all panels. Scale bars: 100 μm.
Sema3F signaling is essential for early patterning of cranial neural crest derivatives
Despite unperturbed patterning of neural crest migration at stage T22-T23 in Sema3F CRISPR mutants (Fig. 8B,D), we observed inappropriate patterning of cranial neural crest derivatives in older mutant embryos (Figs 9,10; see Fig. S3 for mutant genotypes). At T26, we observed differentiated neural crest-derived melanocytes in mutants, but these embryos (n=8/10) failed to properly position melanocytes in the anterior head and along the dorsal pharynx (Fig. 9A,B). Mutant embryos (n=7/10) also had severely disorganized cranial ganglia, with apparent fusion and/or uncondensed ganglionic neurons compared with controls (Fig. 9C,D). Vertebrate cranial sensory ganglia, including those of lamprey, are thought to derive from both neural crest and ectodermal placode cell populations (Modrell et al., 2014; Schlosser, 2005). Given this, and based on ectodermal expression of Sema3F and Nrp genes in lamprey (Figs 1-7), we investigated the gene expression profiles of the placode-specific cranial ganglion markers Pax3/7 and Six1/2, which mark the ophthalmic portion of the trigeminal ganglion (OpV; Pax3/7) and petrosal and posterior lateral line ganglia (pet, pLGG; Six1/2), respectively (Modrell et al., 2014; Schlosser, 2005; Zou et al., 2004). Our results showed that OpV ganglia in Sema3F mutants appeared smaller, with reduced expression of Pax3/7 (Fig. S2A,B). Moreover, Six1/2-positive pLLG and pet ganglia could not be discriminated from each other in mutants, suggesting that they formed as a single fused ganglion (Fig. S2C,D). These results suggest that proper patterning of nonectomesenchymal cranial neural crest derivatives, as well as cranial sensory placodes, is dependent on Sema3F/Nrp signaling.
CRISPR/Cas9 knockout of Sema3F results in early mispatterning of melanocytes and cranial ganglia. (A) T26 Control (ContCR) embryo showing normal patterning of melanocytes linearly over the pharynx (arrows) and melanocyte migration into the anterior head (arrowhead) and upper lip (ul). (B) Sema3F CRISPR (Sema3FCR) mutants have mispatterned melanocytes at T26, including a lack of melanocyte migration into the upper lip (asterisk, B) and dispersed melanocytes over the pharynx (arrows). (C) T26 control embryo immunostained for Hu in cranial sensory neurons. (D) Sema3F CRISPR mutants showed defects in patterning of cranial sensory neurons, including a lack of condensation of the opV ganglion, apparent fusion of ganglionic clusters (g+pet+epg?) and splitting of interconnected ganglia (asterisk, epibranchial ganglion, epg). The anterior side is facing left and the dorsal side is oriented up in all panels. Scale bars: 100 μm.
CRISPR/Cas9 knockout of Sema3F results in early mispatterning of melanocytes and cranial ganglia. (A) T26 Control (ContCR) embryo showing normal patterning of melanocytes linearly over the pharynx (arrows) and melanocyte migration into the anterior head (arrowhead) and upper lip (ul). (B) Sema3F CRISPR (Sema3FCR) mutants have mispatterned melanocytes at T26, including a lack of melanocyte migration into the upper lip (asterisk, B) and dispersed melanocytes over the pharynx (arrows). (C) T26 control embryo immunostained for Hu in cranial sensory neurons. (D) Sema3F CRISPR mutants showed defects in patterning of cranial sensory neurons, including a lack of condensation of the opV ganglion, apparent fusion of ganglionic clusters (g+pet+epg?) and splitting of interconnected ganglia (asterisk, epibranchial ganglion, epg). The anterior side is facing left and the dorsal side is oriented up in all panels. Scale bars: 100 μm.
CRISPR/Cas9 knockout of Sema3F causes patterning defects during early head skeleton development. (A-F) Control CRISPR embryos (ContCR) at T25 and T26. (A) T26 embryo with SoxE1 expression in prechondrocytes of pa3-9, with weak expression in presumptive mucocartilage of pa1-2 (velum outlined) and upper and lower lips (ul, ll). (B) T26 embryo with SoxE3 perichondrial expression in the presumptive velum in pa1-2 (velum indicated by dashed lines), upper/lower lips (ul, ll), and pa3-9. (C) T25 embryo with TwistA expression in postmigratory neural crest in pa1-2 (dashed line), upper and lower lips (ul, ll), and pa3-9. (D-F) Horizontal sections through the pharynx of embryos in A-C showing pharyngeal pouches (endoderm outlined), and properly patterned prechondrocytes (arrowheads, D,E) or postmigratory crest (arrowheads in F) surrounding the mesoderm (mes in F). (G-L) Sema3F CRISPR mutant embryos (Sema3FCR) at T25 and T26. (G) T26 mutants have disorganized SoxE1-postive neural crest in pa3-9 (arrowhead) and an abnormally shaped oropharynx (asterisk). (H) Sema3F T26 mutants have disorganized SoxE3-positive neural crest in pa1-2 and upper/lower lips (asterisk) and pa3-9 (arrowhead). (I) T25 mutants have TwistA-positive cells scattered in pa1-2 and upper/lower lips (asterisk), and pa3-9 (arrowhead). (J-L) Horizontal sections through the pharynx of mutants in G-I showing disorganized prechondrocytes and postmigratory crest in formed or partly formed pharyngeal arches (arrows, endoderm outlined) and neural crest cells spanning the boundary (arrowheads) between pharyngeal arches in which the endoderm failed to evaginate (asterisks). The anterior side is facing left in all panels. Dashed lines in A-C,G-I indicate plane of section shown in the indicated panels. Scale bars: 100 μm.
CRISPR/Cas9 knockout of Sema3F causes patterning defects during early head skeleton development. (A-F) Control CRISPR embryos (ContCR) at T25 and T26. (A) T26 embryo with SoxE1 expression in prechondrocytes of pa3-9, with weak expression in presumptive mucocartilage of pa1-2 (velum outlined) and upper and lower lips (ul, ll). (B) T26 embryo with SoxE3 perichondrial expression in the presumptive velum in pa1-2 (velum indicated by dashed lines), upper/lower lips (ul, ll), and pa3-9. (C) T25 embryo with TwistA expression in postmigratory neural crest in pa1-2 (dashed line), upper and lower lips (ul, ll), and pa3-9. (D-F) Horizontal sections through the pharynx of embryos in A-C showing pharyngeal pouches (endoderm outlined), and properly patterned prechondrocytes (arrowheads, D,E) or postmigratory crest (arrowheads in F) surrounding the mesoderm (mes in F). (G-L) Sema3F CRISPR mutant embryos (Sema3FCR) at T25 and T26. (G) T26 mutants have disorganized SoxE1-postive neural crest in pa3-9 (arrowhead) and an abnormally shaped oropharynx (asterisk). (H) Sema3F T26 mutants have disorganized SoxE3-positive neural crest in pa1-2 and upper/lower lips (asterisk) and pa3-9 (arrowhead). (I) T25 mutants have TwistA-positive cells scattered in pa1-2 and upper/lower lips (asterisk), and pa3-9 (arrowhead). (J-L) Horizontal sections through the pharynx of mutants in G-I showing disorganized prechondrocytes and postmigratory crest in formed or partly formed pharyngeal arches (arrows, endoderm outlined) and neural crest cells spanning the boundary (arrowheads) between pharyngeal arches in which the endoderm failed to evaginate (asterisks). The anterior side is facing left in all panels. Dashed lines in A-C,G-I indicate plane of section shown in the indicated panels. Scale bars: 100 μm.
Next, we focused on the possible patterning functions of Sema3F/Nrp signaling in neural crest cells during early head skeleton development in lamprey. The SoxE and Twist families of transcription factors are not only widely recognized for controlling specification of neural crest cells, but are also known to govern patterning of neural crest-derived elements of the head skeleton in vertebrate embryos (Carl et al., 1999; Cattell et al., 2011; Cheung and Briscoe, 2003; McCauley, 2008; Soo et al., 2002). Similar to jawed vertebrates, lamprey homologs of these genes (SoxE1, SoxE3 and TwistA) are also expressed during early development of the head skeleton and are required for development of cellular cartilage (T25-T26) (Lakiza et al., 2011; McCauley and Bronner-Fraser, 2006; Sauka-Spengler et al., 2007). In particular, SoxE1 expression largely occurs in pharyngeal prechondrocytes in pa3-9 (Cattell et al., 2011; McCauley and Bronner-Fraser, 2006), whereas SoxE3 and TwistA transcripts mark prechondrocytes in all pas, including mucocartilage elements in pa1-2 and upper and lower lips (McCauley and Bronner-Fraser, 2006; Sauka-Spengler et al., 2007; Zhang et al., 2006) (see also Fig. 10A-F). As expected from our results above (Fig. 8), we found that CRISPR targeting of Sema3F did not prevent SoxE1+, SoxE3+ or TwistA+ cells from migrating into the pharynx (Fig. 10). However, the cells expressing these genes did not become organized into serially repeating stacks of prechondrocytes in pa3-9 compared with control embryos (SoxE1, n=18/20, Fig. 10A,G; SoxE3, n=16/20, Fig. 10B,H; TwistA, n=15/20; Fig. 10C,I). Sectioning of these Sema3F CRISPR mutants revealed that, in some cases, the pharyngeal endoderm had failed to evaginate properly and did not contact the ectoderm laterally (asterisks in Fig. 10J-L, compare to Fig. 10D-F), leading to SoxE1, SoxE3 and TwistA-expressing cells being able to cross pharyngeal arch boundaries along the anteroposterior axis (arrowheads, Fig. 10J-L), a result similar to that of previous work in lamprey suggesting that correct chondrogenesis and patterning of cartilage precursors is dependent on proper pharyngeal pouch formation (Jandzik et al., 2014). However, even in pharyngeal arches that had outpocketed completely or nearly so, we still observed postmigratory neural crest cells that had failed to completely condense into pharyngeal prechondrocytes with sharp boundaries (arrows in Fig. 10J,K,L), suggesting that proper patterning of pharyngeal prechondrocytes in Sema3F mutants involves not only an indirect effect on pharyngeal pouch morphogenesis, but also a specific neural crest patterning function upon proper pouch formation. We also observed that Sema3F mutant embryos had patterning defects in pa1-2, which will form the mucocartilage-based elements of the mouth and velar skeleton. We observed failure of proper mouth development in embryos, including improperly patterned SoxE3+ and TwistA+ cartilage elements in pa1-2 (compare pa1-2 outline in Fig. 10A-C with asterisks in Fig. 10G-I). Taken together, these results highlight an important role for Sema3F signaling in patterning the neural crest-derived head skeleton in lamprey embryos.
Sema3F signaling is essential for long-term patterning of pigment, cranial sensory ganglia and cartilage elements of the head skeleton
Finally, given that we observed early defects in the patterning of cranial neural crest-derived melanocytes, cranial sensory ganglia and prechondrocytes during earlier embryonic stages in Sema3F CRISPR mutants (T25-T26, Figs 9,10), we asked whether these phenotypes persisted into ammocoete larval stages (T30), or were corrected later in development by a compensatory mechanism. We found that, compared with control larvae, CRISPR/Cas9-mediated knockout of Sema3F did not prevent differentiation of melanocytes, yet most mutant embryos (n=9/10) had melanocytes that appeared to be scattered randomly throughout the head and failed to become properly patterned into a segmental organization with each of the pharyngeal arches (compare Fig. 11A and Fig. 11F; see Fig. S3 for mutant genotypes). Moreover, in two embryos, in addition to the apparent random location of melanocytes, these cells were small with a stellate appearance and lacked the dendritic appearance of melanocytes in control embryos (compare Fig. 11B and Fig. 11G).
CRISPR/Cas9 knockout of Sema3F results in disorganized melanocytes and cranial ganglia. (A-E) Control CRISPR larvae (ContCR) at T30. (A) Control larvae have melanocytes arranged segmentally in correlation with pa1-9. (B) High magnification of melanocytes in control embryo with typical size and stellate appearance. (C) Fluorescent Hu immunostaining showing properly patterned cranial ganglia in control larva. (D) Higher magnification image of inset in C showing individual cranial ganglia. (E) Fluorescent Hu immunostaining in the trunk of control larva showing segmental organization of DRG (arrowheads) and enteric neurons (en, arrow). (F-J) Sema3F mutant larvae (Sema3FCR) at T30. (F) Mutants had disorganized melanocytes (arrowheads) and (G) melanocytes were smaller and lacked the stellate branches (arrowheads). (H-J) Sema3F mutants had mispatterned cranial ganglia (inset, I), but DRG (arrowheads in J) and enteric neurons (en, arrow in J) appeared unaffected. Question mark indicates unidentifiable ganglionic protrusion from part of the mmV ganglion. e, eye; epg, epibranchial ganglion; g, geniculate ganglion; mmV+g+va, fusion of geniculate and vestibuloacoustic ganglia; pllg, posterior lateral line ganglion; va, vestibuloacoustic ganglion. The anterior side is facing left and the dorsal side is oriented up in all panels. Scale bars: 100 μm.
CRISPR/Cas9 knockout of Sema3F results in disorganized melanocytes and cranial ganglia. (A-E) Control CRISPR larvae (ContCR) at T30. (A) Control larvae have melanocytes arranged segmentally in correlation with pa1-9. (B) High magnification of melanocytes in control embryo with typical size and stellate appearance. (C) Fluorescent Hu immunostaining showing properly patterned cranial ganglia in control larva. (D) Higher magnification image of inset in C showing individual cranial ganglia. (E) Fluorescent Hu immunostaining in the trunk of control larva showing segmental organization of DRG (arrowheads) and enteric neurons (en, arrow). (F-J) Sema3F mutant larvae (Sema3FCR) at T30. (F) Mutants had disorganized melanocytes (arrowheads) and (G) melanocytes were smaller and lacked the stellate branches (arrowheads). (H-J) Sema3F mutants had mispatterned cranial ganglia (inset, I), but DRG (arrowheads in J) and enteric neurons (en, arrow in J) appeared unaffected. Question mark indicates unidentifiable ganglionic protrusion from part of the mmV ganglion. e, eye; epg, epibranchial ganglion; g, geniculate ganglion; mmV+g+va, fusion of geniculate and vestibuloacoustic ganglia; pllg, posterior lateral line ganglion; va, vestibuloacoustic ganglion. The anterior side is facing left and the dorsal side is oriented up in all panels. Scale bars: 100 μm.
Sema3F mutant larvae also had mature cranial sensory ganglia, yet the ganglia in most larvae (n=11/12) were malformed compared with controls (compare Fig. 11C,D with Fig. 11H,I) as we also observed earlier at stage T26 (Fig. 9). These patterning defects included misshapen ganglia (e.g. pLLG; Fig. 11I), and unidentifiable ganglionic protrusions (? in Fig. 11I), to apparent fusion of multiple ganglionic clusters [e.g. maxillomandibular (mmV)+geniculate (g)+vestibuloacoustic (va), Fig. 11I]. In some cases, we noted individual ganglia appeared reduced in size [e.g. petrosal ganglion (pet), Fig. 11I], but these effects varied among mutant embryos. However, in contrast to defects in cranial sensory ganglia, these embryos (n=12/12) had normally patterned trunk neural crest-derived dorsal root ganglia and enteric neurons (compare Fig. 11E and Fig. 11J; see Fig. S3 for mutant genotypes).
Among ∼30 Sema3F mutants that survived to stage 30, larvae developed cellular cartilage of the head and pharyngeal skeleton, including the ‘stack-of-coin’ cartilage bars in pa3-9, as well as mucocartilage elements in pa1-2 and the upper and lower lips (Fig. 12). However, there were consistent and moderate to severe patterning defects in each of these skeletal elements. In all larvae examined (n=15/15 analyzed), the mucocartilage of pa1-2 failed to properly condense into a velum, the oral skeletal element that functions in agnathan respiration (compare Fig. 12A control with Sema3F mutants in Fig. 12D,G). Rather, these larvae had loosely arranged mucocartilage and Alcian Blue-positive cellular debris scattered throughout the head and mouth (arrowheads in Fig. 12D,G; see Fig. S3 for mutant genotypes). There were also abnormalities in the cartilage bars in pa3-9, with the most frequently observed phenotype comprising severely bent or disjointed cartilage bars (n=13/15) (compare Fig. 12B,E,H, and Fig. 12C,F,I). These larvae also had ectopic clusters of fused cartilage nodules from adjacent bars (arrowheads in Fig. 12F,I), or disconnected bars that were not fused together (asterisks in Fig. 12F,I).
CRISPR/Cas9 knockout of Sema3F results in a disorganized head skeleton. (A) Control CRISPR larva (ContCR) with Alcian Blue staining of the head skeleton. Mucocartilage elements include upper/lower lips (ul, ll), medial velar skeleton (mvs) and pa1-2 (outlined). Cartilage bars of pa3-9 are indicated. (B) Same larva in A but with fluorescence in pa3-9. (C) Higher magnification image of inset in B showing cartilage bar morphology (top arrowhead) and ventral fusion of the branchial basket (bottom arrowhead). (D-F) Sema3F CRISPR mutant larva (Sema3FCR). (D) Alcian Blue staining reveals disorganized mucocartilage of pa1-2 (arrowheads). (E) Fluorescence (E, inset in F) of the same larva shows disarticulation of cartilage bars in pa4-6 (asterisks, F) whereas the middle of these cartilage bars are clustered and disjointed (arrowhead, F). (G-I) Another Sema3FCR larva. (G) Mucocartilage cells are scattered throughout the oral skeleton that never condensed into the velum in pa1-2 (arrowheads). (H) Fluorescent imaging (H, enlarged in I) of the same larva shows disarticulated (asterisk in I, arch 6) and fused and bent cartilage bars in pa3-9 (arrowhead, I, arch 8). The anterior side is facing left and the dorsal side is oriented up in all panels. Scale bars: 100 μm.
CRISPR/Cas9 knockout of Sema3F results in a disorganized head skeleton. (A) Control CRISPR larva (ContCR) with Alcian Blue staining of the head skeleton. Mucocartilage elements include upper/lower lips (ul, ll), medial velar skeleton (mvs) and pa1-2 (outlined). Cartilage bars of pa3-9 are indicated. (B) Same larva in A but with fluorescence in pa3-9. (C) Higher magnification image of inset in B showing cartilage bar morphology (top arrowhead) and ventral fusion of the branchial basket (bottom arrowhead). (D-F) Sema3F CRISPR mutant larva (Sema3FCR). (D) Alcian Blue staining reveals disorganized mucocartilage of pa1-2 (arrowheads). (E) Fluorescence (E, inset in F) of the same larva shows disarticulation of cartilage bars in pa4-6 (asterisks, F) whereas the middle of these cartilage bars are clustered and disjointed (arrowhead, F). (G-I) Another Sema3FCR larva. (G) Mucocartilage cells are scattered throughout the oral skeleton that never condensed into the velum in pa1-2 (arrowheads). (H) Fluorescent imaging (H, enlarged in I) of the same larva shows disarticulated (asterisk in I, arch 6) and fused and bent cartilage bars in pa3-9 (arrowhead, I, arch 8). The anterior side is facing left and the dorsal side is oriented up in all panels. Scale bars: 100 μm.
Although our gene expression analyses suggested that early patterning of the head skeleton (T25-T26) was severely disrupted in Sema3F CRISPR mutants (Fig. 10G-L), the larvae we examined at T30 still displayed some evidence of identifiable dorsal-ventral and anteroposterior patterning of cartilage bars in pa3-9 that allowed us to putatively assign some of their identities (Fig. 12E,F,H,I). We attribute this difference in the severity of early- versus late-stage phenotypes to lethality effects we observed shortly after T26: ∼98% of 5000 injected embryos survived to T26 (15 days post fertilization), yet only ∼30 larvae (0.6%) had survived to T30 (30 days post fertilization). We speculate that this sharp increase in mortality might be attributed to the fact that embryos with the most severely disrupted pharyngeal development during early head skeleton patterning were unable to properly pattern their oropharyngeal skeleton, which is required for ventilation and survival at later larval stages. Thus, larvae that survived to T30 were those that showed only moderate disruption of Sema3F function and pharyngeal development, which allowed for their subsequent examination.
DISCUSSION
The new head hypothesis proposes that the origin of vertebrates was catalyzed by a series of evolutionary modifications to the chordate head region (Gans and Northcutt, 1983). These modifications included a muscularized, pumping pharynx that was supported by a rigid cellular skeleton, as well as a peripheral nervous system containing elaborate, paired sensory organs (Donoghue and Keating, 2014; Gans and Northcutt, 1983). Each of these innovations was made possible by the acquisition of the neural crest and, therefore, one of the primary aims in vertebrate evolutionary-developmental biology has been to dissect the molecular, cellular and genetic origins of neural crest cells. However, it has remained unclear how the neural crest and its underlying gene regulatory network became integrated developmentally into the ancestral chordate body plan and acquired the ability to construct the novelties that define the vertebrate head and pharynx.
Our gene expression and CRISPR/Cas9 functional analyses together suggested that the ability of neural crest cells to become integrated and assembled into distinct structures organized along the anteroposterior axis (e.g. pharyngeal skeleton and cranial ganglia) within the head of lamprey embryos is driven in part by the deployment of Nrp-Sema3F signaling, an intercellular signaling system that originated before the divergence of jawless and jawed vertebrates over 500 million years ago. In jawed vertebrates, Sema3 proteins, especially Sema3F, function to repel migratory and postmigratory neural crest cells that express complementary Nrp1 and Nrp2 receptors, thereby positioning groups of neural crest to differentiate into specific derivatives along the embryonic anteroposterior axis (Gammill et al., 2007). Our results in lamprey, a basal jawless vertebrate, suggested that these complementary patterns of Nrp and Sema3F expression are strikingly similar to those in jawed vertebrates and correlate with key stages of cranial neural crest patterning events. Moreover, our CRISPR/Cas9 functional experiments revealed a crucial role for Nrp/Sema3F-mediated patterning of multiple neural crest-derived structures in the vertebrate head, including the craniofacial skeleton and sensory ganglia, and suggest that Nrp-Sema3F signaling is a deeply conserved function that ancestral vertebrates used to pattern cranial neural crest cells. In light of our findings in lamprey and comparisons with jawed vertebrates, we propose that deployment of intercellular guidance cues, such as Sema3, along with their corresponding Nrp receptors, was instrumental in organizing neural crest cells for the first time into derivatives in the vertebrate head by assuming a role in patterning events inserted temporally between earlier (neural crest specification) and later (neural crest differentiation) steps of neural crest development.
Once the core structural components of the neural crest (head skeleton and cranial sensory ganglia) were fixed in early vertebrates, changes in the spatial and temporal patterns of intercellular signaling and patterning systems via Nrp/Sema3F and others could have enabled further modifications to these and other neural crest-derived structures in the vertebrate head. For example, the anterior-most pharyngeal arches in early jawless vertebrates and modern lampreys are differentiated into a velum, a cartilaginous skeletal element in the agnathan oropharynx that functions in respiration (Forey, 1995; Mallatt, 1984, 1997; Miyashita, 2016; Square et al., 2016b; Yasui and Kaji, 2008). Our CRISPR/Cas9-mediated knockout results showed that condensation and patterning of the lamprey velum required proper Nrp/Sema3F signaling, suggesting that modifications to the early vertebrate oropharyngeal skeleton were mediated, at least in part, by Nrp/Sema3F signaling. In higher jawed vertebrates, the ancestrally homonomous series of neural crest-derived pharyngeal cartilage bars was gradually transformed into a series of individuated structures, including jaws and hyoid as well as elements of the inner ear and facial skeleton (Gegenbaur, 1878; Kardong, 2002; Kuratani, 2004; Mallatt, 2008; McCauley and Bronner-Fraser, 2006; Romer, 1950; Shigetani et al., 2005). These changes were driven primarily by spatial repositioning of the ancestral pharyngeal structure and new cell-mesenchyme interactions, rather than by the origin of new cell types (Dupret et al., 2014; Gegenbaur, 1878; Gillis et al., 2013; Kardong, 2002; Shigetani et al., 2002). Similarly, our functional results in lamprey suggest that rearrangement and novel patterning of the pharyngeal and head skeleton, along with other novelties throughout vertebrate evolution (Noguchi et al., 2017), was achieved, in part, by altering the spatial and temporal activity of a combinatorial repulsion-guidance code of signaling molecules involving Sema3F, among others, for neural crest cells in the head of vertebrate embryos.
Although our findings in lamprey implicate an ancient role for Nrp/Sema3F signaling in patterning neural crest-derived structures in the vertebrate head (cartilage, sensory neurons and pigment), we also observed important differences compared with jawed vertebrates. In mouse and chicken embryos, for example, Sema3F functions early in neural crest development to enforce the segregation of the three primary cranial neural crest streams, which express Nrp1/2 receptors (Gammill et al., 2007; Kulesa et al., 2010). Functional perturbation of Sema3F activity results in intermingling or complete fusion of these streams, leading to inappropriate patterning of cranial ganglia and the pharyngeal skeleton (Gammill et al., 2006, 2007; Roffers-Agarwal and Gammill, 2009). By contrast, lamprey does not appear to use Nrp/Sema3F signaling to pattern or segregate migratory crest, and it instead functions primarily in the structural organization of neural crest derivatives, especially craniofacial cartilage and sensory ganglia. This suggests an alternative signaling mechanism that mediates segregation of cranial neural crest streams in jawless vertebrates, although the exact guidance cues are unknown. Alternatively, the lack of a Sema3F mutant phenotype for neural crest migration might reflect the relaxation of migratory constraints in the lamprey head, as previously described (McCauley and Bronner-Fraser, 2003). In addition to early patterning of migratory crest, Nrp/Sema3F signaling in jawed vertebrates is also crucial for patterning trunk neural crest derivatives, such as dorsal root ganglia (DRG) (Gammill et al., 2006, 2007). Similar to jawed vertebrates, lampreys also have DRG arranged along the trunk in a segmental pattern, and recent work shows that lamprey has a population of trunk neural crest-derived enteric neurons (Green et al., 2017). Both DRG and enteric neurons form in stereotypical positions in the lamprey trunk and, therefore, are patterned using intercellular signaling cues (Green et al., 2017). However, Nrp/Sema3F signaling appears to be dispensable for patterning of the trunk neural crest subpopulation in lamprey. This suggests the operation of patterning cues for neural crest in the lamprey trunk that are distinct from those in the head, a situation that also occurs in jawed vertebrates (Krull et al., 1997; Kulesa et al., 2010; Robinson et al., 1997). A comprehensive comparative analysis of a wider repertoire of neural crest-patterning mechanisms in lamprey and hagfish, another agnathan group, could help to address whether ancestral vertebrates patterned cranial versus trunk neural crest subpopulations using distinct or overlapping intercellular signaling mechanisms.
Although bona fide migratory neural crest cells and the structures that they form are vertebrate innovations, there is compelling evidence that the closest extant relatives of vertebrates, the invertebrate chordates, have ‘proto-neural crest cells’ that have a similar gene regulatory profile and can give rise to similar cell types, such as sensory neurons and melanocytes (Abitua et al., 2012; Stolfi et al., 2015). Although some of these cells can migrate endogenously over a short distance, long-range migration is only possible when neural crest transcription factors, such as Twist, are forcibly expressed (Abitua et al., 2012). These migratory cells colonize the pharynx as ectomesenchyme (Abitua et al., 2012), but there has been no gene expression or functional analysis of the contribution of receptor-ligand guidance cues in these cells. We hypothesize that the molecular deployment of Nrp/Sema3 signaling in the head of early vertebrates, in conjunction with co-option of other guidance and repulsion cues into a combinatorial receptor-ligand patterning ‘code’, was an important step that allowed stem vertebrates to organize neural crest cells for the first time into many of the hallmark traits that define the new vertebrate head. Therefore, it would be interesting to determine whether and how migratory neural crest-like cells in invertebrate chordates deploy intercellular patterning systems.
MATERIALS AND METHODS
Embryo collection
To collect embryos, gravid adult sea lampreys (P. marinus) were obtained from the Hammond Bay Biological Station, Millersburg, MI, and shipped to the University of Oklahoma. Adults were housed at 14°C in a recirculating water system. Eggs were stripped manually from gravid females into a beaker of water (∼200 ml) and mixed with sperm expressed from a male directly onto the eggs. Embryos were reared in small Pyrex dishes under constant flow in deionized water supplemented with 0.05X Marc's Modified Ringers solution (MMR) chilled to 18°C. All procedures involving adult lampreys were performed with approval from the University of Oklahoma Institutional Animal Care and Use Committee (IACUC, R15-027).
Molecular phylogenetics
To determine sequence orthology of lamprey Sema3s and Nrp receptors, we constructed neighbor-joining phylogenetic trees, using gnathostome Sema7 and Nrp and Tolloid-Like (NETO) genes as outgroups, respectively. Untrimmed sequences were aligned in MEGA version 7.0 using MUSCLE, and a JTT+G model for protein evolution was chosen for phylogeny reconstruction (Kumar et al., 2016). Results were obtained after 1000 parametric bootstrap replicates. Gene sequences analyzed and corresponding accession numbers (in parentheses) included: Dr, Danio rerio (Sema3B: NP_001121818.1; Sema3C: XP_017210807.2; Sema3D: AAI62510.1; Sema3Fa: AAI63764.1; Sema3Fb: AAW56082.1; Nrp1a: AAI63888.1; Nrp2: NP_998130.1); Gg, Gallus gallus (Sema3A: NP_990308.2; Sema3C: NP_989574.1; Sema3D: NP_990704.1; Sema3E: NP_989573.1; Sema3F: NP_989589.1; Sema3G: XP_015148335; Sema7A: NP_001186678.1; Nrp1: NP_990113.1; Nrp2: NP_989615.1); Mm, Mus musculus (Sema3A: AAH90844.1; Sema3B: AAH90669.1; Sema3C: NP_038685.3; Sema3D: NP_083158.3; Sema3E: NP_035478.2; Sema3F: AAH10976.1; Sema3G: NP_001020550.1; Nrp1: AAH51447.1; NETO1: EDL09346.1); Pm, P. marinus (Sema3F: AAU94360.1; all other putative lamprey Sema and Nrp sequences were obtained from manual searches of the 2010 version of the sea lamprey genome assembly); Rn, Rattus norvegicus (Nrp2: NP_110496.1); Xt, Xenopus tropicalis (Sema3A: AAK38166.1; Nrp2: AAI36102.1); and Xl, Xenopus laevis (Sema3B: AAI66183.1; Sema3D: NP_001087589.1; Sema3F: NP_001011157.1; Nrp1: NP_001081380.1; NETO2: NP_001072912.1).
Gene cloning, in situ hybridization, immunostaining and Alcian Blue staining
Partial clones for Sema3F (553 bp), NrpA1 (670 bp), NrpA2 (550 bp), NrpB (584 bp) and Six1/2 (706 bp) were isolated by direct amplification from a sea lamprey cDNA library (primers: Sema3F forward: 5′-CCACGGAATCTGGCAACCAGAA-3′; Sema3F reverse: 5′-GCGATGCGCGTGAACTTGTA-3′; NrpA1 forward: 5′-CTGAGATTGTCCTGCGATTCCAC-3′, NrpA1 reverse: 5′-CGCACGAACCGCGTCAGCAC-3′; NrpA2 forward: 5′-ATGCTCGCACATGTTCACAGC-3′, NrpA2 reverse: 5′-CGGATCATCTCTGCTGGGCG-3′; NrpB forward: 5′-GGATCCTCTCGCTCTCCTTC-3′, NrpB reverse: 5′-GGAGATGTGACAGCCGTAGA-3′; and Six1/2 forward: 5′-TCCACAAGAACGAGAGCGTG-3′, Six1/2 reverse: 5′-TGCTGAGACATGTGGCTCTG-3′) (kindly provided by J. Langeland Kalamazoo College, MI, USA), ligated into a pGEM-T-easy vector and sequenced. All other clones (SoxE1, SoxE2, SoxE3, nMyc, TwistA and Pax3/7) were isolated from previous cDNA library screenings (McCauley and Bronner-Fraser, 2006; Sauka-Spengler et al., 2007). Clones were then used to generate antisense riboprobes for single or double chromogenic in situ hybridization as previously described (York et al., 2017). To visualize differentiated cellular cartilage, Alcian Blue staining was performed as previously described (Martin et al., 2009). For Hu immunostaining, the primary antibody (Hu C/D, mouse IgG2b; Invitrogen) was diluted (1:300) in 10% sheep serum, and detected using either goat anti-mouse IgG conjugated to horseradish peroxidase followed by DAB staining or Alexa 544-conjugated goat anti-rabbit IgG (1:300).
CRISPR/Cas9 experiments
For all CRISPR experiments, lamprey zygotes were microinjected before first cleavage (∼5 nl) with 1 ng-µl−1 Cas9 protein (PNA Bio), 500 pg guide RNA (gRNA) and 10% fluorescein dextran tracer in nuclease-free water. Approximately 5000 injected embryos were screened by fluorescence 4 days after injection and those lacking fluorescence were discarded. Injected embryos were reared to desired stages, fixed in MEMFA, dehydrated and stored at −20°C in 100% methanol.
Sema3F CRISPR experiments
To disrupt lamprey Sema3F function, we microinjected a gRNA that efficiently and specifically (Figs S3-6) targets the Sema3F genomic coding sequence [Sema3FgRNA1: 5′-GGAGCACCTTCCTGAAGGCCCGG-3′; protospacer adjacent motif (PAM) sequence is underlined]. This gRNA construct was carefully selected to recognize only a single region of Sema3F and to avoid off-target cleavage effects based on the following stringency criteria described previously for lamprey (Square et al., 2015; York et al., 2017): 50-80% GC content; targeted regions as close as possible to the presumptive start codon (or 5′ end of available genomic sequence); and no potential nonspecific/off-target hits to the known P. marinus genome that had >80% similarity by BLAST analysis.
Control CRISPR experiments
To rule out the possibility that Sema3F mutant phenotypes result from a general artifact of Sema3F gRNA1 construct injection and/or toxicity, we performed negative control experiments in which we microinjected a gRNA with a ‘scrambled’ sequence of nucleotides (5′-AATAAGTTGGGGTTTCCA-3′) into zygotes from the same batch of eggs for which we performed our Sema3F CRISPR injections.
Genotyping of Sema3F individual CRISPR mutants
After immunostaining or in situ hybridization of putative Sema3F CRISPR mutants, we genotyped individual embryos from Figs 8-12 and Fig. S2 (see Results) to directly link mutant genotypes to phenotypes. To ensure that tissue fixation or damage to genomic DNA during the in situ hybridization or immunostaining protocols did not generate false positive mutations in our sequencing reactions, we compared the sequences of putative mutant embryos to negative control CRISPR embryos that were also fixed and then assayed by in situ hybridization or immunostaining. To this end, following gene expression analyses and imaging, embryos were incubated for 24-48 h with 0.1 mg ml−1 proteinase K before extraction of genomic DNA (Sive et al., 2000). Oligonucleotides (Sigma) flanking the Sema3F (forward: 5′-TCAATGTCACGAGTTGCAAG-3′; reverse: 5′-TTAATCGAATCGCTAGCTAG-3′) genomic CRISPR target site were used to PCR amplify and sequence 742 bp of the Sema3F genomic locus. For each individual embryo in which we performed this protocol, we sequenced four different clones to verify mutagenesis.
Efficiency of mutagenesis at the Sema3F locus
We estimated the general efficiency of CRISPR-Cas9-mediated mutagenesis of Sema3F gRNA1 at the Sema3F locus by pooling five randomly selected Sema3F gRNA1 CRISPR-injected embryos at ∼T26 (∼embryonic day 15), isolating genomic DNA per standard methods, PCR amplifying the targeted genomic locus (oligonucleotide sequences listed above), and then sequencing 50 clones. Efficiency (%) of mutagenesis at the Sema3F target locus was then calculated by dividing the number of mutant genotypes by the total number of clones sequenced. We found that our Sema3F gRNA1 construct was, in general, highly efficient at inducing mutations at the targeted Sema3F locus in randomly selected embryos, with an estimated mutagenesis efficiency of 98% (Fig. S4). Graphical representation as a box and whisker plot (Fig. S5) of mutant genotypes obtained from individual and pooled embryos (Figs S3, S4) was prepared in R (R Core Team, 2013) using the package ‘ggplot2’ (Wickham, 2010).
Genomic analysis of off-target CRISPR sites
Although our Sema3F gRNA1 construct was designed explicitly to minimize the potential for off-target cleavage (see ‘Sema3F CRISPR experiments’ above), we nonetheless sought to verify that our Sema3F mutant phenotypes were specific to cleavage at the Sema3F locus and, therefore, were not likely to be attributable to mutagenesis of other loci (‘off-target’ effects) by Sema3F gRNA1. To this end, we performed an in silico analysis in which we conducted BLAST searches of the Sema3F gRNA1 sequence against the 2010 version of the sea lamprey genome (https://genome.ucsc.edu/cgi-bin/hgGateway) (Servetnick et al., 2017). From our searches, the top five potential off-target genomic sites that had an intact PAM cleavage sequence (NGG) still had two or more mismatches in the 13-bp ‘seed sequence’ proximal to the PAM site (Table S1). It was shown previously that two or more mismatches within the gRNA seed sequence are sufficient to inhibit Cas9-mediated cleavage at off-target sites (Hsu et al., 2013; Pattanayak et al., 2013). Therefore, these top potential off-target loci are not likely to be cleaved by Sema3F gRNA1. Nonetheless, to ensure that Sema3F gRNA1 did not cleave potential off-target sites, we isolated genomic DNA from the same five pooled Sema3F CRISPR injected embryos (T26) from which we calculated on-target mutagenesis efficiency (see ‘Efficiency of mutagenesis at the Sema3F locus’ above) and sequenced ten clones from the top three off-target regions [Hox3, g-variable lymphocyte receptor (gVLR), ABCB7; Table S1]. Our results suggested that our Sema3F gRNA was highly specific, with no evidence of a tendency to induce mutations at the top three potential off-target loci (Fig. S6). The following primers were used to amplify potential off-target regions: Hox3, forward: 5′-AGCAGGGTGCCTACAACATC-3′, reverse: 5′-GCTGTCCACGTATCCTCCTC-3′; gVLR, forward: 5′-CCGCTCACTACCAAACCATT-3′, reverse: 5′-AACATACGTTTTGGGGCAAG-3′; and ABCD7, forward: 5′-GAGAGAGACGCAAGGAAGG-3′, reverse: 5′-GGCTGAGTAGACCCAACTCG-3′.
Sectioning and imaging
Whole-mount embryos stained by in situ hybridization and immunohistochemistry were mounted in 75% glycerol and photographed on a Zeiss Discovery V12 stereomicroscope. For sectioning, selected embryos were embedded in 5% agarose and sectioned using a Vibratome (Pelco 101, Series 1000) (20 μm). Sections were mounted in 75% glycerol on a coverslipped glass slide and photographed on a Zeiss Axioimager Z1 compound microscope using Zeiss Axiovision software (v 4.7). Fluorescent visualization of Alcian Blue-stained lamprey cellular cartilage bars is described elsewhere (Martin et al., 2009). Image stacks of fluorescent lamprey cartilage and high-magnification Hu immunostaining of cranial sensory ganglia were rendered as maximum intensity projections using the Inside 4D module of the Axiovision software package. All figures were assembled using Adobe Photoshop CS5.5.
Acknowledgements
We thank Skye Fissette and Dr Weiming Li for collecting and shipping spawning adult lampreys. We also thank Drs Michael Hansen and Nicholas Johnson, and the Hammond Bay Biological Station staff for providing resources to collect and ship the lampreys used in this study.
Footnotes
Author contributions
Conceptualization: J.R.Y., D.W.M.; Methodology: J.R.Y., D.W.M.; Validation: J.R.Y.; Formal analysis: J.R.Y.; Investigation: T.Y., O.L., D.W.M.; Writing - original draft: J.R.Y.; Writing - review & editing: D.W.M.; Supervision: D.W.M.; Project administration: D.W.M.; Funding acquisition: D.W.M.
Funding
Funding was provided by the University of Oklahoma Faculty Investment Program to D.W.M., and by the University of Oklahoma M. Blanche Adams and M. Frances Adams Graduate Student Scholarship and Graduate Student Senate to J.R.Y.
References
Competing interests
The authors declare no competing or financial interests.