Development of the viscerocranial skeleton during embryogenesis of the sea lamprey, Petromyzon Marinus



Evolution of the skeleton was a key transition in early vertebrates. Lampreys lack a mineralized skeleton but possess cartilaginous neurocranial and viscerocranial elements. In lampreys, the visceral skeleton develops as a fused branchial basket supporting the pharynx. Here, we have adapted Alcian blue staining of lamprey cartilage and show this method results in cartilage fluorescence that we used to describe development of the branchial skeleton in Petromyzon marinus between 17 and 63 days of development. We show that skeletal rods develop from condensations of flattened discoidal chondrocytes and may involve cellular intercalation. Lamprey trabecular, parachordal, and subchordal cartilages consist of aggregations of polygonal chondrocytes positioned on the ventral and lateral surfaces of the notochord. We speculate that morphological differences relate to functional differences in the cartilage. We show that differentiated skeletal rods are derived from neural crest. Finally, we show how branchial muscles intercalate with skeletal rods of the branchial basket. Developmental Dynamics 238:3126–3138, 2009. © 2009 Wiley-Liss, Inc.


The presence of an internal supporting skeleton is a defining character of vertebrates. Vertebrate skeletal elements are of dual embryonic origin: the axial and appendicular skeletal elements as well as parts of the skull are derived from mesoderm, whereas splanchnocranial (branchial) elements and the remainder of the skull develop from the neural crest, another hallmark innovation in vertebrate evolution (De Beer, 1937; Gans and Northcutt, 1983; Northcutt and Gans, 1983; Couly et al., 1993; Le Douarin and Kalcheim, 1999; Gross and Hanken, 2005; Hanken and Gross, 2005). Among basal vertebrates, the neural crest-derived visceral, or branchial, skeleton supports gills and muscles of the pharynx, and the evolution of a muscularized pharynx was likely to have been a key innovation in the transition from a sedentary filter-feeding lifestyle to active predation in early vertebrate evolution (Northcutt and Gans, 1983). Thus, an increased understanding of skeletogenesis in extant basal vertebrates may provide a greater understanding of the origin and evolution of the vertebrate skeleton. Unfortunately, little is known about origins of skeletogenesis in the vertebrate ancestor. The first vertebrate endoskeleton was likely cartilaginous (Janvier, 1996), but the origins of neural crest and mesodermal contributions are not fully understood (Hall, 1999).

Jawless vertebrates, or agnathans, arose more than 400 million years ago (Janvier, 1996), but only lampreys and hagfish survive as relicts from this group and most likely represent a monophyletic “Cyclostome” clade (Stock and Whitt, 1992; also reviewed in Osorio and Retaux, 2008). Despite recent major advances in hagfish embryology (Ota et al., 2007), only lamprey embryos are still easily accessible for study. Inquiries into the evolution of cartilage in vertebrates call for the need to understand chondrogenesis in these relict species.

In lampreys, the branchial skeleton exists as a cartilaginous fused branchial basket supporting the muscularized pharynx (Fig. 1). Chondrogenesis of the lamprey branchial skeleton was characterized at cellular and ultrastructural levels by Morrison and her colleagues (2000) and development of the branchial basket was recently described using the Weigert staining method for elastin (Yao et al., 2008). The lamprey branchial basket is thought to be derived from the neural crest, as inferred from the results of ablation and cell labeling studies (Langille and Hall, 1988; McCauley and Bronner-Fraser, 2003). A pair of trabecular neurocranial cartilages supports the brain rostrally. The trabecular cartilage has been suggested to be a rostral extension of mesodermally derived parachordal cartilage (Sewertzoff, 1916; Johnels, 1948), and of mesodermal origin itself, a view that was supported by Kuratani and colleagues (Kuratani et al., 2004). Alternatively, ablation experiments in which the neural crest was removed from regions along the anteroposterior axis have suggested a neural crest origin for the trabecular cartilage (Langille and Hall, 1988). However, the possibility remains that mesoderm might have accompanied the removal of neural crest in these experiments, leaving uncertain the embryonic origin of the lamprey trabeculae. Lateral and subjacent to the notochord, lamprey parachordal and subchordal bars lie in close proximity to the notochord and stabilize dorsal projections of the branchial basket skeletal bars (Fig. 1), but the nature of the lamprey parachordal and subchordal contacts with putatively neural crest-derived branchial basket skeletal rods is not understood. Ventrally, the skeletal rod in each arch (PA3–9) is fused to the hypobranchial bar; seven gill openings are supported by epitrematic and hypotrematic skeletal processes that project rostrally from each rod with the rostral-most processes on pharyngeal arch 3 being fused to form a closed looping structure (Fig. 1).

Figure 1.

Diagram of the lamprey branchial basket and trabecula. A left-side view illustrating the trabecula and branchial basket of a stage 30 proammocoete shows the cranial and pharyngeal cartilages and their arrangement along the anteroposterior axis (dorsal to the top; anterior to the left). Dorsal, T and PC are positioned lateral to the NT and SC cartilages are positioned along its ventral surface. Pharyngeal arch cartilages PA3–9 are fused dorsally to the PC and SC cartilages, and ventrally to the HBB. Projecting rostral from each PA are the EP and HP that support external openings along the pharynx. EP and HP are fused to each other on PA3 to form a closed looping structure. EP, epitrematic process; HBB, hypobranchial bar; HP, hypotrematic process; NT, notochord; PA, pharyngeal arch; PC, parachordal; SC, subchordal; T, trabecula.

Lamprey cartilage has been determined to differ from gnathostome vertebrate cartilage in its protein composition (Wright et al., 1983, 2001; Robson et al., 1993; McBurney and Wright, 1996). While type II collagen is the major matrix protein of gnathostome vertebrates, the piston (lingual), annular (mouth), and neurocranial cartilages in lampreys consist of a lamprey-specific matrix protein, lamprin, and the branchial cartilages that support the pharynx and gills are composed of an as yet unidentified protein that is different from lamprin (Morrison et al., 2000; Wright et al., 1983). Moreover, both of these proteins appear to have biochemical properties similar to elastin (Wright et al., 1988). Indeed, a recent study has found that the Weigert staining method to detect the presence of elastin fibers is able to stain the lamprey branchial basket and trabecular cartilage elements (Yao et al., 2008). An additional study has shown that the elastin-like proteins of lamprey branchial cartilages are covalently linked by lysyl pyridinoline cross-linkages (Fernandes and Eyre, 1999), similar to the hydroxylysyl pyridinoline cross-links characteristic of collagen, although pyridinolines have not been found in vertebrate elastin (Fernandes and Eyre, 1999).

Apart from the recent study by Yao and colleagues (2008), whole-mount studies of chondrogenesis in the lamprey have been hindered by the inability of cartilage markers, such as the common chondrogenic stain Alcian blue, to stain intact cartilage in whole-mount ammocoete larvae. However, Alcian blue has been used to stain lamprey cartilage in histological sections (Langille and Hall, 1988, McBurney and Wright, 1996). It is not understood if the unique protein composition of cartilage in lampreys contributes to these difficulties in staining chondrogenic elements.

To date, methods to analyze developing lamprey cartilage in whole-mount fixed embryos at high resolution have not been available. Alcian blue methods have been described only on sections, and the recent whole-mount description of chondrogenesis using the Weigert method was constrained due to optical limitations of light microscopy (Yao et al., 2008).

We have now developed a reliable protocol to stain intact cartilage in fixed whole-mount developing embryos using Alcian blue. Furthermore, after Alcian blue staining, the cartilage was found to fluoresce, facilitating optical sectioning and three-dimensional reconstruction of the developing branchial basket. Here, we show precise changes that occur during chondrogenesis in lamprey embryos and illustrate differences in cellular morphology among chondrocytes of the different cartilage bars in the branchial basket, parachordal, subchordal, and trabecular cartilages. We also show the developing branchial basket cartilage in the context of the surrounding branchial muscles, further facilitating their study in the context of the developing larva. Finally, we use the lineage tracer DiI (1,1′, di-octadecyl-3,3,3′,3′,-tetramethylindo-carbocyanine perchlorate) to demonstrate for the first time that differentiated chondrocytes in the skeletal bars are definitively derived from neural crest, as has been inferred in previous ablation and cell lineage studies (Langille and Hall, 1988; McCauley and Bronner-Fraser, 2006).

Results described here form a foundation for understanding chondrogenic developmental mechanisms in lampreys and will be useful for interpreting results of ongoing perturbation experiments that address the evolution and development of the vertebrate viscerocranial skeleton.


Fluorescence Spectroscopy

Branchial skeletons were excised from several 30-day-old individual ammocoetes, placed into a cuvette, and analyzed for fluorescent properties by spectroscopy. The excitation spectrum of the Alcian blue-stained cartilage peaked at 365 nm with a 360–375 nm FWHM (full width at half maximum). Spectra of the cartilage were acquired on the cartilage bars and compared with spectra of the background away from the cartilage (Fig. 2). All spectra were corrected for dark noise in the spectrometer as well as the beam splitter emission filter transmission spectra. A series of emission filters were used so that a wider spectral range of emission could be explored and results were averaged into a single graph (Fig. 2m). The mounting media away from the cartilage bars exhibited a broad fluorescence spectrum FWHM ranging between 475 and 590 nm. The cartilage bar spectra showed a distinct peak at 488 nm with a FWHM of 40 nm, suggesting the cartilage exhibited a distinct fluorescent property. We have used a simple multiplicative scaling factor to make the spectra taken with different emission filters overlap. This procedure was necessary because the optical alignment with the specimen was different after changing the emission filters. Different cartilage samples from multiple individual ammocoetes produced the same emission spectra.

Figure 2.

Initiation of chondrogenesis in lamprey. Images are oriented anterior facing left. a–c: Seventeen-day-old larva; skeletal rods are present in pharyngeal arches (PA) 3–5. b, inset in a; c, same as b, shows fluorescence overlapping Alcian blue stained tissue. d: Eighteen-day-old larva; five Alcian blue stained skeletal rods are visible in PA3–7. e: Inset in d; fluorescence in each skeletal rod. Fluorescent cells are first observed at the mid-dorsoventral region within each arch. f: Nineteen-day-old larva; mucocartilage is prominent ventrally and in PA1, but does not show fluorescence. g,h: Epitrematic (ep) and hypotrematic (hp) processes extend rostrally beginning at 19 days (g) but are not present on caudal skeletal rods (h). i,j: Twenty-day-old larva; epitrematic and hypotrematic processes present in PA3 fuse to form a D-shaped loop; condensation of prechondrocytes to form a loop is visible at 20 days (i) but fluorescence at the rostral tip of the loop is absent (j). Skeletal rods are present in PA3–8 (j) but epitrematic and hypotrematic process are visible only on PA3–5 at 20 days (black arrows in i). Dorsally, PA3 projects lateral to the notochord (arrow in j); the ventral end of PA3 projects caudally, toward PA4 (arrowhead in j). k,l: Transverse section through PA3 at high magnification shows Alcian blue-stained fluorescent chondrocytes in the mid-region of PA3 in an 18-day-old larva. Undifferentiated prechondrocytes are not labeled with Alcian blue and are visible dorsal and ventral to labeled cells. The same population of cells that are Alcian blue-stained are also fluorescent (compare k to l). Note the presence of a stained fluorescent cell immediately adjacent to an unlabeled nonfluorescent cell (arrows in k,l). m: cartilage fluorescence measured in counts per second with an emission peak at approximately 488 nm (asterisk). The graph shown represents the averaged value for emission spectra measured using five filter sets and was subtracted from background fluorescence in the same media measured away from the cartilage sample (background fluorescence not shown). PA pharyngeal arch, ep epitrematic process, hp hypotrematic process. Scale bars = 100 μm in a–f,i,j, 50 μm in g,h, 25 μm in k,l.

Cartilage Differentiation in the Developing Lamprey

Condensation of prechondrogenic cells in branchial arches of P. marinus begins by stage 26, at approximately 15 days of development under standard rearing conditions (Piavis, 1961; Morrison et al., 2000; McCauley and Bronner-Fraser, 2006). In lampreys, the first and second pharyngeal arches give rise to the velum and supporting tissue but do not contribute to the branchial skeleton. Instead, the most rostral branchial basket element develops within the 3rd pharyngeal arch, PA3 (Fig. 1). Here, we follow previously established nomenclature in identification of the location of branchial basket cartilage from the third pharyngeal arch (PA3) to the most caudal elements present in pharyngeal arch 9 (PA9; Sewertzoff, 1916, 1917; Johnels, 1948; Takio et al., 2004, 2007; Yao et al., 2008).

Our protocol to label cartilage in developing lampreys induced the differentiated cartilage to fluoresce to the exclusion of other tissues (Fig. 2). Of interest, fluorescence properties were distinct from the Alcian blue chromogenic property because some Alcian blue-stained structures, such as mucocartilage that began to develop in the larval head by 18–19 days (Fig. 2d–f), were not fluorescent and did not obscure our ability to image underlying cartilaginous skeletal elements. Fluorescence of cartilage also was not observed in cleared unstained animals (data not shown). We noted broad fluorescence emission spectra with maximal emission at λ488nm (Fig. 2m). The broad fluorescence signal allowed for imaging using either a DAPI (4′,6-diamidine-2-phenylidole-dihydrochloride) filter set (Zeiss Filter Set 49: Ex365; Em445/50) or a green fluorescent protein (GFP) filter set (Zeiss Filter Set 38: ExBP470/40; EmBP525/50).

The staining of chondrocytes by Alcian blue suggested cartilage differentiation had begun by stage 27 because the first stained cartilage cells were visible in PA3 at this stage, 17 days after fertilization (Fig. 2a–c). At this time, differentiated chondrocytes were visible by both transmitted light and fluorescence (compare Fig. 2b and Fig. 2c). After 18 days, differentiated cartilage cells appeared in PA3–7 in the mid-region of each arch but no elements were yet visible in the most dorsal or ventral regions of any arch (Fig. 2d,e). By 19 days after fertilization, rostral projections of epitrematic and hypotrematic processes were visible on PA3 (Fig. 2g). At 20 days of development, epitrematic and hypotrematic processes were seen on PA3, and were visible on PA4 and PA5 in some Alcian-stained embryos (Fig. 2i). By 20 days, the ventral most part of the skeletal rod in PA3 projected caudally, and at its most dorsal limit projected medially toward the lateral surface of the notochord caudal to the otic vesicle. Both the dorsal and ventral-most projections of the remaining caudal bars of the developing branchial basket (PA3–8) were all found to project in the rostral direction and dorsally all were located subjacent to the notochord (Fig. 2j). Alcian blue staining, an indicator of cartilage differentiation, appeared to initiate from a single region within the precartilage condensation, in the mid-dorsoventral region of pharyngeal arches PA3–5 (Fig. 2a–c) and proceeded in both dorsal and ventral directions (Fig. 2k,l). Note the interface of Alcian blue stained fluorescent cells (below arrows in Fig. 2k,l) immediately adjacent to unstained nonfluorescent prechondrocytes immediately dorsal (above arrows in Fig. 2k,l). By 22 days postfertilization, the epitrematic and hypotrematic processes projecting from PA3 were fused to form a loop and both epitrematic and hypotrematic processes were visible in PA3–5 (Fig. 3a). Ventrally, the hypobranchial bar had fused PA3 to PA4. Cells of the bilateral trabecular cartilages could be seen adjacent to the notochord by day 22 (Fig. 3a) and projected in the rostrolateral direction beyond the rostral limit of the notochord by day 28 (Fig. 3a–d). Trabecular elongation appeared to occur in the rostral direction as has been suggested previously (Johnels, 1948; McBurney and Wright, 1996). After 24 days, epitrematic and hypotrematic processes were seen on all but the most caudal branchial bar (PA9) and all bars were fused ventrally at this time (Fig. 3b). After 26 days, trematic processes were visible on the PA9 skeletal rod (Fig. 3c) and subchordal bars had fused PA7–9 (Fig. 3c). By 28 days, the parachordal protrusion of PA3 had projected rostrally and lay adjacent to the notochord medial to the caudal hemisphere of the otic vesicle (Fig. 3c,d).

Figure 3.

a: Twenty-two-day-old larva; eight skeletal rods are visible (PA3–8). Cells of trabeculae are visible (arrowhead in a) and PA3 is joined to PA4 by condensation of the hypobranchial bar. Fusion of loop in PA3 is completed. b: By 24 days of development, seven skeletal rods are present (PA3–9). Epitrematic and hypotrematic processes are visible (PA4–8), but lacking on PA9. The hypobranchial bar has fused all skeletal rods ventrally. Extension of the trabecular is prominent. c,d: At days 26 and 28, the trabecular continues to elongate and extends beyond the most rostral extension of the notochord. The parachordal cartilage lies adjacent to the notochord medial to the caudal hemisphere of the otic vesicle visible in c and d and is fused with the dorsal extension of the PA3 skeletal rod. d: Caudal skeletal rods (PA4–9) attach to subchordal cartilage condensations beneath the notochord. hbb, hypobranchial bar; n, notochord; ov, otic vesicle; pc, parachordal cartilage; sc, subchordacartilage; t, trabecular. Scale bars = 100 μm in a–d.

The complete branchial basket had formed by 30 days postfertilization (Fig. 4). High resolution imaging allowed us to observe differences in chondrocyte morphology in situ as previously described in sections (McBurney and Wright, 1996; Morrison et al., 2000). Although chondrocytes of the branchial basket have been described as having the appearance of stacked cells (Morrison et al., 2000), not all cells show this morphology. Of interest, cartilage cells contacting the notochord (trabecular, parachordal, subchordal) have a more polygonal irregular appearance than the tightly stacked discoidal cartilage cells of the vertical branchial bars (Fig. 4b–e). Cells of the fused loop present in PA3, the epitrematic and hypotrematic processes of the remaining caudal arches (PA4–9) and the hypobranchial bars also do not appear to have the “stack of coins” morphology of the skeletal rods in PA3–9 (Fig. 4).

Figure 4.

a: Thirty-day-old larva. Chondrogenesis of the branchial basket is complete by 30 days epitrematic and hypotrematic processes are present on skeletal rods in each pharyngeal arch. All rods are attached to the ventral hypobranchial bar, and dorsally to parachordal or subchordal elements. b–k: Images correspond to enlargements of insets in a. b: Trabecular cartilage is made up of loosely aggregated cells ∼four to six cell diameters thick. c: Medially, the trabecular and parachordal cartilage elements attach to the notochord medial to rostral and caudal hemispheres of the otic vesicle, respectively. d,e: Skeletal rods in each arch attach to subchordal cartilages at the ventral surface of the notochord. f,h,i,k Skeletal rods in pharyngeal arches are comprised of chondrocytes stacks arranged as disks (2–3 cells per layer) in a regular arrangement. g: enlargement of inset g′ in (a). the hypobranchial bar forms from rostral extensions of each skeletal rod that fuse to the next most rostral skeletal rod. i-k: The cells that form the fused loop in PA3, epitrematic, and hypotrematic processes, and the hypobranchial bar are less regularly arranged. h: enlargement of inset h′ in a. The dorsal and ventral regions of skeletal rods in PA8 and PA9 are discontinuous (arrowheads in h). j,k: Note the morphological difference in cells of the hypobranchial bar (j) and epitrematic process (k) from the adjoining skeletal rods. j: The hypobranchial bar forms as rostral extensions at the ventral end of each skeletal rod that fuse with the adjacent anterior rod (arrowhead). Anterior faces left for each panel. ep, epitrematic; hbb, hypobranchial bar; ov, otic vesicle; PA, pharyngeal arch; pc, parachordal; sc, subchordal; t, trabecular. Scale bar = 100 μm in a.

Dorsoventral Separation of Branchial Basket Skeletal Rods

In many embryos, we observed that the dorsal condensation of branchial bars did not appear to be fused as a single unit to the ventral condensation within the same skeletal rod (Fig. 4a,h; PA8 and PA9; see also Fig. 4a,c). Of 68 proammocoetes observed (ages 26–54 days old), 21 animals (31%) possessed mid-ventral separation of skeletal rods within the branchial basket. Most commonly (15/21), the dorsoventral separation was present on PA9, the most caudal branchial arch (Fig. 4a,h), but these skeletal bar separations were also found to be present on each branchial arch with the exception of PA7 (Table 1). In addition, three prolarvae examined contained separate dorsal and ventral cartilage elements along multiple branchial arches (Table 1; Fig. 3h) and in another three animals the dorsoventral separation was present on both bilateral halves of the branchial basket, in each case on the skeletal rod in PA9. A 42-day-old proammocoetes larva was found to have the dorsoventral cartilage separation at the mid-ventral region of PA3 (Fig. 5c), and in this individual the cells of the epitrematic region of the loop in PA3 appeared loosely clustered, whereas the cells of the hypotrematic region were more stack-like in their appearance. These observations suggest there may be dorsoventral variation in development of the lamprey branchial basket.

Figure 5.

a: Forty-two-day-old larva; inset in a shown in c. Cells of the epitrematic process (ep) have the rounded morphological appearance of trabecular or parachordal cells (compare with Fig. 3b), whereas cells of the hypotrematic process (hp) maintain the regular morphology observed in skeletal rods. c: Discontinuity between dorsal and ventral halves of the PA3 skeletal rod were also observed (arrowheads). b: The eye (ey) was visibly fluorescent by 54 days of development. n, notochord; t, trabecular. Scale bars = 100 μm.

Table 1. Separation of Dorsal and Ventral Cartilage within Skeletal Rods of the Branchial Basket
 Pharyngeal arch position 
Number of larvae with cartilage separation in individual skeletal rods21110415 
Arch positions of multiple separated skeletal rods within 3 individual larvae+ +    
Number of larvae with no cartilage separation       47

Our results show that the hypobranchial bar (hbb) is not a single cartilage element of the branchial basket, but instead develops as a ventrocaudal extension of the PA3 skeletal bar that fuses with a ventrorostral extension of PA4, and rostral extensions of skeletal bars in PA5–9 (Fig. 4a,g,j) that fuse ventrally with each adjacent rostral skeletal rod. Chondrocytes of the hypobranchial bar and the epitrematic and hypotrematic processes also appear more irregular in shape than the discoidal chondrocyte stacks of the vertical skeletal rods in each PA (Fig. 4j,k).

To more fully characterize the two cellular morphologies of differentiated chondrocytes among the cartilage elements, we imaged the cartilage in a 63-day-old (8-mm) ammocoete (Fig. 6). A previous description of lamprey cartilage (Morrison et al., 2000) reported the flattened discoidal appearance among the cells that form the skeletal rods and polygonal cells forming the trabecular cartilage. Both cell morphologies appear in the same individual but do not appear to mix (see below). In the 8-mm 63-day-old ammocoetes larva, polygonal cells of the trabecular cartilage lay apposed to the notochord and projected in the anterior direction as elongate clusters of cells (Fig. 6a,d,e; pseudo-colored red in Fig. 6e). At its most caudal point, the trabecular cartilage lay medial to the rostral hemisphere of the otic vesicle, adjacent to the notochord, but was not fused with the rostral extension of the parachordal cartilage that had attached to the discoidal chondrocytes of the skeletal bar located in PA3 (Fig. 6d,e; discoidal chondrocytes pseudo-colored green in Fig. 6e,f). This observation differs from the reports of Johnels (1948) and Parker (1883) that the trabecular and parachordals were fused in the early ammocoetes. Developing eyes were visible lateral to the trabecule located at the level of the most rostral extension of the notochord (Fig. 6a,b). Subchordal cartilage bars are shown in Figure 6d,f. Polygonal chondrocytes of subchordal cartilages on the ventral surface of the notochord (Fig. 6f; pseudo-colored red) were fused to one another only between PA7–9 (Fig. 6d,f), and each subchordal cluster of cells appeared to join each of the skeletal rods to its dorsomedial position at the ventral surface of the notochord (Fig. 6d,f). We also observed an abrupt transition of the polygonal parachordal and subchordal chondrocytes into the discoidal chondrocytes characteristic of the skeletal bars in each PA (arrowheads in Fig. 6d,f). Occasionally, we noted the presence of a cartilage nodule as a short rostral extension within the epitrematic region of the cartilage loop present in PA3 (Fig. 6b), although the significance of this cartilage extension is not known. Ventrally, the bilateral hypobranchial bars were positioned lateral to the endostyle (en) from PA3 caudal to the position of PA6, but lay adjacent to one another from PA7 to PA9. However, caudal to PA6 the two hypobranchial bars were not fused (Fig. 6c) as has been described elsewhere (Johnels, 1948).

Figure 6.

Sixty-three-day-old 8-mm larva. a: Dorsal view shows attachment of trabecular and parachordals to notochord, medial to the otic vesicle. a,b: Eyes are positioned dorsolateral to the trabecular cartilages, at the level of the most rostral extension of the notochord. c: Ventral view shows the anteroposterior extension of the hypobranchial bars. Caudal to the midpoint between pharyngeal arch 6 (PA6) andPA7, bilateral hypobranchial bars lie adjacent to one another but are not fused. Rostral to this point, the hypobranchial bars are separated by the ventrally positioned endostyle and extend anteriorly to fuse with PA3 (b,c). d: Ventral view of the notochord. Trabecular cartilages are positioned lateral to the notochord rostral to the otic vesicle. PA3 is attached to the parachordal cartilage lateral to the notochord. Remaining caudal skeletal rods (PA4–9) attach to the subchordal cartilages on the ventral surface of the notochord. e,f: Enlargement of insets in d, pseudo-colored to highlight the attachment of type I chondrocytes in the skeletal rods (pseudo-colored green) to the type II chondrocytes that give rise to the trabecular, parachordal, and subchordal cartilage (pseudo-colored red; arrowheads in d,f). Subchordal cartilage condensations have fused PA7–9 on one side (concave arrowhead in d) but only PA8–9 are attached to a single fused subchordal condensation on the opposite side (d,f). Cartilage bars rostral to PA7 are attached to individual subchordal condensations (PA4–6). ey eye, ep epitrematic, hp hypotrematic, hbb hypobranchial bar, ov otic vesicle, pc parachordal, sc subchordal, t trabecular. Scale bar = 100 μm.

Actin filaments were labeled using fluorescein-conjugated phalloidin to determine the distribution of cartilage in the context of the surrounding pharyngeal muscles (Fig. 7a,b). Pharyngeal muscles shown in Figure 7 have been pseudo-colored magenta to differentiate these branchial muscles from the overlapping emission spectra of the branchial basket cartilage (pseudo-colored green). Of interest, we found that the straight muscle fibers in each arch lie adjacent to curved sections of the cartilage bar (Fig. 6b). This suggests branchial muscle contraction may induce bending of the adjacent skeletal rod. After muscle contraction, the skeletal rods in each arch of the branchial basket would recoil into the relaxed position.

Figure 7.

a: Lateral view of 30-day-old larva, phalloidin-stained to highlight branchial muscles in each pharyngeal arch. Muscles have been pseudo-colored magenta; cartilage bars are pseudo-colored green. b: Z-series reconstructed from a 30-μm thick transverse section at the level of pharyngeal arch 5 (PA5) illustrates the position of branch muscles (bm) adjacent to the PA5 cartilage bars. Straight muscle fibers (magenta) are positioned adjacent to bowed sections of the skeletal rod (green). c: Section of PA3 skeletal bar showing fusion of hypotrematic process (hp) to PA3 skeletal rod. c: Each layer in the PA3 chondrocyte stack contains two or more cells (arrow indicates cell membrane separating two cells within a layer), whereas the hypotrematic process contained a single cell along its diameter (arrowhead). d–f: DiI (1,1′, di-octadecyl-3,3,3′,3′,-tetramethylindo-carbocyanine perchlorate) -labeled neural crest cells present in the cartilage bar of a 30-day-old larva. Differential interference contrast microscopy (d), DiI (e), and the merged image (f) show DiI-labeled cells within the cartilage stack. Multiple adjacent layers of labeled cells (arrowheads in e) suggest neural crest cells continue to divide after the completion of migration into the prechondrocyte stack. Pigmented melanocytes (m) are also DiI-labeled, indicating their neural crest origin (compare arrows in d, e; d,e merged in f). bm, branchial muscle; hp, hypotrematic process; m, melanocyte. Scale bar = 100 μm in a,b, 25 μm in c,d (d applies to e,f).

We reconstructed a Z-series collected from optical sections of the cartilage loop in PA3 in a 30 day larva where the hypotrematic region fuses with the branchial bar of PA3 (Fig. 7c). Cells of the hypotrematic process appeared to be intermediate to the discoidal cells of the chondrocyte stacks within the branchial bars, and the irregularly shaped polygonal chondrocytes adjacent to the notochord (compare with Fig. 6e,f). Whereas each layer in the stack of chondrocytes contains at least two cells (Fig. 7c; see also Morrison et al., 2000), there appear to be only one to two cells along the narrower diameter of the trematic processes (Fig. 7c).

Neural Crest Origin of Branchial Basket

Previous data have been used to infer that cartilage of the branchial basket is derived from the neural crest. Our previous cell labeling experiments with the lipophilic dye DiI (McCauley and Bronner-Fraser, 2003, 2006) suggested that prechondrocytes in a stage 26 embryo were derived from neural crest cells. However, in those experiments, we were not able to show that differentiated chondrocytes were derived from neural crest. Here, we show in a 1-month-old proammocoete larva that neural crest cells labeled with DiI during early development (stage 22; 6 days) contributed to the cartilage stacks of the branchial bars (Fig. 7d–f). This result clearly demonstrates the neural crest origin of chondrocytes within the branchial basket skeletal rods, but it is still unclear if other cartilages (trabecular, parachordal, subchordal) are also of neural crest origin because labeled neural crest cells were never found to contribute to these structures (Langille and Hall, 1988; Kuratani et al., 2004). Results from our cell-labeling experiment suggest that neural crest cells migrating into a prechondrocyte stack may continue to divide after migration. Labeled cells were not randomly distributed along the skeletal rod. Instead, we found that within the cartilage bars DiI-labeled cells were positioned in clusters that each contained four to five layers of labeled cells (Fig. 7d–f). However, we also noted the presence of single DiI-labeled cells within the skeletal rod that were immediately adjacent to unlabeled cells (Fig. 7e,f). We found that contiguous DiI-labeled clusters of neural crest were also separated by unlabeled cells. These observations suggest the condensation of prechondrocytes into stacks may involve cell intercalation. Finally, we also noted the presence of DiI-labeled melanocytes within the branchial arches, indicating their neural crest origin (Fig. 7d–f).


We have described at cellular resolution the spatial and temporal development of branchial basket skeletal rods, trabecular, parachordal, and subchordal cartilages during various stages of development in the sea lamprey. We show that differentiation of the branchial basket in P. marinus occurs between 17 and 30 days after fertilization. Cartilage differentiation within the stacked prechondrocytes of developing skeletal rods appears to begin at the mid-dorsoventral region of each branchial arch and proceeds in both dorsal and ventral directions (Fig. 2) after the intercalation of neural crest cells to form the stacked skeletal rods (Fig. 7).

Our results highlight morphological differences in the chondrocytes that contribute to the different elements of the branchial basket. We show that the dorsal subchordal chondrocytes are morphologically distinct from ventral cartilages of the hypobranchial bars. In 31% of animals observed, we also found a separation of dorsal and ventral elements within the branchial bars, and always at the mid-dorsoventral region of the cartilage bars (Figs. 3, 4). These breaks were not confined to the cartilage within any one branchial arch but could be present at this position in any cartilage bar along the anteroposterior axis of the branchial basket (Table 1; Figs. 3h, 4c). It is unclear if the dorsoventral cartilage separation we observed in the developing branchial basket skeletal rods represents normal variation seen during development but it is tempting to speculate that such variation could have been important for establishing the articulation present in the gnathostome branchial skeleton. Nevertheless, these differences suggest that the dorsal and ventral elements of the branchial basket may consist of two separate embryonic components. In jawed vertebrates, the nested expression of dlx genes is important for establishing dorsoventral (proximodistal) polarity in branchial arches (Panganiban and Rubenstein, 2002; Depew et al., 2005), although Langeland and his colleagues (Neidert et al., 2001) reported no dorsoventral patterning of dlx expression in lamprey. However, our results and those reported elsewhere (Yao et al., 2008) suggest dorsoventral differences in the branchial cartilage elements may be present in this basal vertebrate and the ability to separate these skeletal elements developmentally might have been a key factor in the evolution of branchial articulation, and cartilage segmentation within vertebrates (Crotwell and Mabee, 2007). A recent study suggests that developmental mechanisms patterning the branchial arches and paired fin skeletons are shared in vertebrates (Gillis et al., 2009). Considering the observation that the lamprey median fin may be patterned by similar mechanisms that regulate development of paired appendages (Freitas et al., 2006), it will be informative to determine the degree to which developmental mechanisms patterning the gnathostome branchial arches are conserved in lampreys (McCauley and Bronner-Fraser, 2004).

Recent investigations have begun to shed light on the molecular mechanisms that regulate chondrogenesis in lampreys (McCauley and Bronner-Fraser, 2006; Zhang et al., 2006; Ohtani et al., 2008; McCauley, 2008). It has been demonstrated that the major extracellular matrix (ECM) protein of the lamprey piston (tongue), annular (mouth), neurocranial, and trabecular cartilage in lampreys is lamprin, an elastin-like protein that is unique to lampreys (McBurney et al., 1996a; McBurney and Wright, 1996; Wright et al., 1983) but lamprin was not detected in branchial or pericardial cartilage (McBurney et al., 1996b). Although the protein content of branchial cartilage is not identical to that of trabecular cartilage (McBurney et al., 1996b), similar Alcian blue staining and fluorescence properties of these cartilage elements allowed us to analyze cellular morphology in whole-mount embryos and larvae. A recent description of cartilage development in lamprey used the Weigert method to stain elastin fibers (Yao et al., 2008). In that study, both trabecular and branchial cartilages were visible, suggesting some elastin-like proteins of the trabeculae and branchial cartilage have similar biochemical properties.

A previous description of development of the branchial basket has suggested the branchial bars are fused dorsally by the subchordal bar, and mid-ventrallly by hypotrematic bars (Morrison et al., 2000). Our results indicate that, by 63 days of development, the subchordal bars are present as condensations on the ventral surface of the notochord but are fused only in the caudal region between PA7 and PA9 (Fig. 5d). At this time, hypotrematic processes are also present in each arch but do not interconnect adjacent branchial arches (Fig. 5b). The clustered appearance of irregularly shaped polygonal parachordal and subchordal cartilage cells abruptly transitions into the organized stacks of flattened discoidal cells that form each branchial bar (Fig. 5d–f). The trabecular cartilage contains only polygonal chondrocytes. The abrupt transition between these two cell morphologies and the position of such irregularly shaped cells only adjacent to the notochord suggests that morphological differences may be linked to the different functions of these cells within the skeleton. The discoidal chondrocytes of skeletal rods support the gills and other tissues of the pharynx. As the branchial muscles contract, curvature within the skeletal rods may allow them to flex. The elastin-like proteins of the skeletal rods (Morrison et al., 2000) could then permit a return to the relaxed skeletal morphology after cessation of branchial muscle contraction. Note that discoidal chondrocytes are located only in the dorsoventral axis of the branchial skeleton parallel to the axes of contraction of branchial muscles (Fig. 6b). Longitudinal cartilage elements (trabecular, parachordal, subchordal, epitrematic, hypotrematic, hypobranchial) would not be subject to the contractile forces generated by dorsoventrally positioned branchial muscles. Instead, the parachordal and subchordal cartilage elements appear to provide commissural attachments for the skeletal rods developing in each pharyngeal arch.

Parker (1883) described the composition of the lamprey skeleton as arising from two types of true cartilage that he called “soft” and “hard.” Basicranial elements (trabeculae, parachordals) develop as “hard” cartilage and the branchial bars forming the branchial basket are made up of “soft” cartilage. These distinctions related to differential staining properties originally described by Müller (Müller, 1839; Parker, 1883) were also described by Gaskell (1908). Although our descriptions of chondrocytes as discoidal or polygonal cells are based on cellular morphological differences, discoidal chondrocytes appear to correspond to Parker's “soft” cartilage of the branchial basket, whereas the irregularly shaped polygonal cells make up the “hard” cartilage of the trabeculae and parachordals (and subchordals). Discoidal chondrocytes also appear similar in appearance to the zellknorpel (cellular cartilage) that support gill filaments in teleosts but lack extensive intercellular matrix deposition (Schaffer, 1930; Benjamin, 1990).

Our description of the chondrocytes that comprise the lamprey viscerocranial skeleton are strikingly similar to descriptions of pharyngeal cartilage in zebrafish (Kimmel et al., 1998). Ceratobranchial cartilages have a similar coin stack appearance as the discoidal chondrocytes of the lamprey. The “stack of coins” arrangement of chondrocytes that describe the symplectic region of the hyosymplectic transitions through an interhyal joint region into the polygonal chondrocytes of the hyomandibular cartilage. This gradual transition in form differs from the abrupt transition from discoidal to polygonal chondrocytes seen in lampreys between the skeletal rods and subchordal or parachordal cartilages adjacent to the notochord (Fig. 6). Additionally, whereas in the zebrafish each layer in a “stack” consists of a single cell, in lamprey skeletal rods, a layer in the stack may contain more than a single chondrocyte (Fig. 7c; see also Morrison et al., 2000).

We have shown in this study that chondrocytes contributing to the branchial skeletal rods are derived at least in part from neural crest that migrate into the presumptive pharyngeal region (McCauley and Bronner-Fraser, 2006). However, it still remains possible that skeletal rods receive contributions from both mesoderm and neural crest cells because labeled cells are separated within stacks by unlabeled cells of unknown origin. Condensation of the skeletal rods may involve intercalation of prechondrogenic cells to form precartilage stacks that subsequently undergo differentiation. Of interest, cellular intercalation was also suggested as a mechanism to drive elongation of the zebrafish symplectic cartilage (Kimmel et al., 1998).

The origin of parachordal and subchordal cartilage elements is not known. Our data suggest independent origins for the trabecular and parachordal cartilages. The lamprey trabecular cartilage has been suggested to be a rostral extension of the parachordals (Parker, 1883; Johnels, 1948) and of mesodermal origin (Kuratani et al., 2004). We show here that the trabecular and parachordals arise independently in positions rostral and caudal to the otic vesicle respectively and after 2 months, an 8-mm ammocoetes larva still maintained separate parachordal and trabecular cartilage elements (Fig. 5a,e), perhaps reflecting the different functions of these cartilages; trabeculae support the rostral extension of the brain in the prochordal region; parachordals support the PA3 skeletal rods in their medial position. Subchordal cartilage condensations were described as early as 1879 (Schneider, 1879) and it was recognized early that the subchordal cartilages were most conspicuous in the regions where the skeletal bars arise (Gaskell, 1908). Schaffer (Schaffer, 1896) speculated the subchordal cartilage rod was discontinuous in the early developing ammocoete. Our results confirm Schaffer's speculation on the discontinuity of early subchordal cartilage and demonstrate that chondrocytes forming subchordals are of the same polygonal morphology as trabecular and parachordal cartilages.

The fluorescent property of lamprey cartilage remains enigmatic. Lamprey cartilage was induced to fluoresce at λ488nm. The same Alcian blue staining protocol applied to embryonic zebrafish cartilage did not induce fluorescence (data not shown), suggesting the fluorescent property observed here may be unique to lamprey cartilage and possibly related to the elastin-like ECM proteins that are specific to lamprey. Of interest, the major cross-linkages in lamprey branchial cartilage proteins have been found to be lysyl pyridinoline (LP) and hydroxylysyl pyridinoline (HP), with LP predominating. An earlier study showed that after branchial cartilage digestion by pancreatic elastase, a broad peak of fluorescence was detected that was indistinguishable from the fluorescence of a pyridinoline standard (Fernandes and Eyre, 1999). This observation suggests that cartilage fluorescence shown in this study may be related to the presence of LP cross-linkages present in the lamprey branchial cartilage.

In summary, we have shown at high resolution development of the lamprey viscerocranial skeleton and trabeculae that occur between 17 and 30 days after fertilization, with few further changes through 63 days of development. Two types of cartilage cells give rise to the visceral skeleton. Morphological differences between discoidal and polygonal chondrocytes may relate to their individual functions within the skeleton relative to the position of branchial muscles and correlate, respectively, to the “soft” and “hard” true cartilage described in the lamprey by Parker (1883) and Gaskell (1908). We show that discoidal chondrocytes of the skeletal bars are derived from neural crest cells but the origin of polygonal chondrocytes that form the remaining visceral skeletal elements remains unclear. Finally, our data suggest that previous descriptions of the lamprey branchial basket as a unitary fused structure with no dorsoventral differences ignores developmental variation that might have played a role in the evolution of the articulated viscerocranial skeleton found in jawed vertebrates.


Embryo Culture

Gravid adult lampreys were collected from streams near the Hammond Bay Biological Station (Millersburg, MI) and shipped to the University of Oklahoma. During the summer of 2008, eggs were collected from ovulating females and fertilized by expression of milt from spermiated males directly onto eggs. Fertilized eggs were rinsed two times with spring water and reared at 18°C (Piavis, 1961). Embryos were cleaned daily to remove the dead and transferred to clean water. Embryos were fixed at stages 25–30 in MEMFA as described (McCauley and Bronner-Fraser, 2003), dehydrated and stored in 100% methanol at −20°C. Embryos and pro-larvae were staged according to the developmental staging table described by Tahara (1988) for Lampetra reissneri. Animals used in this study were housed according to protocols approved by the University of Oklahoma Institutional Animal Care and Use Committee.

Alcian Blue Skeletal Staining

Whole-mount stage 26 embryos through 8-mm pro-ammocoete larvae (63 days after fertilization) were fixed for 1 hr at room temperature in MEMFA (4% formaldehyde, 0.1 M MOPS [pH 7.4], 2 mM EGTA, 1 mM MgSO4), rinsed 3 times 15 min in MEM salts (MEMFA lacking 4% formaldehyde), dehydrated through a methanol series, and then stored in 100% methanol (−20°C). Before Alcian blue staining, embryos were rehydrated into PBST (phosphate buffered saline, 0.1% Tween 20). Embryos were bleached in a 3% H2O2/1% KOH solution (10 min) and rinsed through an acid alcohol solution (0.37% HCl/70% EtOH). Alcian blue stock solution (0.4% Alcian blue/70% EtOH) was prepared and diluted to yield a final working solution of 0.1% Alcian blue in acid alcohol. Embryos were placed into the staining solution and rocked overnight (16–24 hr) at room temperature, then destained through a EtOH series (75% overnight, 50% 6 hr, 25% overnight) and transferred to glass vials. Embryos were rinsed 2 × 5 min in 30% saturated sodium tetraborate followed by 90-min treatment in 0.2 mg/ml trypsin, 0.2% Triton X-100, 30% saturated sodium tetraborate. Embryos were cleared with rocking in a glycerol/detergent solution (18% glycerol/0.8% KOH/0.2% Triton X-100) until the ectoderm peeled away. Embryos were cleared further in 50% glycerol/0.25% KOH for 24 hr, followed by 50% glycerol/0.1% KOH (6–24 hr) and stored in 80% glycerol/0.01% thymerisol at +4°C for subsequent observation and imaging.


Developmental stages were observed throughout chondrogenesis, beginning at stage 26–27 through stage 30 (Tahara, 1988). Cleared embryos were mounted on glass slides in 80% glycerol and cover slipped using clay feet to prevent lateral compression during observation. Imaging was performed on a Zeiss Axioimager Z1 equipped with a motorized Z stage, HBO100 Hg lamp, and the Zeiss Apotome module for optical sectioning. Alcian blue staining was imaged with transmitted light using differential interference contrast (DIC) optics. Induced fluorescence of the cartilage was imaged using a DAPI filter set (Ex365;Em445/50; Zeiss Filter Set 49), or GFP filter set (ExBP470/40;EmBP525/50; Zeiss Filter Set 38) due to the broad fluorescence emission spectrum. Z-stacks were rendered using the Inside4D module of the Zeiss Axiovision software (v4.7) and were transferred to a personal computer for assembly using Adobe Photoshop CS3. Rendered Z-stacks were saved as maximum intensity projections using Inside4D and exported as single images or as Quicktime movies.


Excitation and emission filters used for spectroscopy in different emission bands included BP 445/50, HQ600/200M, HQ480/100M, HQ540/120M, D460/120X, and D550/200M (Chroma). Local spectroscopy was performed using an Avantes AvaSpec-2048TEC-USB2, 280–800 nm, Resolution: 2.4 nm (full width at half maximum; FWHM) spectrometer coupled to the Axioimager system by means of a 200-μm core optical fiber mounted at the focal plane of the camera port. The optical fiber only collects the light falling on its core, which for the objective used for spectroscopic analysis (Zeiss 40x/0.75 EC Plan Neofluar) defines a 5-μm-diameter area of the sample. Spatial correlation between the image and the spectra (fiber core) was established using an x–y translator in the optical fiber mount to align it to the eye-piece cross-hair reticle. The excitation spectrum was measured using a 50-μm optical fiber connected to a diffuser (Ocean Optics CC-3) placed in the sample position.

Muscle Staining

Actin fibers were stained using phalloidin coupled to fluorescein (FITC). Briefly, FITC-phalloidin dissolved in 100% methanol was dried onto a clean microscope slide and resuspended in phosphate buffered saline. Alcian blue-stained embryos were incubated in a FITC-phalloidin solution 10 min and washed through several changes of PBS, mounted in 80% glycerol, cover slipped and observed using a GFP filter set. Due to overlap in emission spectra between cartilage and phalloidin-stained muscles, the branchial basket and muscles were pseudo-colored using Adobe Photoshop CS3, where both tissues were observed together.

DiI Labeling

Neural crest cells were labeled with DiI (Molecular Probes, Invitrogen) by pressure injection at stage 22 when neural crest cells are undergoing migration, as described previously (McCauley and Bronner-Fraser, 2003). Labeled embryos were cultured at 18°C for 30 days, fixed in MEMFA, sectioned (20–100 μm) using a Vibratome (Pelco 101 Series 1000, St. Louis), and imaged for fluorescence (ExBP550/25;EmBP605/70).


We thank Eric Lee, Rosemary Knapp, and Jim Langeland for insightful comments on this manuscript. We also thank two anonymous reviewers for insightful comments. D.W.M. was funded by the NIH and the NSF, and L.A.B. was funded by NSF CAREER, the Center for Physics in Nanostructures, NSF MRSEC, and AFOSR.