Logo of npgopennature.com homepage
Nature Communications
Nat Commun. 2011 Jun; 2: 373.
Published online 2011 Jun 28. doi:  10.1038/ncomms1355
PMCID: PMC3157150

Identification of vertebra-like elements and their possible differentiation from sclerotomes in the hagfish


The hagfish, a group of extant jawless fish, are known to lack true vertebrae and, for this reason, have often been excluded from the group Vertebrata. However, it has yet to be conclusively shown whether hagfish lack all vertebra-like structures, and whether their somites follow developmental processes and patterning distinct from those in lampreys and gnathostomes. Here we report the presence of vertebra-like cartilages in the in-shore hagfish, Eptatretus burgeri. These elements arise as small nodules occupying anatomical positions comparable to those of gnathostome vertebrae. Examination of hagfish embryos suggests that the ventromedial portion of a somite transforms into mesenchymal cells that express cognates of Pax1/9 and Twist, strikingly similar to the pattern of sclerotome development in gnathostomes. We conclude that the vertebra-like elements in the hagfish are homologous to gnathostome vertebrae, implying that this animal underwent secondary reduction of vertebrae in most of the trunk.

Vertebrae arise as segmental endoskeletal elements associated with the notochord, and are regarded as a vertebrate synapomorphy1,2,3,4,5. No such structures, nor the somite-derived vertebral primordial, the sclerotome, are present in non-vertebrate chordates such as amphioxus and the tunicates6,7. The vertebrae in gnathostomes, the jawed vertebrates, consist of dorsal and ventral elements1,2,3,4,5,8,9,10, of which only the dorsal elements are present in lampreys11, one of the two extant agnathan taxa (jawless fishes)1,2. Based on evidence that the vertebrae of the gnathostomes and lampreys are developmentally derived from the sclerotome12,13, it is thought that their developmental mechanisms share the same ancestral origin. However, another extant agnathan, the hagfish, has generally been thought to lack vertebrae; the axial supporting tissue of this animal has been described as consisting only of notochord and cartilaginous fin rays1,2,3,4,14,15,16,17. Because of this morphological trait, the hagfish was excluded from the Vertebrata in some textbooks1,2,5, although its vertebrate affinities, as observed in the brain and skull, are widely recognized1,2,3,4,5. This raises a number of questions; specifically, whether the hagfish truly lacks any form of vertebra-like structure throughout its body and life cycle, and whether hagfish somites in the embryonic trunk are dissimilar to those in lampreys and gnathostomes. Importantly, one of the early reports on hagfish anatomy, in fact, described segmental cartilaginous elements (putative vertebral elements) on the ventral aspect of the caudal notochord in Bdellostoma more than a century ago (Supplementary Fig. S1)18. These cartilaginous structures have not been re-examined, and have largely been overlooked by subsequent researchers16,17.

To address these questions, we conducted detailed anatomical and histological analyses on the in-shore Japanese hagfish, Eptatretus burgeri, and discovered putative vertebral elements in the caudalmost region of the body axis. These skeletal elements are small cartilaginous nodules mostly located on the ventral aspect of the notochord, reminiscent of the gnathostome hemal arch, which, if not segmented, are in register with myomeres. Furthermore, we obtained a series of early- to late-stage hagfish embryos. Their examination suggests that the ventromedial portions of somites transform into mesenchymal cell masses that express cognates of Pax1/9 and Twist, similar to vertebral primodia in gnathostomes12. On the basis of these observations, we propose that the apparent lack of vertebrae in the hagfish stems from secondary developmental suppression of the vertebral differentiation programme for most of the trunk somites.


The adult skeletal morphology

In adult specimens of E. burgeri, we found putative vertebral elements in the post-cloacal axis (Fig. 1a–e)18. Histochemical observation indicated that these structures were cartilaginous nodules; the chondrocytes in each nodule were separated by a thin layer of extracellular matrices that were stained with Alcian blue, which detects mucopolysaccharides, as in gnathostomes (Figs 1c–e and 2a,b)2,19. These cartilages consisted of a pair of elements on both sides of the dorsal aorta, a single median element beneath the aorta, and another pair adhering to the ventral aspect of the notochordal sheath (Fig. 1c,d). The distribution of these nodules was restricted to a part of the post-cloacal region that roughly corresponded to the position between the 80th and 100th myomeres and was dorsal to the mucus glands (Fig. 1a,b). Unlike the vertebrae of lampreys and gnathostomes (Fig. 2c–f), these cartilages did not show a clear segmental pattern in register with myomeres. The overall morphology of these elements was reminiscent of hemal arches in gnathostomes (Figs 1c and 2a,e)1,2. On the dorsal side of the notochord at the same axial level, the median plate-like cartilaginous element was associated with dorsal fin rays, which were also present at the post-cloacal level of the trunk. This cartilage was previously designated as the 'median dorsal bar'18. As seen in the transverse section of the most caudal level, the median dorsal bar was attached to a pair of cartilaginous arches on both sides of the notochord (Fig. 1e). The anatomical configuration of these cartilages is strikingly reminiscent of the neural spine and the paired neural arches in gnathostomes (Fig. 2e)1,2,3,4,5,18.

Figure 1
The putative vertebral elements of E. burgeri.
Figure 2
Comparison of axial skeletal elements.

Somites and their derivatives

To examine possible developmental similarities in putative vertebral elements between the hagfish and gnathostome vertebrae, we observed their embryonic development. Although we previously succeeded in collecting several early-stage hagfish embryos20, these were insufficient to trace the developmental processes of the somites and their derivatives. Hence, we conducted long-term incubation of fertilized hagfish eggs (Methods), and obtained 42 fertilized eggs at various developmental stages (Supplementary Table S1). We selected three different stages of hagfish embryo, designated the early-, middle-, and late-pharyngula stages, for analysis of somite development (Fig. 3; Supplementary Table S1). We used these three embryos to conduct detailed histological observations and used in situ hybridization to detect expression of Twist and Pax1/9 homologues (Figs 3 and 4a–l; Supplementary Figs S2 and S3), known to be specifically expressed in the sclerotomes of some gnathostomes12.

Figure 3
Three stages of E. burgeri embryos.
Figure 4
Somite derivatives of E. burgeri.

The early pharyngula of the hagfish developed less than 80 somites, which showed several different phases that enabled us to investigate the somitic differentiation process along the anterior–posterior embryonic axis (Figs 3a,d and 4a–f). At the level of the 60th to 70th somites, the ventromedial region of the epithelial somite, located at the lateral aspect of the notochord, expressed Twist and Pax1/9 genes (Fig. 4b,c), and at the level of the 30th to 40th somites, mesenchymal cells expressed the same genes at the ventromedial side around the notochord (Fig. 4d–f). We also observed that these mesenchymal cells (ventromedial cells) are subdivided into rostral and caudal populations, showing metameric patterning (Fig. 4d), as in some gnathostome sclerotomes4. However, this segmental pattern did not appear to be maintained in the adult cartilaginous state, as noted above (Fig. 2a,b).

Most of the trunk somites had formed by the middle pharyngula stage (Fig. 3b). We thus observed caudal somite derivatives (80th somite level) in the prospective post-cloacal region wherein the putative vertebrae differentiate (Fig. 3e). Histological observations at the caudal levels in this stage showed epithelial somites and mesenchymal cells at the lateral aspect of the neural tube and notochord (Fig. 4g). Expression of Pax1/9 and Twist were also observed in those cells at caudal levels in the middle pharyngula stage (Fig. 4h). In particular, Twist expression was observed extensively in mesenchyme ventral to the notochord (Supplementary Fig. S4), whereas Pax1/9 expression was restricted to the abovementioned ventromedial cells at the caudal level in this stage (Fig. 4h). The dorsolateral epithelial portion of the somites consistently expressed Pax3/7 and MyoD, which are expressed in the dermomyotome and myotome, respectively, in gnathostomes (Fig. 4i,j; Supplementary Fig. S5)21,22. These two genes showed mutually exclusive expression in these mesodermal cells. Similar to their gnathostome cognates21,22, Pax3/7 expression was detected in the lateral part of the somites, whereas MyoD expression was restricted to the medioventral region (Fig. 4i,j). This shows that dermomyotome and myotomal primodia can be identified as derivatives of a hagfish somite, with corresponding gene expression patterns and in comparable mediolateral relative positions to those in gnathostomes (Fig. 4i,j)21,22. Furthermore, these are found at the same axial level as that from which the putative vertebral elements are formed (Fig. 1b). This strongly suggests that the Pax1/9-positive mesenchymal cells at the ventromedial aspect of the somite epithelia were developmentally equivalent to the sclerotome in gnathostomes (Fig. 4h)12,21,22.

At the 90th somite level in the late pharyngula (the level of the vertebra-like elements; see Fig. 3c,f), Pax1/9-positive cells were distributed in more extensive domains than those in the previous embryo, with clusters detectable at the ventrolateral aspect of the notochord (Fig. 3k,l). This led us to infer that these Pax1/9-positive cells may migrate further ventrally, surrounding the dorsal aorta, to differentiate into cartilaginous tissues on the ventral aspect of the notochord in later embryos (Fig. 4m,n). Given the metameric distribution patterns of the mesenchymal cells at the early pharyngula stage (Fig. 4d), the segmentally arranged sclerotomes, which originate from the ventromedial part of the somite (Fig. 4a–c,e,f), are likely to form the putative vertebral elements found in the adult (Fig. 1) via molecular mechanisms common to those in gnathostomes (Fig. 4m,n)12,21,22. These results suggest that despite the minute size and disrupted segmental pattern of the putative vertebral elements in hagfish (Fig. 1b–e), they can be homologous to vertebrae.


From the above observations, it seems plausible to assume that all the hagfish somites form sclerotomes and dermomyotomes at all axial levels in positions corresponding to those in gnathostomes, and that these somatic components arise via molecular developmental mechanisms similar to those in gnathostome embryos (Fig. 4). At the post-cloacal level, the sclerotome-derived cells are distributed in regions prospectively occupied by the vestigial cartilaginous nodules in adults. Thus, in the hagfish, the somite appears to respond to signals derived from the notochord to specify cells destined towards skeletal differentiation. In this connection, the potential skeletogenic property of hagfish paraxial mesoderm is suggested in the chordal cranium (the caudal portion of the neurocranium known as the parachordals; Supplementary Fig. S6)23,24. Therefore, it seems likely that the notochordally derived signal is primarily capable of inducing chondrification in the paraxial mesoderm of the hagfish, and that this mechanism is lost secondarily from the pre-cloacal part of the hagfish trunk.

Our observations are consistent at the anatomical and developmental level with the presence of axial cartilage in hagfish that is homologous to gnathostome vertebrae. This indicates that the presence of cartilaginous axial skeletal elements can be considered as a characteristic common to gnathostomes, lampreys and hagfishes. Hence, we can no longer exclude the hagfish from the Vertebrata simply due to the absence of a cartilaginous axial skeleton, which has now been disproved (Fig. 5).

Figure 5
Hypothetical scenario of vertebral evolution.

Although our developmental study suggested a sclerotomal origin for the ventral elements, the origin of the dorsal element, reminiscent of the neural arches and spine of gnathostomes, remains unclear (Fig. 1b). Nevertheless, given the anatomical similarity of this element to the gnathostome counterpart (Figs 1e and 2e,f), it is highly possible that they share the same developmental patterns, suggesting that the paraxial mesoderm of the entire common ancestor of gnathostomes, lampreys and hagfishes could form cartilaginous elements on both the dorsal and ventral sides of the notochord. Taking into account the recently reconstructed phylogenetic trees based on molecular data, which support a monophyly of hagfishes and lampreys25,26,27,28, it seems reasonable to hypothesize that the dorsal and ventral vertebral elements were obtained from a common ancestor of the entire vertebrates, and that subsequent secondary losses would have led to loss of the dorsal vertebral elements in the hagfishes on one hand, and to a ventral vertebral element in the lamprey on the other (Fig. 5). To test this hypothesis, further comparative developmental studies will be required to exclude the possibility of independent evolutionary occurrence of vertebral elements in the extant agnathan lineages, because it is still unclear whether the entire part of the vertebral elements of extant agnathans is derived from sclerotomal cells, as occurs in gnathostomes12. We still cannot rule out completely the possibility that some parts of the vertebral elements of lamprey and hagfish are derived from non-sclerotomal cells (for example, neural crest cells).

Finally, it remains to be determined whether the morphological pattern of hagfish axial skeletal elements represents a secondary loss of segmentation or an ancestral state before acquisition of the segmental pattern. However, the former scenario is more plausible under the assumption of the monophyly of the extant agnathans (Fig. 5). This scenario implies that the common ancestor of the extant gnathostomes and agnathans had the developmental mechanism to form a segmental axial skeleton consisting of two topographically different cartilaginous nodules, one located at the level of the myoseptum and the other at the intermediate level between the two myosepta, along the anterior–posterior axis, as seen in the vertebral elements of lampreys and gnathostomes (Figs 2 and and5).5). The origin of this developmental mechanism could date back to around 500 million years ago based on the divergence time between agnathans and gnathostomes26,27. Also problematic is the absence of cartilaginous tissue from the pre-cloacal trunk, as well as the absence of dorsal skeletal elements in most of the post-cloacal region (Fig. 1). Further study is required to elucidate whether such a loss can be simply explained by adaptation to a hagfish lifestyle or by the long evolutionary period2. The origin of the axial skeleton remains an issue in evolutionary developmental biology, and will require consideration of paleontological, embryological and molecular data12,13,25,26,27,28,29,30. We are hopeful that further advances in hagfish embryology will provide more insights into the evolution of the axial skeletal element in the early vertebrates, especially into the origin of its segmental pattern.


Sample collection and aquarium maintenance

Adult males and females of E. burgeri were collected using eel traps from a depth of 25–100 m in the Japan Sea off the Shimane and Yamaguchi prefectures during September and October. After the hagfish were transferred to a laboratory aquarium (at 16 °C) and sexed by manipulation, about 100 adult specimens were maintained in an aquarium tank lined with potentially favourable substrates, including fine-grain sand and oyster shells (1,000 l, 16–17 °C) (refs 31,32). The deposited eggs were individually incubated in a plastic container (4 cm×4 cm×5 cm) in the same aquarium tank32. Fertilized eggs were identified by visual inspection and were prepared for further analysis. All of the embryos are listed in Supplementary Table S1.

Molecular cloning

Total RNA was extracted from embryos using TRIZOL reagent (Invitrogen). Degenerated RT–PCR was performed to amplify fragments of Twist, Pax1/9, and MyoD messenger RNA using the primer sets listed in Supplementary Table S2. These fragments were isolated using the TOPO TA Cloning Kit Dual Promoter (Invitrogen) and sequenced using an ABI 3130XL automated sequencer (Applied Biosystems). The sequence data were submitted to the DDBJ database (AB594746, AB594747, AB594748). To identify the orthologous genes of the isolated fragments, comparable sequence data were surveyed using the NCBI protein database and a BLAST search, and multiple sequence alignments were generated using the CLUSTALW multiple alignment program33. Multiple sequence alignments were corrected by visual inspection. Based on these alignments, gene-specific residues and motifs were identified (Supplementary Figs S3, S4, S6).

Histology and in situ hybridization

Hagfish specimens were fixed in Serra's fixatives for 24–48 h, dehydrated using an ethanol series, placed in xylene, embedded in paraffin, and sliced at a thickness of 8 μm. The sliced sections were deparaffinized and stained with haematoxylin, eosin and Alcian blue34. In situ hybridization was carried out using a Ventana automated machine (Ventana Medical Systems). Detection of signals and counterstaining were performed using the BlueMap NBT/BCIP substrate kit and the nuclear fast red equivalent reagent, ISH RED (Ventana).

Author contributions

Y.O. and K.G.O. performed sample collections and maintained the aquarium tank. S.F. was involved in the molecular cloning of the Pax1/9, MyoD, and Twist genes and in the in situ hybridization. K.G.O. performed the molecular evolutionary analysis. K.G.O. and S.K. wrote the manuscript. All authors discussed the results and commented on the manuscript.

Additional information

Accession codes: The sequence data have been deposited to the DDBJ database under accession codes AB594746, AB594747 and AB594748.

How to cite this article: Ota, K. G. et al. Identification of vertebra-like elements and their possible differentiation from sclerotomes in the hagfish. Nat. Commun. 2:373 doi: 10.1038/ncomms1355 (2011).

Supplementary Material

Supplementary Information:

Supplementary Figures S1-S6 and Supplementary Table S1-S2.


We thank O. Kakitani of Shimane fisheries cooperative, K. Yamada of Kotoku Inc., and K. Kayano of Sekikatu Inc. for sample collection, S. Aota, F. Sugahara, H. Nagashima, S. Kuraku, N. Adachi, and H. Higashiyama for their valuable advice, and R. Ladher and D. Sipp for critical reading of this manuscript. This work was supported by Grants-in-Aid from the Ministry of Education, Culture, Sports, Science and Technology of Japan.


  • Janvier P. Early Vertebrates (Oxford University Press, 1996).
  • Kardong K. V. Vertebrates Comparative Anatomy, Function, Evolution (McGraw-Hill Higher Education, 2008).
  • Goodrich E. S. Vertebrata Craniata. (First Fascicle; Cyclostomes and Fishes) (Adam and Charles Black, 1909).
  • Goodrich E. S. Studies on the Structure and Development of Vertebrates (McMillan, 1930).
  • Kent G. C. & Carr R. K. Comparative Anatomy of the Vertebrates (McGraw-Hill, 2001).
  • Shimeld S. M. & Holland P. W. H. Vertebrate innovations. Proc. Natl Acad. Sci. USA 97, 4449–4452 (2000). [PMC free article] [PubMed]
  • Scaal M. & Wiegreffe C. Somite compartments in anamniotes. Anat. Embryol. 211 (Suppl 1), 9–19 (2006). [PubMed]
  • Miles R. S. Remarks on the vertebral column and caudal fin of acanthodian fishes. Lethaia 3, 343–362 (1970).
  • Dennis K. & Miles R. S. A pachyosteomorph arthrodire from Gogo, Western Australia. Zool. J. Linn. Soc. 73, 213–258 (1981).
  • Coates M. I. & Sequeira S. E. K. A new stethacanthid chondrichthyan from the lower Carboniferous of Bearsden, Scotand. J. Vert. Paleontol. 21, 438–459 (2001).
  • Tretjakoff D. Die Wirbeläule des Neunauges. Anat. Anz. 61, 387–396 (1926).
  • Christ B., Huang R. & Scaal M. Formation and differentiation of the avian sclerotome. Anat. Embryol. 208, 333–350 (2004). [PubMed]
  • Zhang G., Miyamoto M. M. & Cohn M. J. Lamprey type II collagen and Sox9 reveal an ancient origin of the vertebrate collagenous skeleton. Proc. Natl Acad. Sci. USA 103, 3180–3185 (2006). [PMC free article] [PubMed]
  • Müller J. Vergleichende Anatomie der Myxinoiden (Akademie der Wissenschaften, 1834).
  • Parker K. W. On the skeleton of the marsipobranch fishes. Part I. The Myxinoids (Myxine, and Bdellostoma). Philos. Trans. R. Soc. Lond. 174, 373–409 (1883).
  • Cole F. J. A monograph on the general morphology of myxinoid fishes, based on a study of Myxine. Part 1. The anatomy of the skeleton. Trans. R. Soc. Edinb. 41, 749–788 (1905).
  • Marinelli W. & Strenger A. Vergleichende Anatomie und Morphologie der Wirbeltiere. 2. Myxine Glutinosa (L.) (Franz Deuticke, 1956).
  • Ayers H. & Jackson C. M. Morphology of the myxinoidei. I. Skeleton and musculature. J. Morphol. 17, 185–226 (1900).
  • Hall B. K. Bones and Cartilage: Development and Evolutionary Skeletal Biology (Elsevier, 2005).
  • Ota K. G., Kuraku S. & Kuratani S. Hagfish embryology with reference to the evolution of the neural crest. Nature 446, 672–675 (2007). [PubMed]
  • Goulding M. D., Chalepakis G., Deutsch U., Erselius J. R. & Gruss P. Pax-3, a novel murine DNA binding protein expressed during early neurogenesis. EMBO. J. 10, 1135–1147 (1991). [PMC free article] [PubMed]
  • Rudnicki M. A. et al. . MyoD or Myf-5 is required for the formation of skeletal muscle. Cell. 75, 1351–1359 (1993). [PubMed]
  • Couly G. F., Coltey P. M. & Le Douarin N. M. The triple origin of skull in higher vertebrates: a study in quail-chick chimeras. Development 117, 409–429 (1993). [PubMed]
  • Kuratani S. & Ota K. G. Primitive versus derived traits in the developmental program of the vertebrate head: views from cyclostome developmental studies. J. Exp. Zool. B Mol. Dev. Evol. 310B, 294–314 (2008). [PubMed]
  • Mallatt J. & Sullivan J. 28S and 18S rDNA sequences support the monophyly of lampreys and hagfishes. Mol. Biol. Evol. 15, 1706–1718 (1998). [PubMed]
  • Kuraku S., Hoshiyama D., Katoh K., Suga H. & Miyata T. Monophyly of lampreys and hagfishes supported by nuclear DNA-coded genes. J. Mol. Evol. 49, 729–735 (1999). [PubMed]
  • Takezaki N., Figueroa F., Zaleska-Rutczynska Z. & Klein J. Molecular phylogeny of early vertebrates: monophyly of the agnathans as revealed by sequences of 35 genes. Mol. Biol. Evol. 20, 287–292 (2003). [PubMed]
  • Heimberg A. M., Cowper-Sal-lari R., Sémon M., Donoghue P. C. J. & Peterson K. J. microRNAs reveal the interrelationships of hagfish, lampreys, and gnathostomes and the nature of the ancestral vertebrate. Proc. Natl Acad. Sci. USA 107, 19379–19383 (2010). [PMC free article] [PubMed]
  • Janvier P. & Arsenault M. Palaeobiology: calcification of early vertebrate cartilage. Nature 417, 609 (2002). [PubMed]
  • Shu D. et al. . Head and backbone of the Early Cambrian vertebrate Haikouichthys. Nature 421, 526–529 (2003). [PubMed]
  • Ota K. G. & Kuratani S. The history of scientific endeavors towards understanding hagfish embryology. Zoolog. Sci. 23, 403–418 (2006). [PubMed]
  • Ota K. G. & Kuratani S. Developmental biology of hagfishes, with a report on newly obtained embryos of the Japanese inshore hagfish, Eptatretus burgeri. Zoolog. Sci. 25, 999–1011 (2008). [PubMed]
  • Thompson J. D., Higgins D. G. & Gibson T. J. CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res. 22, 4673–4680 (1994). [PMC free article] [PubMed]
  • Ota K. G. & Kuratani S. Expression pattern of two collagen type 2 α1 genes in the Japanese inshore hagfish (Eptatretus burgeri) with special reference to the evolution of cartilaginous tissue. J. Exp. Zool. B Mol. Dev. Evol. 314B, 157–165 (2010). [PubMed]
PubReader format: click here to try


Save items

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


  • Nucleotide
    Primary database (GenBank) nucleotide records reported in the current articles as well as Reference Sequences (RefSeqs) that include the articles as references.
  • Protein
    Protein translation features of primary database (GenBank) nucleotide records reported in the current articles as well as Reference Sequences (RefSeqs) that include the articles as references.
  • PubMed
    PubMed citations for these articles
  • Taxonomy
    Taxonomy records associated with the current articles through taxonomic information on related molecular database records (Nucleotide, Protein, Gene, SNP, Structure).
  • Taxonomy Tree
    Taxonomy Tree

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...