An axial Hox code controls tissue segmentation and body patterning in Nematostella vectensis

See allHide authors and affiliations

Science  28 Sep 2018:
Vol. 361, Issue 6409, pp. 1377-1380
DOI: 10.1126/science.aar8384

Hox code in segmentation and patterning

Hox genes encode conserved transcription factors that are best known for their role in governing anterior-posterior body patterning in diverse bilaterian animals. He et al. used a combination of CRISPR mutagenesis and short hairpin RNA–based gene knockdowns to interrogate Hox gene function in a cnidarian, the sea anemone Nematostella vectensis (see the Perspective by Arendt). Four homeobox-containing genes constitute a molecular network that coordinately controls the morphogenesis of radial endodermal segments and the patterning of tentacles. Thus, an ancient Hox code may have evolved to regulate both tissue segmentation and body patterning in the bilaterian-cnidarian common ancestor.

Science, this issue p. 1377; see also p. 1310


Hox genes encode conserved developmental transcription factors that govern anterior-posterior (A-P) pattering in diverse bilaterian animals, which display bilateral symmetry. Although Hox genes are also present within Cnidaria, these simple animals lack a definitive A-P axis, leaving it unclear how and when a functionally integrated Hox code arose during evolution. We used short hairpin RNA (shRNA)–mediated knockdown and CRISPR-Cas9 mutagenesis to demonstrate that a Hox-Gbx network controls radial segmentation of the larval endoderm during development of the sea anemone Nematostella vectensis. Loss of Hox-Gbx activity also elicits marked defects in tentacle patterning along the directive (orthogonal) axis of primary polyps. On the basis of our results, we propose that an axial Hox code may have controlled body patterning and tissue segmentation before the evolution of the bilaterian A-P axis.

Cnidarians (corals, jellyfish, and sea anemones) occupy a key position within the animal phylogeny, serving as an essential out-group for understanding the evolution of developmental processes throughout Bilateria (13). The characteristic polyp bauplan is a unifying feature within Cnidaria, typified by a tubelike body column with a single oral opening surrounded by tentacles (4, 5). Bilaterian body plans are more diverse, exhibiting increasing levels of complexity that correlate with the expansion of genomically clustered Hox genes (6, 7). In arthropods and chordates, for instance, Hox genes are expressed along the anterior-posterior (A-P) axis in staggered domains, setting up a molecular code that determines body segment identity and directs the formation of distinct appendages (8, 9). Among early-branching phyla, true Hox genes are found only within Cnidaria, indicating an ancient evolutionary origin predating the bilaterian-cnidarian split approximately 600 million years ago (1, 10, 11). Nevertheless, functional requirements for cnidarian Hox genes are unclear, leaving the possible ancestral role of this crucial developmental gene cluster poorly understood (12).

The sea anemone Nematostella vectensis is a model anthozoan cnidarian that has multiple Hox genes (13, 14). Under the control of bone morphogenetic protein (BMP) signaling, Anthox1a, Anthox6a, and Anthox8, together with Gastrulation brain homeobox (Gbx) (a Hox-linked subfamily gene), exhibit partially overlapping endodermal expression patterns in planula larvae (10, 14, 15). During this stage, the developing endoderm undergoes morphogenetic segmentation into eight sectors along the directive axis (Fig. 1, A to C), generating internal anatomical subdivisions that further correlate with the positioning of the first four tentacles in metamorphosed polyps (fig. S1). To determine whether the larval Hox-Gbx expression domains specifically define endodermal segment boundaries, we performed fluorescence in situ hybridization (FISH) on mid–planula stage larvae. Anthox1a, Anthox8, Anthox6a, and Gbx exhibited sharp expression territories, cell-autonomously defining segment boundaries (Fig. 1, D to K, and fig. S2). Further, each nested Hox expression domain formed in concert with the stepwise sequence of endodermal boundary morphogenesis (fig. S3), indicating a temporal correlation between Hox-Gbx expression and tissue segmentation. These observations suggest that a Hox-Gbx–dependent code controls the formation of the eight endodermal segments (segments s1 to s8) and specifies distinct positional identities along the directive axis (Fig. 2A).

Fig. 1 Nematostella Hox-Gbx expression patterns cell-autonomously correlate with endodermal segment boundaries in planula larvae.

(A) Nematostella life cycle. Embryos enter a free-swimming planula larva stage at approximately 48 hours postfertilization. (B and C) Wild-type mid-planula larvae stained to label F-actin. (B) Side view with oral pole (asterisk) facing upward, showing the focal plane for oral view images. (C) Oral view depicting the formation of eight endodermal segments. Scale bars, 50 μm. (D to G) Fluorescent in situ hybridizations, stained with Hoechst (DNA) to demonstrate the expression patterns of Nematostella Anthox1a, Anthox8, Anthox6a, and Gbx in planula larvae. Scale bar, 50 μm. (H to K) Magnified images from (D) to (G) illustrating the sharp expression boundaries for each gene, corresponding todistinct endodermal segment boundaries. Scale bar, 25 μm.

Fig. 2 Nematostella Hox-Gbx genes control tissue segmentation and tentacle patterning.

(A) Color-coded expression pattern of individual genes in wild-type larvae. (B to F) Oral views of wild-type versus Hox-Gbx KD planula-stage larvae. (B) Wild type [n = 24 larvae (image is representative of 24 of 26 wild-type larvae)]; (C) Anthox1a KD (n = 31/32); (D) Anthox8 KD (n = 26/28); (E) Anthox6a KD (n = 30/36); (F) Gbx KD (n = 26/33). Scale bar, 50 μm. hpf, hours postfertilization. (G to K) Oral views of wild-type versus Hox-Gbx KD polyps. (G) Wild type (n = 197 of 264 polyps); (H) Anthox1a KD polyps (shRNA1, n = 214/253; shRNA2, n = 165/192); (I) Anthox8 KD polyps (shRNA1, n = 144/162; shRNA2, n = 112/132); (J) Anthox6a KD polyps (shRNA1, n = 92/126; shRNA2, n = 123/162); (K) Gbx KD polyps (shRNA1, n = 84/181; shRNA2, n = 50/70; shRNA3, n = 24/38). Arrowheads indicate missing tentacles. Scale bar, 100 μm. dpf, days postfertilization.

To test the developmental requirements for the Hox-Gbx code during segmentation of the larval endoderm, we took advantage of the distinctive specificity of the cnidarian microRNA pathway (16) to establish a robust short hairpin RNA (shRNA)–based gene knockdown (KD) technique (figs. S4 to S6 and movies S1 and S2). KD of Anthox1a, Anthox8, Anthox6a, and Gbx elicited clear segmentation defects that directly correlated with the genes’ endogenous expression domains (Fig. 2, B to F). In Anthox1a KD larvae, the boundaries flanking segment s5 were abolished, resulting in the fusion of s4, s5, and s6 into a single large segment (s4-6) (Fig. 2C). In Anthox8 KD larvae, segment boundaries between s3 and s4 and between s6 and s7 were lost, resulting in two enlarged endodermal segments flanking s5 (s3-4 and s6-7) (Fig. 2D). In Anthox6a KD animals, the boundaries between s2 and s3 and between s7 and s8 were lost, generating fusion segments s2-3 and s7-8 flanking s1 (Fig. 2E). Lastly, in Gbx KD larvae, the boundaries flanking s1 were abolished, resulting in the fusion of s8, s1, and s2 into a single large segment (s8-2) (Fig. 2F). Similar phenotypes were observed for at least two independent shRNAs targeting each gene (Fig. 2 and materials and methods). Quantitative polymerase chain reaction (qPCR) validation further confirmed substantial reduction of the target mRNA level for each shRNA (fig. S7). These results demonstrate that a Hox-Gbx–dependent code drives morphogenetic tissue segmentation in the endoderm of developing Nematostella larvae.

To characterize later roles of Anthox1a, Anthox8, Anthox6a, and Gbx genes in body patterning, experimental larvae were reared through metamorphosis to the polyp stage. KD of each gene elicited marked and highly penetrant tentacle-patterning defects, each of which corresponded to the position of larval segmentation abnormalities (Fig. 2, G to K). In wild-type controls, four tentacles of equal size developed in stereotyped radial positions corresponding to the endodermal segments s2, s4, s6, and s8 (tentacles t2, t4, t6, and t8) (Fig. 2G and figs. S1C and S8A). In contrast, Anthox1a KD polyps exhibited a single large tentacle replacing t4 and t6, resulting in animals with three tentacles. The tip of this enlarged tentacle was frequently bifurcated, suggesting a possible fusion of t4 and t6 (t4-6) (Fig. 2H and fig. S8B). Anthox8 KD polyps specifically lost tentacles t4 and t6 (Fig. 2I and fig. S8C), resulting in animals with only two tentacles. Anthox6a KD polyps maintained four tentacles, although t2 and t8 were enlarged and partially fused with the adjacent tentacles t4 and t6, respectively (Fig. 2J, fig. S8D). Lastly, Gbx KD polyps consistently lost tentacles t2 and t8, resulting in two-tentacled animals with a mirror-image phenotype to Anthox8 KD animals (Fig. 2K and fig. S8E). Similar developmental requirements were not observed for the other Nematostella homeodomain-containing genes, Anthox6, Anthox7, and Cdx (fig. S9). Collectively, these experiments demonstrate key roles for Anthox1a, Anthox8, Anthox6a, and Gbx in controlling tentacle patterning, revealing a link between the process of endodermal segmentation and the specification of tentacle primordia.

Previous studies have demonstrated a role for BMP signaling in regulating both larval Hox expression and the development of mesenteries, digestive and reproductive organs that form during larval development at the boundaries between endodermal segments (fig. S6, A to C) (15). To explore additional roles for Hox genes in axial patterning, we developed a method to image Anthox1a, Anthox8, and Anthox6a KD animals by using selective plane illumination microscopy (SPIM) (17). We observed only six mesenteries in each KD condition (movies S1 and S3 to S5), consistent with the earlier defects in boundary formation. Taken together with earlier results, these findings confirm that Hox-dependent segmentation of the larval endoderm establishes key elements of the polyp bauplan, including the positioning of mesenteries and the patterning of tentacle primordia.

To validate the developmental function of the Nematostella Hox-Gbx code with a technically independent approach, we next mutated Anthox1a, Anthox6a, Anthox8, and Gbx by using CRISPR-Cas9–mediated genome editing (1820). After the injection of Cas9–guide RNA (gRNA) complexes into embryos (with either single or paired gRNAs), F0 indel mutations were recovered in all four loci, along with low frequencies of the expected tentacle phenotypes (Fig. 3 and fig. S10). Through subsequent controlled crosses, we obtained genetically null mutants for Anthox1a, Anthox8, and Anthox6a. In each case, these animals exhibited tentacle-patterning defects identical to those of the cognate shRNA KD animals (figs. S11 to S14). Putative Gbx mutants showed severe growth defects, and only heterozygous animals were recovered from F0 founder crosses, suggesting that Gbx has additional roles in later developmental processes. Taken together, these complementary CRISPR-Cas9– and shRNA-based methods illuminate clear developmental requirements for a Hox-Gbx network during cnidarian development (Figs. 2 and 3 and figs. S7 to S14).

Fig. 3 CRISPR-Cas9–mediated mutagenesis of Anthox1a confirms its function in tentacle patterning.

(A) gRNA strategy for the Anthox1a locus. F0 animals injected with either gRNA carried both frameshift and non-frameshift indel mutations (arrowheads indicate mutation positions). HD, homeodomain; PAM, protospacer adjacent motif; D, deletion of the specified number of bases. (B and C) In contrast to GFP gRNA–injected controls, putative Anthox1a F0 founders displayed the characteristic t4-6 tentacle fusion phenotype observed in KD experiments. Scale bars, 100 μm. (D) Quantification of t4-6 fusion phenotypes observed in uninjected controls [wild type (WT)], GFP gRNA controls (F0) (gGFP), Anthox1a gRNA–injected animals (F0) (g1, g2, and g1+g2), and two independent shRNA KD groups (sh1 and sh2).

In bilaterian systems, Hox-dependent patterns typically arise through extensive cross-regulatory interactions (2124). We therefore performed a series of double-KD experiments to determine whether Nematostella Hox genes exhibit genetic interactions during early development. Consistent with independent requirements for each locus, strictly additive phenotypes were observed in Anthox1a-Anthox8 (loss of t2 and t4), Anthox1a-Gbx (t4-6 fusion and loss of t2 and t8), and Anthox8-Gbx (no tentacles) double-KD animals (fig. S8, F to J). To further interrogate the molecular regulation of Hox-Gbx gene expression, we took advantage of a transgenic Anthox8>GFP reporter line that faithfully recapitulates endogenous Anthox8 expression in endodermal segments s4, s5, and s6 (Fig. 4, A to C). Notably, the enhancer region in this construct contains a previously identified Meis/Pbx-Hox binding site (25). Because Pbx is a key binding partner for bilaterian Hox genes that biochemically interacts with several Hox proteins in Nematostella (2528), we tested whether Pbx regulates Anthox8 expression. Larval Anthox8 transcription was undetectable after Pbx shRNA KD, which caused a severe and uniform loss of endodermal segmentation and subsequent metamorphic failure (Fig. 4D and fig. S8, K and L). To identify the putative Hox cofactor responsible for Pbx-dependent activation of Anthox8, we knocked down all Hox-Gbx genes and assayed Anthox8>GFP expression. Only shRNAs targeting Anthox8 substantially reduced green fluorescent protein (GFP) intensity (Fig. 4E and fig. S15). Further validated by FISH experiments (fig. S16), these results demonstrate Pbx-dependent autoregulation of Anthox8. In parallel experiments to explore cross-regulatory interactions, we also found that Anthox1a unidirectionally repressed Gbx expression in developing endodermal segment s5 (fig. S17). Combined, these findings hint at the possibility of similar regulatory mechanisms between cnidarian and bilaterian Hox networks.

Fig. 4 Nematostella Hox-Gbx genes comprise a spatial code that directs axial patterning.

(A) Design of the Anthox8>GFP reporter construct. (B and C) Although the Actin promoter alone drives ubiquitous GFP expression, addition of the Anthox8 upstream enhancer region restricts GFP expression to segments s4, s5, and s6 at the mid-planula stage (n = 10 of 10 larvae). (D) Pbx shRNA–injected transgenic animals lost all endodermal segmentation and failed to activate the Anthox8 reporter (n = 6 of 6 animals). (E) shRNAs targeting Anthox8 substantially decreased the GFP signal (n = 5 of 6 animals), restricting it to segment s5. Scale bars in (B) to (E), 50 μm. (F) Comparison between the functional Hox codes in a representative cnidarian (Nematostella) and a representative vertebrate (mouse). Although the Hox codes respond to different upstream signaling pathways along distinct axes [BMP in Nematostella and retinoic acid and fibroblast growth factor (RA and FGF) in mouse], similar downstream molecular programs drive segmental patterning. Data on potential homologies between Nematostella and bilaterian Hox genes are from previous publications (12, 14).

In summary, this work leverages both classical genetics and a robust gene KD methodology to demonstrate the existence of a functional Hox code in a developing cnidarian. Reminiscent of the sophisticated Hox networks that operate in arthropods and chordates, Nematostella Hox-Gbx genes encode axial identities, thus governing the patterning of secondary structures such as tentacles and mesenteries (Fig. 4F). Hox-Gbx–dependent endodermal segments may be established in a manner analogous to bilaterian posterior prevalence, whereby posteriorly expressed Hox genes generally override the effects of genes that are more anterior (2931). According to this logic, Anthox1a expression in Nematostella segment s5 would reflect a dominant pole of the anthozoan directive axis, with Anthox8, Anthox6a, and Gbx operating to define successive segment boundaries toward the opposite end (fig. S18). Although understanding the direct or indirect nature of these Hox-Gbx interactions will be an important area for future studies, our findings further demonstrate the existence of a Pbx-dependent functional network and provide initial evidence for both Hox auto- and cross-regulation (Fig. 4). Despite limited data regarding other cnidarian species, the phylogenetically basal position of Anthozoa (32) permits speculation that the Nematostella Hox code reflects a conserved gene regulatory module, one that could have been co-opted to direct A-P patterning in the ancient urbilaterian.

Supplementary Materials

Materials and Methods

Figs. S1 to S18

Tables S1 and S2

References (3346)

Movies S1 to S5

References and Notes

Acknowledgments: We thank R. Krumlauf (Stowers Institute), P. Cartwright (University of Kansas), and D. Lambert (University of Rochester) for suggestions and critical reading of the manuscript. We also thank M. Kirkman and K. Delventhal for genotyping assistance and the Stowers Institute Aquatics Core facility for animal husbandry. Funding: This study was supported by the Stowers Institute for Medical Research. Author contributions: S.H. and M.C.G. designed and analyzed the experiments. S.H. developed the shRNA approach and performed all RNA interference experiments. A.I. generated the Actin>GFP transgenic line and tested shRNA perdurance. S.H. generated the Anthox8>GFP transgenic line. S.H. and F.D.V. performed CRISPR-Cas9 genome editing. C.-Y.C. optimized the FISH protocol and performed time-course reverse transcription–qPCR analysis. S.H. and C.-Y.C. performed the FISH experiments. A.E.K., A.I., and S.H. performed SPIM imaging and SPIM data analysis. S.H. and M.C.G. wrote the manuscript. All authors discussed the experiments and read and approved the manuscript. Competing interests: None declared. Data and materials availability: Original data underlying this manuscript can be accessed from the Stowers Original Data Repository at All data needed to evaluate the conclusions in the paper are present in the paper or the supplementary materials.
View Abstract

Stay Connected to Science

Navigate This Article