Introduction

Hox genes are crucial regulators of body patterning and development in bilaterian animals, controlling the identity and position of body structures along the anterior-posterior axis1. These body patterning genes were initially discovered in Drosophila melanogaster fruit flies through the striking phenotypes they gave rise to when genes were mutated2. In these homeotic phenotypes, one part of the body along the anterior-posterior axis is transformed into the likeness of another part of the body, such as Antennapedia mutations that transform fly antennae into legs3. In many bilaterian animals, these genes are organised into clusters in the genome, with the order of the genes reflecting their spatial expression during embryonic development (called collinearity)2. The last common ancestor of protostomes and deuterostomes is thought to have had a cluster of at least seven Hox genes characterised by a common transcriptional orientation and by colinearity in the order of the genes and their expression domains along the anterior-posterior axis4,5,6.

Figure 1
figure 1

Cladogram of the Phylum Nematoda showing Clades relevant to this work. The clade naming convention was introduced by Blaxter et al. in 1998.

Full size image

Vertebrate Hox gene clusters have a distinctive structural and transcriptional organisation. This is true of all of the Hox cluster copies present following the whole genome duplications in vertebrates7,8. Despite the generally larger size of vertebrate genomes, vertebrate Hox genes are very densely packed and Hox cluster span is much smaller than in many invertebrates. The clusters have low repeat densities, and while protein-coding genes unrelated to the Hox family are excluded, small RNA genes which regulate cluster expression can be interspersed between Hox genes. Indeed, clustering is strongly associated with the mechanisms of cross- and co-regulation of Hox gene expression9,10.

Clustering and collinearity are not always conserved. Numerous animal taxa have managed to escape the constraint of having a tightly regulated, compact and complete Hox cluster. In several bilaterian phyla, Hox genes have undergone significant evolutionary changes including gene duplication, loss, and gain as well as changes in cluster organisation and expression patterns11. Some taxa have Hox clusters that are less tightly organised but still intact when compared to vertebrate counterparts, including the cephalochordates (Branchiostoma)12, sea urchins13, and insects (Apis14,15,16, Anopheles17,18 and Tribolium castaneum19,20). The Hox cluster in Drosophila melanogaster is split into two subclusters21. Splitting has occurred repeatedly and independently in other drosophilids, at different positions along the cluster16,22. Disaggregation of the Hox cluster can be observed in tunicate chordates Ciona intestinalis (solitary tunicate) and Oikopleura dioica (appendicularian) where Hox loci are present but not tightly linked in the genome23,24.

By comparing the complements and developmental roles of Hox genes in different groups, it is possible to propose models of how the evolution of these genes has influenced the morphological diversity within various metazoan phyla. Loss of Hox genes can lead to body plan simplification while gene duplication can result in the acquisition of novel developmental processes25,26. Indeed the body plan of an animal can be a consequence of Hox gene loss as thought for the tardigrade Hypsibius exemplaris26,27. Hox gene loss is a common feature in Tardigrada, where a reduced complement of seven HOX genes including a local duplication of Hox9-13 was found28.

Nematomorphs are the closest relatives to Nematoda and are all parasitoids in arthropods. Both groups have relatively simple and vermiform body plan. A previous survey of Hox gene content using both PCR surveys and transcriptome analysis in Paragordius revealed a Hox gene complement of five ancestral ortholog groups. The presence of Hox2 (Pb) in Paragordius yet lost in the nematode counterparts revealed loss in the common ancestor of nematodes28,29. Recently published high-quality genomes from Nematomorpha (Acutogordius australiensis, Nectonema munidae, and Gordionus montsenyensis) show a relatively low number of protein coding genes (11,114, 8,717, 10,819 respectively) and lack a high proportion of universal single-copy metazoan orthologs30,31.

Nematoda comprises ecologically and functionally diverse organisms32,33,34,35 that have provided valuable insights into the genetics of development in general. The model nematode species C. elegans was considered a representative of the developmental biology and genomics of all Nematoda for many decades, but nematodes are highly diverse with over 28,000 described species36,37 and potentially more than 1 million species in the phylum38,39. They are found in a wide range of habitats including soil, freshwater and marine environments. Phylum Nematoda is split into three classes32: Dorylaimia (also known as Clade I), Enoplia (Clade II), and Chromadoria (Clade C). Within the primarily aquatic Chromadoria, the largely terrestrial order Rhabditida is nested and includes three speciose suborders (Spirurina or Clade III, Tylenchina or Clade IV, and Rhabditina or Clade V). C. elegans is a member of Rhabditina (Clade V) in Fig. 1. It has now become apparent that nematodes show substantial divergence in regard to cellular pattern formation in early development40,41,42, post-embryonic development43,44,45 and their general gene content46,47,48,49.

The C. elegans Hox gene set has been shown to differ substantially from counterparts in other animal phyla. The six Hox genes include representatives of only four core Hox orthology groups (ceh-13 from the HOX1 group, lin-39 from HOX4, mab-5 from HOX6-8 and egl-5, php-3 and nob-1 from HOX9). The genes are linked but largely unclustered: they are distributed over an approximately>4 Mb span of C. elegans chromosome III, with up to 44 unrelated genes interspersed25,50. While there is some residual collinearity, the HOX1 and HOX4 orthologues of ceh-13 and lin-39 are inverted compared to the relative order of HOX6 and HOX9. In recent years it has become evident that many features of nematode embryonic development (including timings of cell division, cell lineage, and gene expression) are highly variable across the phylum40,51, and, like nematode genomes, evolve rapidly29,48,52,53. While Hox gene evolution has been well studied in C. elegans and a taxonomically biased sample of parasitic species from both Clade I and III29, our knowledge of Hox gene repertoire, Hox Cluster organisation and linkage in other nematode lineages, representing most of genomic diversity of the phylum, is limited. Determination of the full complement of Hox genes in recently available nematode genomes across the Nematoda is fundamental for understanding the link between the genomic and morphological evolution of this phylum. Here we have analysed the Hox gene complements of a wide range of newly-sequenced, high-quality genomes from diverse nematodes to understand the evolution of this key set of body patterning genes.

Results

Hox gene complements in Nematoda

Figure 2
figure 2

Nematoda and Nematomorpha Hox gene complements. Genes belonging to the same orthology group (HOX1-HOX9) are coloured similarly. X indicates the absence of the orthologue. x2 indicated the duplication event. The cladogram on the left shows the phylogenetic relationships of the analysed nematode species based on37,54.

Full size image

HOX gene evolution in nematodes is largely characterised by consecutive losses

Hox genes play critical roles early in the embryonic development in determining cell fates so that the correct body parts are formed in the right places from nematodes to vertebrates. While it is possible to develop interim catalogues of Hox gene repertoires by directed PCR, transcriptomics or draft genome sequencing, to validate presence-absence patterns and explore synteny relationships high quality genome sequences are required. Across the analysed nematode species we identified loci attributable to six Hox orthology groups; HOX1 (arthropod lab, C. elegans ceh-13), HOX3 (zen), HOX4 (Dfd, lin-39), HOX6-8 (ftz, mab-5), HOX6-8 (Antp), HOX9-13 (AbdB, egl-5/php-3/nob-1). We did not find loci that were classifiable as HOX2 (pb), HOX5 (scr) and the Ubx/AbdA subtypes of HOX6-8 loci in any of the nematode genomes analysed. The maximum number of Hox loci in a single species was seven (in members of Spirurina / Clade III), and the minimum found number of genes was four (Oncholaimus oxyuris) (Fig. 2). Loci attributed to HOX orthogroups HOX9-13 showed variable presence across the nematode phylogeny.

Our pipeline reliably recovers C. elegans HOX loci

The pipeline is dependent on a reference database consisting of highly curated HMM profiles for HOX homeodomains. These profiles are obtained through protein alignments from closely related species, where C. elegans and Amphioxus Hox homeodomains are employed as the in-group and outgroup, respectively. This choice holds particular significance due to the heightened sensitivity of sequence similarity search techniques to the reference database’s protein quality. Furthermore, the rapid evolutionary nature of nematode homeodomains adds to the rationale for this approach29. We applied the analysis pipeline (Fig. 5) in order to test the ability to recover all the C elegans expected Hox loci. All the six expected loci were recovered (Fig. 2) in our tests.

Hox6-8 (Antp) loci variation in Rhabditomorpha (including an ingroup Strongylomorpha=Strongyoidea) crown group nematodes

Nematodes belonging to the infraorder Rhabditomorpha form a diverse group exhibiting various lifestyles, ranging from being free-living bacterial feeding or predatory to obligate parasites in vertebrates and invertebrates. While C. elegans misses Hox6-8 (Antp) loci, we found it to be present in crown lineage except for Teladorsagia circumcincta and Nippostrongylus brasiliensis. In the members of this lineage included in our analysis, all C. elegans Hox genes are present except for nob-1 (Fig. 2).

nob-1 is a recent duplication within Caenorhabditis

We expanded the scope of the study to investigate the Hox gene complements in diverse Caenorhabditis nematodes beyond C. elegans (Fig. 3). The Hox gene repertoire resembled that of C. elegans, with an exception: the detection of a recent nob-1 gene duplication29 in species belonging to the “elegans” and “japonica” subgroups including C. briggsae, C. brenneri, C. nauraguensis, and C. japonica. The nob-1 gene was absent in the earlier branching Caenorhabditis members of the “drosophilae” subgroup (Fig. 2). Among the three posterior Hox gene class (egl-5, php-3, and nob-1) in C. elegans, the same genes were present in close relatives. However, other Caenorhabditis species have two HOX9-13 loci corresponding to egl-5 and php-3, while no other nematode displayed more than two members of HOX9-13. The inclusion of Diploscapter as a very close outgroup provides strong support for the idea that nob-1 is unique to a lineage within the genus Caenorhabditis. (Fig. 2).

Hox3 gene loss in Tylenchina and Spirurina

During our investigation, we noted that all the examined members within the Tylenchina clade (Clade IV) lacked HOX3 loci. Within the Spirurina clade (Clade III), specific instances of Hox3 loss were identified in Syphacia muris, Enterobius vermicularis, and Dracunculus medinensis, which are illustrated in Fig. 2. We also recorded a single loss of ant-1 in Tylenchina, except in the most early branchind Steinernema carpocapsae, as depicted in Fig. 2. However, despite the presence of ant-1 loci in all Spirurina species included in our study, Syphacia muris and Anisakis simplex were exceptions where this locus was found to be absent. Furthermore, we observed separate instances of Hox9-13 (egl-5) loss among the members of the Tylenchoidea crown group that includes root-knot and cyst nematodes parasites of plants.

Hox gene complements in early-branching chromadoreans

Previous investigations into the Hox gene complement among early-branching members in Chromadorea were limited due to insufficient genomic resources, impeding a comprehensive understanding of Hox gene evolutionary patterns28. In this study, we expanded our analysis by utilizing newly available draft genomes in early-branching Chromadorea. In certain species, we observed Hox3 loss (Terschellingia longicaudata, Daptonema setosum, and Axonolaimus paraspinosus). Furthermore, the Antp-like locus was absent in Plectus sambesii, Sabatieria pulchra, and Eleutherolaimus stenosoma. Additionally, we identified independent cases of Hox9-13 (egl-5) loss in Leptolaimus papilliger, Daptonema setosum, and Eleutherolaimus stenosoma. Other members of this Clade collectively possessed a total of seven Hox genes, as depicted in Fig. 2.

An independent loss of ant-1 locus in Enoplida, and a single HOX9-13 (php-3) in Enoplean nematodes

The Clades I and II (Enoplea) class encompasses the earliest-branching lineages within the Nematoda. Previous investigations into the Hox gene composition of Enoplea were limited due to the lack of genomic data for most of the non-parasitic enoplean lineages. In this study, we delve deeper into the analysis, utilizing nine genomes from Enoplia (Clade II) and nine genomes from Dorylaimia (or Clade I). Initially, our findings revealed the presence of a collective total of six Hox gene loci in most Enoplean members (depicted in Fig. 2). Among these, a solitary HOX9-13 orthologue was identified, which was classified as php-3. However, Oncholaimus oxyuris, a species with only four Hox orthologue groups, lacked any HOX9-13 loci, specifically the one (php-3) that would be anticipated. It’s important to note that this assembly originated from a single specimen amplified for long-read sequencing, and the absence might stem from residual assembly issues rather than actual loss (as shown in Fig. 2). The egl-5 locus was absent in both Enoplea (Clade II) and Dorylaimia (Clade I) species. Furthermore, we observed independent losses of the Antp-like locus within sampled Clade II (Enoplida) members among Enoplia (Fig. 2).

Nematomorpha have atleast six Hox loci

Nematomorpha, the sister phylum of Nematoda55, represent a group of parasitoid worms commonly known as horsehair worms or Gordian worms. Earlier investigations into the Hox gene complements within Nematomorpha yielded inconclusive results, primarily attributed to inadequate sequence data that hindered the assignment of numerous Hox genes to specific orthologue groups29. In our investigation of their Hox gene complements from four recently available genomes, we found Hox loci corresponding to Hox1 (lab), Hox2 (Pb), Hox4 (Dfd), Hox6-8 (Antp), Hox6-8 (Abd-A) and Hox9-13 (Abd-B). We did not find Hox loci corresponding to Hox3 (zen) and Hox6-8 (Ftz) (Fig. 2). We also found a duplication of Hox4 (Dfd) in G. aquaticus and G. montsenyensis (Fig. 4).

Figure 3
figure 3

nob-1 Hox gene complement across Caenorhabdtis lineage.

Full size image
Figure 4
figure 4

Hox gene clusters for eight selected species of Nematoda and for all three studied species of Nematomorpha showing synteny relationships of Hox cluster loci. Identified Hox genes and their respective transcriptional orientations are indicated by arrow boxes which are coloured by orthology group. Dashed horizontal lines indicate cluster breakage in scaffolded genomes. Lineage-specific duplications include paralog groups Hox6-8 and Hox9-13. Cross indicate a break that resulted from gaps in un-scaffolded draft genome. The cladogram on the left summarises the relationships between species.

Full size image

Hox cluster organization in the Nematoda

Clustering and collinearity are prominent features observed in the evolutionary path of HOX genes, and this pattern has proven to be consistent across a wide spectrum of species. In contrast to the loose clustering of Hox genes in three loci pairs on chromosome III in C. elegans, the analysis of nematode genomes in the present study reveals more condensed Hox gene clusters. These clusters exhibit variations in the number of genes, the sequence of gene arrangement, and the orientation of transcription. Furthermore, we observe interruptions within these Hox gene clusters by the presence of non-Hox genes.

Unique characteristics of Hox gene organization in Bursaphelenchus

We conducted a comprehensive investigation into clustering, gene order, and orientation within Hox gene sets across the phylum. Our analysis revealed potential HOX clusters in several species. Notably, in the two Bursaphelenchus species, four out of the five Hox loci were found to be closely linked, with a separation of less than 40 kb in B. xylophilus, while the HOX9-13 locus, php-3, was separated by 302 or 660 kb. Within this core set of four loci, the genes maintained the same relative order, but in B. okinawensis, the core cluster exhibited an inversion with respect to the position of php-3 compared to its arrangement in B. xylophilus. The order of genes in the core did not align with the expected HOX1-HOX6-8 ordering (Fig. 4).

Some members of Spirurina (Clade III) display Hox clusters containing a greater complement of Hox genes.

Both Brugia malayi and Onchocerca volvulus possess expanded Hox clusters containing seven linked Hox genes, of which six are closely clustered, and both species have a greater number of Hox genes compared to most other members of the Nematoda group. This core cluster comprises six members (ceh-13, hox-3, lin-39, mab-5, ant-1, egl-5), while the HOX9-13 php-3 locus is situated at a distance of 0.9 Mb (B. malayi) or 2.5 Mb (O. volvulus). Although the gene order within the core group remains consistent between the two species, it does not adhere to the typical HOX1-HOX6-8 structure, and there are no differences in the transcriptional orientation of the Hox loci between these species (Fig. 4).

Hox cluster organization in basal Chromadorea

In Plectus sambesii and Sabatiera pulchra (Chromadoria) clusters containing five Hox genes, the homologues of ceh-13, hox-3, lin-39, mab-5 and ant-1, were identified. The cluster in P. sambesii was particularly compressed (less than 40 kb) (Fig. 4). The clusters differed by an inversion of the mab-5 and egl-5 pair, but in P. sambesii the order of the loci corresponds to the model structure, albeit the transcriptional order does not correspond (Fig. 4). The HOX9-13 locus php-3 was located on a different genomic scaffold in Plectus sambesii (Fig. 4).

Hox loci in Trichuris are linked together in CiS and adhere to the model order

In the instance of Trichuris suis, a member of the Dorylaimia (Clade I) lineage, a cluster covering about 100 kb was detected, possessing six Hox loci that were interconnected and transcribed from distinct DNA strands (Fig. 4).

Hox cluster arrangement in Nematomorpha

The Hox genes in Gordionus montsenyensis exist as a fragmented cluster (Fig. 4); anterior paralogs (Hox1(lab) and Hox2(Pb)) are seperated from the three central group paralogs (Hox4 (Dfd), Hox6-8(Antp) and Hox6-8(Abd-A)) by approximately 5.5 Mb (Fig. 4). The posterior Hox9-13 (Abd-B) is further way on the same scaffold by more than 4 Mb (Fig. 4). The order of the genes in the core does reflects the expected HOX1-HOX6-8 ordering except that HOX6-8/AbdA and Hox6-8/Antp genes are inverted with respect to the order found in ancestral cluster. Hox genes in Gordius aquaticus are linked on scaffold 5 (Fig. 4). Acutogordius australiensis has members of Hox genes present in 3 seperate genomic scaffolds. Scaffold_2 contains Hox4 (Dfd), scaffold_3 contains 4 Hox loci, scaffold_4 contains Hox6-8 (Abd-A).

Hox clusters in nematodes are interrupted by non-Hox genes

Although vertebrate genomes are typically larger, the arrangement of vertebrate Hox genes is notably compact, with Hox clusters spanning a smaller region compared to many invertebrates. These clusters exhibit low repeat densities and exclude protein-coding genes unrelated to the Hox family. One of the few gene types that are found inside the clusters are small RNA genes (microRNAs) that have roles in Hox gene regulation. This suggests a strong constraint on the Hox cluster in vertebrates, likely because of the intimate co-regulation of the Hox loci. We explored whether similar constraint acts on nematode Hox clusters. We identified non-Hox genes between Hox loci in nematode species analysed (Table 1). Where Hox loci were tightly clustered, the number of genes was reduced, but even in the smallest clusters (B. xylophilus, P. sambesii, T. muris), non-Hox loci were common (Table 1) and (Fig. 4).

Hox clusters in nematodes are interrupted by miRNAs

MicroRNAs play pivotal roles in post-transcriptional regulation, orchestrating mRNA degradation or translational repression. We found miRNA presence within diverse nematode HOX clusters: dpu-miR-993 was identified in both B. malayi and O. volvulus; meanwhile, T. suis, P. sambesii, and S. pulchra exhibited dpu-miR-993. Notably, no microRNAs were detected in the HOX clusters of B. xylophilus and B. okinawaensis. Additionally, our analysis of C. elegans revealed the presence of cel-miR-231-3p, consistent with prior findings of the cel-miR-231 microRNA family’s localization between the lin-39 and ceh-13 Hox loci in C. elegans56.

Table 1 Hox cluster features of Nematoda and Nematomorpha from available high-quality genomes.
Full size table

Discussion

Hox genes are key players in the structuring of animal body plans during development. The developmental role of Hox genes has been well studied in C. elegans and the satellite model nematode species Pristionchus pacificus25,29,57,58. Our knowledge of the Hox gene repertoire, Hox Cluster organisation, and linkage in most nematode lineages was limited until now, with data restricted to some parasitic species scattered across Nematoda29. Most published nematode genomic resources came from Clades I, III, IV and V often missing a higher representation of free-living nematodes in Clade II and early branching Chromadorea37,59,60. Determination of the full complement of Hox genes in recently available nematode genomes across the phylum is fundamental to understanding the link between the genomic and morphological evolution of these animals. We have re-visited the Hox gene complements of Nematoda not only to understand the evolution of this key set of body pattern genes using newly available high-quality genomes but also to classify the loci into their orthologue groups as well as to look for linkage between loci, non-Hox genes and miRNAs interspersing HOX genes in a broader phylogenetic context.

Nematodes are renowned for their simple and highly conserved body plan. However, at the sequence level, they have been shown to possess rapid homeodomain sequence and genome evolution which might be a general feature of Nematoda Hox genes29. We observe that it is difficult to fully prove the absence of a gene and that the missing Hox genes in some genomes could indeed be truly missing from these genomes or they could be a result of extreme homeobox sequence divergence such that the sequence similarity search methods used in this study could not detect extreme sequence variability. In C. elegans, a reduced Hox gene complement (6 genes in 4 orthology groups) has been reported29 and we confirmed this reduction in this study (Fig. 2), illustrating the robustness of our methods. In Syphacia muris, Enterobius vermicularis as well as the common ancestor of Clade IV-V nematodes, an independent loss of Hox3 had been reported28 and we confirmed this loss with our approach as well (Fig. 3). This further supports the fidelity of the analysis methods employed here for detecting gene presence and absence in assembled genomes.

We have expanded the availability of Hox gene complements for a fairly large number of nematode species that were previously missing due to a lack of genomic resources. The Hox complement across the phylum Nematoda is shaped by key gene losses. We found a recent duplication of nob-1 in the model genus. This case illustrates the potential for new in-paralogues to evolve in certain lineages. In C. elegans, loss of nob-1 and php-3 functions leads to severe abnormalities during embryonic development, affecting both the arrangement of posterior structures and the movement of the posterior hypodermis. Additionally, this disruption results in changes from posterior to anterior cell identities and ultimately leads to non-viability25. Further examinations are required to establish if these roles persist across different Caenorhabditis species that possess nob-1.

We found presence of Hox3 in Clades I-III of Nematoda and absence in both Clade IV and V of Nematoda and Nematomorph genomes analyzed here (Fig. 2). The pattern of presence and absence in Nematoda can be explained by an independent derivation in the respective Clades. Within the insect counterparts, orthologs of Hox3 (zerknüllt, zen) have undergone repeated functional alterations, resulting in the relinquishment of their conventional Hox role and subsequent reassignment of their functional domain from embryonic to extraembryonic tissues. In fact, all known zen genes have been found to participate in the development of extraembryonic membranes (EEMs). These EEMs serve to shield embryos from external environmental challenges, and their formation has enabled insects to deposit eggs in diverse habitats, ultimately facilitating their successful colonization of terrestrial environments61. Functional investigations are necessary to clarify the function of Hox3 in nematodes that possess this gene.

Previous analysis using transcriptome datasets and PCR surveys identified loss of ant-1 locus in the Enoplea species Enoplus brevis, the chromadorea Plectus sambesi, S. muris, the ancestor of Tylenchomorpha, the diplogasteromorph P. pacificus, and the ancestor of Caenorhabditis28. We now certify this pattern of presence-absence using currently available high-quality genomes from these species and further extend the loss of ant-1 locus in the pinworm E. vermicularis (Clade III). In B. malayi, it has been demonstrated that the antennapedia gene’s positioning occurs after the Bm-egl-5 Hox gene. Moreover, their homeodomain exons are alternatively joined to the same 5’ exon through cis splicing and this arrangement might signify an intermediate stage in the potential loss of Hox genes due to redundancy in other nematode species29. Furthermore, ant-1 was identified in C. monodelphis in contrast to prior PCR survey-based studies that reported its absence28. The presence of ant-1 exhibited variability in Clade V nematodes (Fig. 2). Nevertheless, additional investigation is required to substantiate this distribution pattern.

We found Hox2/pb, Hox5/scr and Hox6-8/Ubx/AbdA missing in Nematoda (Fig. 2). The missing genes in newly sequenced nematode genomes further validate the previous findings from PCR-based studies28,29. The finding of Hox2/pb and Hox6-8/AbdA in Nematomorpha (Fig. 2), which are closest relatives to Nematoda, indicates that these Hox loci were lost secondarily in the lineage leading to Nematoda. Recent work suggests much more gene loss has occurred in metazoan genome evolution than previously thought30,62,63, and Hox gene loss has been observed in several other lineages28. The absence of Hox5/Scr in both Nematoda and Nematomorpha not only suggests an inheritance characteristic from the common ancestor of all extant Nematoidea (a clade uniting Nematoda and Nematomorpha) but also a potential example of evolutionary conservation.

Why some Hox genes were lost early on in the evolution of the phylum and others were lost or duplicated later on remains unclear. We found that some nematodes possess Hox3 and antennapedia-like Hox gene ant-1 which belong to orthology groups present in other metazoan animals but are absent in C. elegans and some of its relatives (Fig. 2). The loss may be due to a more specialised function that is only needed in certain nematode species or developmental contexts. For instance, Hox3 or Hox6-8/Antp may be involved in the development of specialised structures that are unique to those non-model nematode species and if these structures are lost during the course of evolution, the respective Hox genes may also become dispensable and eventually lost. In contrast, other Hox genes such as ceh-13 which is critical for embryogenesis and influences cell fate specification beyond the anterior body part25,64 and php-3 which regulates specification of the tail65 might have more essential functions that are needed for proper development of nematodes. These genes are less likely to be lost because their loss would have a more significant impact on the nematode’s development and survival. Therefore, the Hox complement in Nematoda is shaped by gene loss and might have played an important role in the evolution of their simplified body plan.

While the C. elegans Hox cluster is dispersed, we found intact and not dispersed Hox clusters in several genomes of non model nematode species (Fig. 4). We found rearrangement, translocation, and disruption by inversion with respect to the transcriptional orientation of the Hox genes in the clusters of analysed nematode genomes (Fig. 4). This pattern of mixed arrangement reflects a highly dynamic and complex history of nematode Hox cluster evolution and organisation similar to a situation observed in Drosophila melanogaster where the Hox cluster is broken into parts (the Antennapedia (ANTC) and bithorax (BXC) complexes). It exhibits aberrations in the transcriptional orientation of some genes as well as both interspersed genes of independent origin and Hox-derived genes that have evolved novel developmental functions11. Additional Hox cluster rearrangements have also been reported in other Drosophila species16,22,66,67 as well as in the silk moth Bombyx mori68. The rearrangement of Hox genes in the Cluster of C. elegans has also been reported before29. This further suggests that the physical orientation and rearrangement of the Hox genes in the clusters are not crucial for normal nematode development and body plan conservation. The dynamics in the organisation of the nematode Hox clusters could be further evidence that strong evolutionary constraints are rather acting on the highly conserved functional domains of the nematode Hox genes to maintain the conserved body plan. Hox gene rearrangements may drive evolutionary changes causing alterations in the pattern of Hox gene expression during development. They could facilitate the evolution of new morphological structures that are characteristic of different species thereby contributing to the process of nematode speciation. As the complete genome sequences of more nematode species become available, the molecular phylogenetics of Hox cluster evolution will become clearer with an emphasis on understanding the evolution of the regulatory modules that partition Hox expression domains.

We identified non Hox genes between Hox loci (Table 1) which may act as Hox cluster co-regulator candidates. Previous works identified T-box proteins as being particularly amplified in Rhabditid nematodes69,70. T-box genes encompass a group of transcription factors that possess a strongly conserved DNA-binding domain called the T-domain, and they are associated with the control of various developmental processes in multicellular organisms. In our work, we found TBX10 in B. malayi and O. volvulus parasitic nematodes (Fig. 4). Tbx10 has been shown to have retained ancestral roles in specific developmental processes, such as ventral somite specification in the tail region in mouse embryogenesis71. Additionally, the presence of “foreign” genes in Hox clusters of nematodes are indicators that the distances between Hox genes are quickly evolving72.

The identification of potentially unified Hox clusters in species occupying phylogenetic positions close to or in the early branching Clades is an indicator that perhaps the common ancestor of all nematodes had a single Hox cluster of at least six Hox genes as observed in Trichuridae (Trichuris suis) and Plectus sambesii and that these clusters have undergone remodification in different nematode lineages as a result of gene loss and Hox paralog specific duplication. The Nematoda Hox cluster organisation is highly dynamic. If present at all, the arrangement of genes in a cluster can be interrupted, shuffled, or even completely disintegrated. Further knowledge on topological islands in the analysed nematode Hox gene clusters herein awaits an investigation.

MicroRNAs are known to play crucial roles in post-transcriptional regulation of gene expression, often by targeting mRNAs for degradation or translational repression73. The presence of asu-miR-993-3p in B. malayi and O. volvulus, dpu-miR-993 in P. sambesii, S. pulchra and T. suis Hox clusters underscore their conservation within closely related nematode species. This conservation of microRNAs implies functional importance. These findings are consistent with previous studies highlighting microRNA presence in C. elegans74,75. The microRNA family cel-miR-231 has been previously reported in ceh-13/lin-39 Hox locus56. While these microRNAs have been identified within nematode Hox clusters, their specific regulatory targets and functional implications within non-model nematode development remain to be elucidated.

In contrast to the nematode body plan’s conservation, essential bilaterian regulators like Hox genes have undergone significant evolutionary flexibility. The loss of some of these genes in the lineage leading to the Ur-Nematode might have played a role in roundworm body plan evolution. While the C. elegans Hox cluster is dispersed, we found species including Plectus sambesii, Trichuris suis in which the cluster is intact and not dispersed but interrupted with non-Hox genes nonetheless. The rearrangement of Hox genes in nematode Hox clusters may have underlying effects including the breakage of the usual colinearity of Hox gene expression in Nematoda. Further research regarding nematode Hox gene complements that we have identified in the present study should seek to understand whether and where the colinear expression is retained in the phylum. As more species become amenable to molecular methods48,51,76, this can be done with tools like in situ hybridization, single-cell RNA-Seq studies, and CRISPR-guided gene knockouts. In summary, our analyses suggest the potential for new and exciting work on Hox genes and regulation of Hox clusters in a broad sampling of Nematoda. New lineages with unified Hox gene clusters will likely emerge as more complete high-quality genomes become available.

Methods

Genome data

We included in our analysis 80 nematode genomes, including newly sequenced genomes from free-living Enoplia (Clade II) and stem-group Chromadoria (outside Clade III-V), lineages that previously lacked genomic resources (see Fig. 1). New draft genomes were generated at both the Sanger Institute and the Worm L̃ab laboratory at the University of Cologne (see Supplementary Table 1). We included four genomes from sister lineage Nematomorpha, Acutogordius australiensis, Nectonema munidae, Gordionus montsenyensis, and Gordius aquaticus, as an outgroup (see Supplementary Table 1).

Figure 5
figure 5

Hox gene identification workflow. The initial inputs of genome or proteome in fasta format and Hox gene reference profiles from the HOX database are colour-coded green. Software tools (and versions) are colour-coded blue while output files are colour-coded grey. This pipeline does not include non-Hox gene and miRNAs identification.

Full size image

Hox Homeodomain identification

We collated a reference dataset of C. elegans, and Branchiostoma floridae (amphioxus) Hox homeodomain proteins from HomeoDB77. In the framework of the highly conserved Hox orthologues, the choice of these species allowed us to search with the reduced Hox gene complement of C. elegans as an ingroup and the full amphioxus set as an outgroup. We performed a sequence similarity search using the BITACORA78 gene family analysis pipeline in both genome and protein mode with default parameters (e-value of 10-5). The output of the first round of searching was used to generate an HMMER79 search profile containing orthologous Hox genes from multiple nematode species. Final predicted proteins and corresponding gene annotation files for each genome were generated through a second round of Hox gene orthologue search in the BITACORA environment using the HMM models. The predicted proteins were aligned with MAFFT80 and corresponding homeodomain sequences were visualised and extracted using Aliview81. Phylogenetic trees were reconstructed using the homeodomain protein sequences by maximum likelihood implemented in IQTREE282. The final trees were visualised using Figtree (http://tree.bio.ed.ac.uk/software/figtree/). (See Fig. 5 for visualized analysis pipeline).

Hox cluster annotation and gene identification

After the identification of HOX homeodomains, we extracted the respective scaffolds or contigs for annotation of complete Hox genes in recently available genomes without genome annotation. To search for potential clusters of Hox genes in assembled genomes, genomic scaffolds or contigs containing Hox genes were extracted from the respective assemblies using seqtk (https://github.com/lh3/seqtk). We used AUGUSTUS (v3.5.0)83 to predict complete protein models in contigs and scaffolds. To improve the accuracy of gene prediction AUGUSTUS was trained on C. elegans and a reference set of Hox gene profiles from C. elegans and B. floridae as hints for Hox gene finding. A database of predicted proteins from AUGUSTUS was constructed for search using BLASTP (v2.9.0)84 with an e-value of 10-5 using a query file containing a reference set of Hox protein sequences. Hox gene loci were manually inspected if they were in clusters (We define a Hox cluster as a group of four or more Hox genes spanning genomic loci with a size less than the size of the loosely connected C .elegans Hox genes (i.e less than 0.3 kb) on the same contig or scaffold. To validate the presence of Hox clusters we used HbxFinder pipeline available at GitHub (https://github.com/PeterMulhair/HbxFinder) to identify homeobox genes in the respective contigs or scaffolds. Hox gene cluster organisation and strand orientation maps from the AUGUSTUS output GFF files were drawn using Affinity Designer 2 (https://affinity.serif.com/en-gb/learn/designer/desktop/).

Non-Hox gene interruptions within Hox clusters

Genomic contigs or scaffolds containing Hox genes were extracted from the genome assembly using seqtk (https://github.com/lh3/seqtk). Subsequently, structural annotation was conducted using AUGUSTUS (v3.5.0)83 and functional annotation of genes within Hox cluster boundaries was performed using eggNOG-mapper v285, leveraging pre-computed eggNOG v5.0 clusters and phylogenies to assign orthologous groups and infer functional annotations. Non-Hox genes present within the Hox cluster were extracted and counted from the respective GFF structural annotation files.

microRNA interruptions within Hox clusters

To examine the microRNAs within the Hox cluster, we employed Caenorhabditis elegans microRNA references from miRBase86, a comprehensive resource for microRNA sequences and annotations. Initially, custom nucleotide databases were built for each species’ Hox cluster. Following this, we performed searches against these custom databases using established mature miRNAs from Caenorhabditis elegans sourced from miRBase, setting the e-value threshold to 10-5 in BLASTN84 analysis.