Congenital and idiopathic scoliosis constitute two major categories of spinal curvature. Idiopathic scoliosis, as defined by the Scoliosis Research Society refers to a lateral curvature of 10˚ or greater on plane radiographs, is not associated with any underlying cause [1]. Congenital scoliosis is a spinal curvature that results from developmental abnormalities in vertebral bodies, which are referred to as congenital vertebral malformations (CVM). These abnormalities may further be divided into disorders of formation (wedge vertebrae, hemivertebrae) or disorders of segmentation (vertebral bar). Both congenital and idiopathic scoliosis are clinically and etiologically heterogeneous. Although the genetic mechanisms responsible for both conditions are not well understood, there is an observed prevalence of 17.3% of congenital scoliosis in families with idiopathic scoliosis, suggesting similar underlying pathogenic mechanisms [2].

Mutations in genes associated with somitogenesis represent ideal candidates for scoliosis. Mesodermally derived somites are paired structures that give rise to the vertebral bodies, ribs, spinal and rib associated muscles and tendons. Somitogenesis occurs by an intricate interplay of patterning genes which are expressed in the presomitic mesoderm. Much of our understanding of the somitogenesis process has been derived from experimental work in mice, chicken and Xenopus species. The Notch, Fgf and Wnt signaling pathways regulated by a molecular oscillator in which the Notch and Fgf genes oscillate in opposite phase to the Wnt genes [3]. Since a large number of human disorders are characterized by aberrant spine development including congenital scoliosis, spondylocostal dysostosis, spondylothoracic dysostosis, Klippel-Feil syndrome (fusion of cervical vertebrae, short neck), hemifacial microsomia (ear tags, microtia, cardiac abnormalities, vertebral abnormalities) and VACTERL syndrome (vertebral anomalies, anal atresia, cardiac abnormalities, tracheo-esophageal fistula, renal abnormalities, limb abnormalities) the clinical relevance for understanding the genetic basis of somitogenesis becomes important. Several prior studies have incorporated a candidate gene approach based on the assumption that human genes that are syntenic to mouse genes are associated with spine development [3-9].

There are inherent difficulties in the identification of genes contributing to scoliosis in humans. Most cases of congenital vertebral malformations (CVMs) represent sporadic occurrences within a single family, thus making traditional linkage approaches difficult to utilize. The large number of potential candidate genes to choose from, compounded by a clinical heterogeneity of CVM phenoltypes, makes this a difficult area to provide genetic diagnosis and counseling for families.

Multiple factors may contribute to the development of idiopathic scoliosis including muscle imbalance and changes in the connective tissue matrix. Linkage and association studies have identified a number of genetic regions associated with idiopathic scoliosis [10-13]. Polymorphisms in CHD7, a chromeodomain helicase which is associated with CHARGE syndrome (Coloboma, Atresia Choanae, Retarded Growth, Ear Anomalies), have been associated with idiopathic scoliosis [11]. CHD7 is associated with embryonic axial development in mice, providing additional evidence that congenital and idiopathic scoliosis may have a unifying pathogenetic mechanism [14].

Prior studies have demonstrated that considerable evolutionary activity exists at the evolutionary breakpoint regions (EBRs) which are located between homologous synteny blocks (HSBs) including reuse, increased gene density, segmental duplication accumulation and the emergence of centromeres and telomeres [15]. EBR is defined as an interval between two adjacent HSBs that is demarcated by the end-sequence coordinates of those HSBs on each side. Because the process of spinal column development is similar among amniotes, we hypothesized that genes associated with scoliosis are conserved in amniotes and their location is within the regions of conserved synteny of chromosomes in different mammals.

Ninety seven patterning genes including genes from the Wnt, Fgf, and Notch signaling pathways in addition to other patterning genes operative in mice somitogenesis and associated with scoliosis phenotypes, were initially identified for synteny block analyses [3]. The analysis was performed in order to determine whether these genes involved in somitogenesis and scoliosis are preferentially located in the regions of mammalian chromosomes that are stable in evolution, or whether they are located in the regions that correspond to positions of EBRs in the genomes of several amniotic species (human, chimp, macaque, mouse, rat, dog, pig, cattle, opossum, chicken).

Mouse gene coordinates corresponding to the human chromosome coordinates for the 77 genes from Table 1 (20 of the original 97 patterning genes did not have corresponding mouse coordinates) were obtained by using Ensemble homology tables [16]. The human genome was divided into 2980, 1 Mbp intervals; the number of the genes from Table 1 was counted in each of those intervals. A determination was made as to which bins are overlapping with positions of the HSBs or EBRs. Student’s t-test analysis with unequal variances, as described previously, was performed to determine whether the somite patterning genes are preferentially located in the EBRs or HSBs [17,18].

Vertebral patterning and scoliosis associated genes in Table 1 were found to be preferentially located away from the EBRs, with approximately twice as many genes on average occurring in other parts of the genome as compared to the breakpoint intervals (p-value = 0.011). While this does not appear to be a large difference, if all genes in the genome are counted, the EBRs on average contain ~2 times more genes than the rest of the genome. In general, breakpoint intervals are significantly enriched for genes, and the results of this analysis indicate that they are not enriched for vertebral patterning genes. Examination of large blocks (>3 Mb) of homologous synteny (approximately 7 of these occur in amniote genomes, which are

>16.3 Mbp in human coordinates) indicated 0.04 genes from Table 1 localized in these blocks on genome average, while 0.03 genes localized to the rest of the genome [19]. This result is not statistically significant, probably because of the small number of genes in this comparison.

Synteny block analysis performed on 77 genes associated with Wnt, Fgf and Notch signaling pathways indicated that these genes are located away from the boundaries of EBRs. The location of vertebral patterning genes away from synteny breakpoints highlights their important and conserved evolutionary function in amniotes. This is the first study to analyze conserved synteny for genes associated with somitogenesis in amniotic species and provides additional genetic evidence for similarities in spine development in amniotes.

A prior analysis of mouse scoliotic phenotypes using the Mouse Genome Database (MGD), followed by use of the Online Mendelian Inheritance in Man (OMIM), yielded 45 genes with possible scoliosis phenotypes. Twenty eight genes were translated to the human genome coordinates using mouse and human synteny maps [8]. These included WNT3A and DLL3 genes, also members of the cohort of genes in Table 1. During this analysis it was not possible to determine whether each vertebral patterning gene was located within EBRs or away from breakpoint intervals. The localization of patterning genes associated with human vertebral development to regions away from synteny breakpoint intervals provides evidence for conservation of the basic vertebral patterning scheme during amniote development.

These data are consistent with prior analyses performed by Larkin et al. [19]. Gene networks associated with major developmental processes such as neuronal development, central nervous system, bone and blood vessel development, some of which were mediated by Wnt and Notch signaling pathways, were significantly enriched in HSBs and, therefore, less likely to be localized at EBRs. Gene networks associated with responses to external stimuli such as inflammatory responses and muscle contraction were more likely to be localized to EBRs. Our study focused on 77 genes associated with somitogenesis including 11 NOTCH, 12 FGF, 29 WNT, 8 HOX, 3 PAX, 2 TGFβ pathway and 12 additional genes associated with scoliosis phenotypes and the results demonstrated these patterning genes were significantly overrepresented in the evolutionary conserved regions.

Using a series of bioinformatic approaches including neighbor-joining (NJ) and maximum parsimony (MP), contained within the PHYLIP (PHYLogeny inference package) software package [20], the evolution of Notch family proteins in species from worm to human was analyzed in C. elegans, D. melanogaster, C. intestinalis, and H. sapiens using Mapviewer, Geneview, and the BlastP and TBlastN algorithm [21]. The chromosomal distribution of PBX (pre-B cell leukemia homeobox), LHX3 (Lim homeobox 3), NRARP (Notch regulated ankyrin repeat protein), BRD (bromodomain) and CAMSAP1 (calmodulin regulated spectrin-associated protein 1) was found to follow the distribution of Notch, providing evidence for co-evolution with Notch signaling pathway genes by segmental duplication. The close proximity of these genes may reflect a functional relationship. For instance, in C. elegans, PBX appears to be responsible for transcriptional control of Notch signaling [22].

This study has several limitations. While genes in the FGF, WNT and Notch signaling pathways were analyzed, genes in other pathways such as the BMP signaling pathway were not studied. Corresponding mouse coordinates were identified for 77 of the 97 patterning human genes originally identified. Due to the relatively small number of genes studied, it was not possible to determine whether the vertebral patterning genes were preferentially localized to large blocks of homologous synteny.

Besides playing a crucial role in somitogenesis, the Fgf, Wnt and Notch signaling pathways also are involved in the embryogenesis of other organs. Fgf’s have important roles in development of the limbs, skin, central nervous system, ear, lungs, liver and have major involvement in the wound healing process [23]. Fibroblast growth factor receptor related disorders in humans include craniosynostosis syndromes such as Apert and Crouzon syndrome, and skeletal dsyplasias, of which achondroplasia is the most common. The Wnt canonical pathway is active in neural tube development [24]. Notch signaling is involved in developmental pathways which affect the vasculature, heart, eye and liver [25]. Both Notch and Wnt pathways are involved in autonomous phenotypes including cellular development and proliferation. It is possible that the conservation of Fgf, Wnt and Notch signaling related genes in HSBs reflects conservation of non-somite signaling functions or conservation of multiple signaling functions. A previous study by Larkin et al. focused on association of HSBs with respect to gene networks, while our study was aimed at localization of genes associated with a specific disease process, namely scoliosis with respect to HSBs [19].

In summary we provide further evidence that developmental pathways associated with somitogenesis are conserved across amniotes which is consistent with their known function.

We appreciate the comments and advice provided by Drs. Harris Lewin and Denis Larkin.