Evolutionary Conservation of Hox Genes in Vertebrate Brain Development

Hox genes are master regulatory genes that share a highly conserved DNA sequence called the homeobox. The eight Hox gene paralogues present in Drosophila, called the homeotic complex, were identified through mutations of the thorax, abdomen, legs, and antennae that caused homeotic transformations ⁶. Over time, more research emerged on these master regulatory genes showing their evolution and conservation among organisms. In Drosophila, the HOM-C complex, which is responsible for the body patterning, segmentation, and positional specification along the anterior-posterior axis in the Drosophila embryo, has conserved homologues in vertebrates, which are expressed in the same pattern of colinearity that is characteristic of Hox gene expression in the Drosophila embryo⁷. The homeotic selector genes antennapedia (Antp) and bithorax are highly conserved and possess derivatives that can be found in paralogous groups on four different chromosomes of the vertebrate genome. The chromosomes are labeled with letters A-D and the paralog groups are numbered from 1-13 with each group containing 9-11 different Hox genes⁶. Vertebrate Hox gene clusters contain the instructions for anterior-posterior and dorsal-ventral axis patterning among vertebrates, which is similar to the HOM-C complex that segments and patterns the body of the fruit fly, Drosophila melanogaster, and other invertebrates.

Lappin et al. ⁷ reviewed the evolution of Hox genes and found that their presence could be mapped back to a common ancestor of plants around 1,000 million years ago. These organisms phenotypically do not possess segmentation, which was prior to this finding, a primary indicator of ancestral homeobox gene possession given that Hox genes are responsible for body patterning and segmentation. They found that plants, fungi, mollusks, echinoderms, and urochordates also possessed conserved forms of homeobox genes. Observation and analysis done on the 13 groups of Hox genes in vertebrates point to the conclusion that groups 1-4 pattern the developing hindbrain and groups 5-13 pattern the developing the spinal cord ^{8, 9}. Viewing these data from a molecular evolutionary perspective, this deviation could be due to numerous whole genome duplications over time. This idea is termed the duplication and divergence model, or the duplication-first model ¹⁰. Hox genes have been highly conserved throughout metazoan evolution, from phylum Cnidaria, class Anthozoa (Sea Anemone) to phylum Chordata, order Primates (Humans). This means that the same genes responsible for segmentation and regionalization in the organismal body plan invertebrates (e.g. Drosophila melanogaster)are still responsible for the specification and determination of parts of the vertebrate body plan as well. Moreover, evidence indicates that the somites involved in body patterning and segmentation are not the primary regions in which Hox genes are expressed in the vertebrate body plan; rather, it is the nervous system in both Drosophila and vertebrates ⁶.

The central nervous system (CNS) is polarized at the earliest stage of brain development in vertebrates and invertebrates. The mechanism by which embryonic ectoderm is polarized to neuroectoderm is a result of a conserved molecular signaling system known as the Decapentaplegic (Dpp)/Bone morphogenetic protein (BMP)signaling cascade ¹¹. BMPs, such as Noggin and Chordin, are essential for the formation and development of the central nervous system (CNS) because they activate the signaling pathway responsible for the differentiation of the non-neural ectoderm into the neural plate. The morphogens released from the signaling cascade produces an antineurogenic effect which represses the identity of the neurons to the dorsal side of vertebrates – mice, chickens, zebrafish - and the ventral side of invertebrates – fruit flies and tunicates. This repressive antineurogenic effect localizes the formation of the neuroectoderm to the respective dorsal or ventral side of the embryo that does not express Dpp/Bmp. Pro-neural genes found in Drosophila – achete, scute, and lethal of scute – have homologues in mammals, which are also pro-neural genes that are necessary for the generation of neuronal tissue and neural stem cells. These homologues include Mash1, Ngn1, and Ngn2 ^{11, 12}. Additionally, anotherancestral gene found in the Drosophila embryo – achaete-scute – is involved in the control of sensory neurons in Drosophila and is controlled by the homeobox gene, Ubx, which is homologous in function and composition to Mash1 in vertebrates. Specifically, in mammals, the orthologue Mash1 has a positive effect on the formation of noradrenergic interneurons; Ubx has the opposite effect in Drosophila given that it suppresses this action along with the formation of bristles in the bithorax. Nonetheless, these two orthologues act in a similar fashion concerning the specific stage of development in which they are activated ^{10, 13}. Subsequent analysis ¹¹ revealed another gene family that is also highly conserved from invertebrates to vertebrates known as the ems/Emx, which is necessary for early patterning and segmentation of the brain and spinal cord in both invertebrates and vertebrates.

Experimental analysis ^{10, 11} demonstrated that the Notch signaling pathway has been highly conserved over time and is essential in both vertebrates and invertebrates for the formation of the neuroepithelium. Despite the identification of these pathways ^{10, 11, 12, 13}, there still seems to be a gap in the evolutionary conservation of the genes responsible for head and brain development among invertebrates and vertebrates. Closer analysis of this conservation and the gap that seems to persist between invertebrates and vertebrates was performed to bring clarity to the genetic and molecular similarities that humans share with other vertebrates and insects. This research ¹⁰ focused mainly on cartilaginous fish and cyclostomes because these phyla are the most promising model system for closing the “gap” of Hox gene conservation among invertebrates and vertebrates. In cartilaginous fishes, Hox genes were found to have well conserved ancestral cluster organization. The first evidence of this cluster organization came from the horn shark, which is the most recently evolved member available for experimentation. The only deviation in cartilaginous fishes in terms of the Hox genes is the absence of HoxC in the spotted dogfish. Otherwise, the entire genome sequence of Hox genes and their derivatives in cartilaginous fishes conserved more ancestral members of Hox genes than the class of bony fishes (Osteichthyes). Furthermore, intergenic sequences within the Hox clusters themselves have been conserved over a period of 250 million years ¹⁰.

The genetic pathways that act in tandem with Hox genes have also been conserved across phyla. For example, Reichert and colleagues ¹¹ showed that the cephalic Gap gene otd/Otx, which is essential for the formation of the anterior brain primordium and for the marking of the anterior portions of the embryo up to the midbrain/hindbrain boundary, is conserved in insects and mice and is responsible for the growth and development of the anterior brain primordium. Otx1 and Otx2 are mouse orthologues of the Drosophila cephalic Gap genes; in both Drosophila and mouse embryos, when the otd/Otx Gap gene was knocked out, the premature anterior portions of the brain did not form properly, thus concluding that this cephalic Gap gene is necessary of the formation of the anterior brain primordium. Otx orthologue expression has also been discovered in human, chick, Xenopus, and zebrafish embryos, supporting the adaptation and conservation of Gap genes across metazoans ¹⁴. Even before the formation of neuronal structures, the pathway responsible for the generation of the neuroectoderm was also highly conserved among animals ^{11, 15}. Conservation of this cluster organization and positioning is also seen in the lamprey; it contains a similar pattern of expression and Hox genecluster organization in comparison to mammals. The only difference between the two is that mammals possess four Hox clusters while lampreys possess six ¹⁶. The most parsimonious explanation for the conservation of clusters and possession of excess clusters among vertebrates is the duplication and divergence model.

The duplication and divergence model of evolution states that the highly conserved Hox genes seen in many different species have duplicated before each one of the lineages diverged overtime from a common ancestor ⁹. This hypothesis is supported by observations of Hox genes in amphioxus (i.e. lancelet; phylum Chordata, subphylum Cephalochordata). Amphioxus does possess ancestral characteristics of chordate anatomy but lacks cranial neurons and neural-crest derived mesenchyme, the primary region where Hox genes act in chordates and more specifically, vertebrates. It is interesting though that these organisms possess somite-like blocks in their spinal cord indicating that Hox genes may play a part in the development of amphioxus. The conclusions of this analysis showed that the amphioxus Hox clusters observed shared many similarities to mouse Hoxb-3 proteins and sequence organization pointing to homologies among vertebrate Hox clusters and amphioxus Hox clusters, further supporting the duplication and divergence model of Hox genes ⁸.

Research done on zebrafish and other teleost HoxD complexes revealed that the majority of teleosts possess all four ancestral clusters of mammalian Hox genes. In zebrafish specifically, the entire 5’ end of the HoxD cluster found in mammals is preserved in zebrafish ¹⁷. The genomic data of hagfish (clade Cyclostomata, Myxini) was also examined and nineteen Hox genes were detected. Genes from paralogous group 1 (PG1) which are significant in rhombomere, reticulospinal, and branchiomotor neuron specification were also detected in the hagfish ¹⁸. Taken together these data reveal that fibrous skeleton fishes (e.g. hagfish) cartilaginous fishes (clade Chondricthyes, Elasmobranchii: sharks and rays), and ray-finned fishes (clade Osteichthyes, clade Teleost, class Actinopterygii: zebrafish) possess extremely well conserved ancestral clusters of Hox genes ^{17, 19}. Conclusions from the data obtained in these studies on the duplication and divergence model point to the idea that even if genes are duplicated, the resulting duplicates could amass numerous deleterious mutations, become a pseudogene, or be completely lost due to independent gene loss in a lineage ^{17, 20}. Nonetheless, these genes could still prove to be essential to development and overall function of the vertebrate embryo. This hypothesis was supported when viewing the duplicated genes of Hoxb1 in zebrafish; both hoxb1a and hoxb1b were found to be necessary for specification of segments in the hindbrain, termed rhombomeres, thus supporting the hypothesis of sub-functionalization ²⁰. Experiments performed on pufferfish, which have fused bones in the head and jaw and lack ribs and pelvic fins, were also found to have duplicated genes that were necessary for development and specification in a more “primitive” vertebral structure. Both pufferfish and zebrafish share Hoxab, Hoxaa, Hoxba, and Hoxbb clusters; the difference among these four duplicates is roughly zero to two genes. This conservation supports duplication before the divergence of the pufferfish and zebrafish lineages. Another example of the duplication/divergence model is the gene Hoxa7a, a pseudogene in the pufferfish lineage; Hoxa7a is functional in fellow clade members’ striped bass and tilapia being necessary for proper skeletal development ²¹. These data show that this gene is necessary for development in striped bass and tilapia but not in pufferfish. This is most likely due to inactivity of the pseudogene in pufferfish given that it had a specified function in closely related species. One conclusion is that given the persistence of Hox gene clusters, the lack of segmentation in the body plan of an organism does not discourage the expression and apparent necessity of these genes for development. Examination of Hox genes and their specific roles in the development of the anterior-posterior (A-P) and dorsal-ventral (D-V) axes in vertebrates and invertebrates has been rapidly emerging since the discovery of Hox genes. The common theme among the majority of this research is that Hox genes do not pattern the body of vertebrates like they do in Drosophila; rather, it is the brain, spinal cord, and spinal column along the A-P and D-V axes, in tandem with branchiomotor, sensory, cranial, and neural-crest derived neurons and neuronal signals, that these Hox genes act upon ²².

In the Drosophila embryo, expression of Hox genes is easily examined early in development given that the embryo, or syncytial blastoderm, is polarized and determined by maternal effect genes. In vertebrates, polarization and determination of the central nervous system (CNS) is not observed because of the numerous gene duplications and lineage divergences of Hox genes from insects to vertebrates. However, conserved segmentation is still seen in the formation of somites, their boundaries, and specification of rhombomeres in the hindbrain and spinal cord. To test the significance, and genomic evolution, of Hox gene expression in these areas, loss-of-function experiments were done to test if Hox genes really were vital to the development of these areas in vertebrates and invertebrates. An investigation ¹¹ on the neural regionalization of vertebrates and invertebrates exhibits that Drosophila genes necessary for the formation of columnar domains (i.e., epithelial tissue made of column-shaped epithelial cells) – und, ind, and msh – have orthologues in vertebrates – Nkx (und), Gsh (ind), and Msx (msh) – which are also responsible for the development of columnar domains in the central nervous system. Similarly, in the development of the hindbrain in vertebrates, a series of seven to eight rhombomeres visibly divide and are controlled by the same conserved pattern of serial segmentation observed in the Drosophila embryo. These patterns have been extensively studied using combinations of loss-of-function and gain-of-function analyses; these analyses revealed that there are distinct cellular and molecular characteristics that have been highly conserved from the earliest invertebrates to vertebrates. This so-called Hox Code is what gives each rhombomere in the premature hindbrain its identity; the Hox Code is vital to the specification and determination of the cells that will create the vertebrate hindbrain ⁹.

The expression of Hox genes is also seen in the formation of the branchial arches, visible masses of tissue that will form the head and neck in vertebrates (subphylum Craniata). Development of the vertebrate head and neck is initiated by Hox gene expression in the branchial arches, which have been determined to help organize, pattern, and segment this primitive head, neck, and brain structure ²³. Hunt and coworkers ²⁴ termed this discovery the ‘branchial Hox code.’ The branchial Hox code proposes an early signaling cascade that lays the foundations of Hox genes, labial genes, and other derivatives in the formation of the neural crest. It also shows the characteristic nested expression of Hox genes within each developing rhombomere of the hindbrain until it reaches the r2/r3 boundary of the A-P axis. This branchial Hox code is also seen in the formation and restriction of Hox-2.5 expression in the spinal cord of the mouse by 10 days post-coitum. These authors ²⁴ concluded that the role of this nested Hox expression, known as colinearity, acts on the ventral regions of mouse embryos as well. Colinearity is what shows us how each gene is activated early embryogenesis. This activation is also determined by the linear order of the clusters along their specific chromosomal clusters. Overall, the significance of colinearity is due to the nesting and overlapping of Hox gene expression if segments that will form the branchial arches and hindbrain of the vertebrate embryo. Each segment forms because this process contributes to the formation of the A-P and D-V axis of the vertebrate embryo. The nesting and overlapping pattern of Hox gene expression also relates back to the Hox code whereby different combinations of Hox proteins are synthesized temporally throughout the body plan of the embryo contributing to the developing hindbrain, nascent spinal cord of the embryo, and somites (vertebral column precursors) ^{9, 25}.

Group B Hox genes, specifically Hoxb3 and Hoxb9 are strongly expressed in the mouse developing spinal cord along the D-V axis forming an ‘M’ shape in the presumptive neural tube ^{9, 24}. While performing a loss-of-function experiment to test the branchial Hox code hypothesis, Hunt and coworkers ²⁴ showed that loss of Hox-1.5 activity was detrimental to normal D-V axis formation; mouse embryos showed defects in the neural crest and neural-crest derived mesenchyme but cranial ganglia developed normally. To date, scant research has been performed on the expression of Hox genes and their derivatives in the cranial mesoderm, which is the tissue that lies just below the neck. However, research on cranial nerves and cranial ganglia and their respective roles in the formation of the neural crest has been extensively studied in mice, chick, zebrafish, Xenopus, and Drosophila embryos ^{14, 19}. While there has been little research performed on the action of Hox genes in the cranial mesoderm - a particular kind of tissue that is derived from the paraxial mesoderm - these data, reviewed above, supports the hypothesis that Hox genes work to pattern the dorsal-ventral axis, the formation of the spinal cord, and the brain in vertebrates.

In the fruit fly, Drosophila melanogaster, genetic mutations of the abdomen, thorax, or head results in that segment of the body taking on the identity of the other (homeotic transformation). For example, a mutation in the Antp (Antennapedia) gene of Drosophila results in an antenna being transformed into a leg. A process similar to homeotic transformation is observed in vertebrates from misexpression (e.g. duplication) of Hox genes in the developing axial skeleton. The ancestral murine gene, Hoxb1, is necessary for proper formation of rhombomere 4 (r4) in the developing hindbrain. Rhombomere 4 is necessary for proper migration of the VIIth facial branchiomotor neurons in mammals ²⁰. Zebrafish possess duplicate copies of Hoxb, hoxb1a and hoxb1b,which have similar functions as the mammalian orthologues Hoxa1 and Hoxb1. Interestingly, morpholino driven down regulation of hoxb1a and hoxb1b in zebrafishresulted in striking changes to cranial/facial development due improper migration and expression of the hoxb1 duplicates. These data demonstrate that while the hoxb1 duplicates completely altered the organization of each rhombomere in the hindbrain itself, the orthologous copy of Hoxb1 found in mice did not alter hindbrain segmentation when lost. Mouse Hoxa1 null animals had phenotype similar to the loss of the hoxb1 duplicates in zebrafish preventing branchiomotor neurons responsible for rhombomere 4 identity from migrating downstream to their specific segment; therefore these orthologues play similar roles in mouse and zebrafish neuronal development “revealing that a ‘function shuffling’ among paralogues has occurred during vertebrate evolution” ²⁰.

In pufferfish ²¹, group two and four Hox genes and their duplicates are responsible for formation of r2-r5 borders in the hindbrain and for the evolutionary advantageous function called “puffing” that gives pufferfish their names. The Hoxa2a gene responsible for this puffing probably resulted from duplication and divergence over time; interestingly, this gene’s function in the pufferfish could be a result of evolution in the motor neurons of the pufferfish lineage given that zebrafish and other teleost fish possess a pseudogene of Hoxa2a that is non-functional in their genome ²¹. Experimentation ⁹ performed on Hoxa2 also revealed that the genefunctions as a transcription factor in the formation of the cranial neural crest. The ability of Hoxa2 to initiate the migration of neural crest cells to the second branchial arch of the hindbrain was dependent on the expression of the ancient Ap-2 gene. If Ap-2 is not expressed in the early development of the hindbrain, Hoxa2 also fails to be expressed indicating that migration of the neural crest cells to the second branchial arch does not occur ⁹. These observations allow for the conclusion that Hox genes are not only working to form boundaries and segment the hindbrain, but they are also responsible for the neural regionalization and migration of specific neural crest cells in the hindbrain and spinal cord.

Work ^{11, 25} on the formation of the hindbrain and spinal cord has established that hindbrain segmentation has a “striking resemblance” to the embryonic body plan of Drosophila melanogaster. Multiple studies on Hox genes in the hindbrain revealed that Hoxb1 plays a major role in the cascade of genes that helps to form the rhombomeres in the hindbrain. Specifically, Hoxb1 functions to form r4. In Kiecker and colleagues ²⁵ review of chordate evolution and hindbrain segmentation, Hoxb1 loss-of-function experiments in mice demonstrate that the identity of r4 is ectopically expressed as r2-like in the branchial arches and other segments of the hindbrain. These authors ²⁵ also clarified that this same phenotype was present in zebrafish embryos lacking the duplicate hoxb1a gene, revealing evolutionary conservation and sub-functionalization as predicted by the duplication and divergence model. The reason for this identity shift is due to the inability of branchiomotor neurons (i.e. VIIth facial branchiomotor neurons) in these mutant animals to migrate to their proper anatomical positions to form the proper boundaries necessary for segmentation, neural migration, and neural regionalization of rhombomeres in the hindbrain ^{20, 25}.

Analysis on the functions of Hoxb1 in hindbrain development reveals that Hoxb1 acts synergistically with Hoxa1, a group 1 paralog. In the vertebrate brain, a gain-of-function of Hoxa1 leads to the transformation of r2 into a r4 identity, which is the opposite reaction of loss-of-function in Hoxb1 demonstrating the transformation of r4 into r2-like expression. Duplicates of the ancestral Hoxa1 and Hoxb1 copies are found in the zebrafish embryo; when the hoxb1 duplicate in zebrafish was analyzed in a gain-of-function experiment, the result was a transformation from r2 identity to r4 identity, which is similar to the gain-of-function experiment done on Hoxa1 in the mammalian brain. This transformation is observed in the ancestral mammalian gene Hoxa1, which shows that the conservation and subsequent shuffling of functions among duplicated genes has occurred over time ^{15, 16, 20}.

The inactivation of Hoxa1 in the hindbrain also led to the deletion of rhombomere segments in the hindbrain. These data support the hypothesis that Hoxa1, coupled with Hoxb1, are responsible for the segmental identity of specific rhombomeres in the hindbrain. Inspection of the specific role that Hoxa1^nullmutants have in relation to Hoxb1 in the hindbrain revealed the coupled reactions of Hoxb1 and Hoxa1 proposed above ¹⁵. Hoxa1 is one of two known paralogs expressed in the CNS. Double heterozygous and homozygous mutant embryos for Hoxa1 and Hoxb1 indicated that heterozygous embryos possessing one functional copy of Hoxb1 and a mutant copy of Hoxa1 showeddramatic alterations to r4 and a significant decrease in proper formation of r4. This shows that without Hoxa1, the one functional copy alone is not enough to maintain proper expression of r4 in the developing hindbrain ¹⁵. The synergistic interactions of Hoxb1 and Hoxa1 can be explained by auto- and cross-regulatory pathways, which are essential for patterning the hindbrain ²⁵.

Auto-regulatory pathways are internal biological processes that are responsible for an organism’s response to external stimuli whereas cross-regulatory pathways are responsible for controlling metabolic pathways through the products made by related, but entirely different pathways. Nolte et al. ⁹ illustrated that these two pathways work in tandem with Hox genes to cross-regulate the expression of other Hox genes or sustain their own expression through auto-regulatory mechanisms. Polycomb and trithorax controlled regulatory pathways, which determine segmental expression in the fruit fly, Drosophila melanogaster, were also found to regulate vertebrate Hox gene expression in the spinal cord and hindbrain; without the mouse homologues PcG and trxG present, Hox gene expression is altered leading to improper expression of Hox genes in rhombomeric segments and skeletal defects. Hox proteins and response elements such as retinoic acid (RA), fibroblast growth factor (FGF), and Krox20 were also found to self-regulate the feedback loops that generate and maintain this segmentation ⁹. The auto and cross-regulatory loops of the hindbrain and spinal cord are most evident in rhombomere 4 (r4), which as previously mentioned is patterned and segmented by Hoxb1 and HoxaHoxa1. In addition to this, retinoic acid response elements (RAREs), fibroblast growth factors (FGFs), Krox20, Pbx, and Kreisler seem to aid in the signaling cascade for either auto- or cross-regulatory loops in tandem with group 1 and group 4 Hox genes in the hindbrain ^{9, 15, 26}.

As mentioned above, paralogous group 1 Hox genes are primarily responsible for patterning r4 in the hindbrain. In Nolte and Krumlauf’s review ⁹ they discuss how these patterning mechanisms showed that retinoic acid (RA), which is a derivative of vitamin A, is directly connected to embryological defects in the formation of the anterior hindbrain and spinal cord when in excess. Ectopic expression of retinoic acid also resulted in the transformation of rhombomere boundaries; whereas, the abrogation of RA in early stages of development resulted in the complete loss of the hindbrain and spinal cord. The most anterior Hox genes in the hindbrain are known as 3’ Hox genes and include Hoxb1 and Hoxa1; the Hoxb1 locus is necessary and sufficient to regulate neural and mesoderm expression, but the 3’ RARE on this locus was found to activate endogenous Hoxb1 in the early neural ectoderm by way of an auto-regulatory loop. Investigation of this hypothesis was done by a so-called “hit-and-run” model which applies and targets a specific germline mutation in the organism; in this case, the mutation was a loss of function of 3’ RARE. Subsequently, a loss of 3’ RARE resulted in improper levels of cis-regulatory elements, which establish proper formation of the neuroectodermal and mesodermal expression of Hoxb1¹⁵.

Another study done on cross-regulation and expression of Hox genes examined retinoic acid receptor ß (Rarb) showed that Hoxb4 and Hoxd4 along with the two Hox genes from PG1 (paralogous group 1) discussed above, required the participation of retinoic acid in order to form rhombomere boundaries and segment the primitive hindbrain and spinal cord. Data from this experiment on the mouse embryo revealed that Hoxb4 and Hoxd4 have congruent expression with Rarb in r6/r7 at around day 8.5 or 9.5 post coitum. A protein known as Raldh2 produced results that were similar to a loss-of-function of RA ²⁶; loss of Raldh2 gives rise to abnormalities along the A-P axis due to disrupted Hox expressions. These authors ²⁶ also showed that there is some auto-regulatory loop function, as described above, in PG1 genes in r4 of the hindbrain, among Rarb, Hoxb4, and Hoxd4 in later stages of embryonic development. Mutation of PG4 genes did not produce significant alterations, but did lead to dramatically reduced neural tube expression in the D-V axis of the mouse embryo. Therefore, RA/receptor complex binding to a distal enhancer showed that RA and PG4 Hox genes are responsible for D-V patterning as well as A-P patterning. Additional results from these experiments illustrate that Hoxb4 and Hoxd4 regulate the early expression of Rarb in the primitive hindbrain, neuroectodermal, and mesodermal tissues ^{15, 26}. While RA involvement in these processes is vital, it is not the only protein and transcription factor needed for appropriate expression in the hindbrain and spinal cord.

Fibroblast growth factor (FGF), Krox20, Kreisler, and Pbx are all necessary for the regulation and function of Hox genes in the vertebrate nervous system. Gain-of-function mutations of FGF can increase endogenous expression of Hox genes in the mesoderm and neural tissue. A loss of FGF results in the opposite of up-regulation in the formation of the vertebrate hindbrain and spinal cord. Specifically, research ⁹ performed in mouse embryos reveals that Fgf2, Fgf3, Fgf8, and Fgf14 have the most effect on central nervous system development. FGF is also vital for the activation of Hox group genes 5-13 which are expressed in a colinear fashion as explained earlier in this review ⁹. Without induction by FGF derivatives, the most caudal region of the hindbrain and the most rostral region of the spinal cord do not form properly. Thus, it appears that FGF is a prerequisite for the signaling cascade activating the expression, regulation, and segmentation action of Hox genes in the hindbrain and spinal cord.

Fgf3, Fgf8, Krox20, and Kreisler are expressed in r4; expression data indicates that Krox20 and Kreisler are part of an auto-regulatory loop, which activates the expression of FGF ligands and receptors. This implies that Krox20 and Kreisler work to initiate proper expression of FGF ligands acting further upstream of the signaling cascade responsible for Hox gene expression. Krox20 expression is also necessary to form primitive boundaries for r3 and r5; expression of this protein following formation of these rhombomeres is significantly down regulated. In order for Krox20 to act in its specific rhombomere segments, however, it must bind to regulatory elements upstream of mammalian Hoxa2, Hoxb2, and Hoxb3. Without Krox20 activation, no expression of these Hox genes occurs and formation of the rhombomere boundaries is not specified or determined. Kreisler functions in a very similar way but it works to form the r5 and r6 boundaries within the hindbrain. Interestingly, a loss of Kreisler results in inner ear and hindbrain malformations. The duplicated version of this gene in zebrafish was also found to regulate hindbrain expression; for example, if Fgf8 expression is lost in the fish and chick hindbrain, then Krox20 and Kreisler are not expressed. This leads to developmental defects in the embryo because Hoxa2 and Hoxb1 are not activated to form the r1/r2 boundary and the r4 boundary while Krox20 and Kreisler are not able to specify and determine the r3, r5, and r6 midbrain/hindbrain boundaries, respectively ^{9, 27}. A loss-of-function experiment¹⁶on Amphioxus hox 1, hoxa2, and hoxa3 showed that without the action of Kreisler and Krox20, segmentally restricted expression of rhombomeres in the hindbrain does not occur properly at the r6/r7 boundary in the hindbrain. These analyses show that Kreisler, Krox20, and FGF are all significant in patterning the hindbrain ^{9, 16, 27}.

Pbx, a Hox cofactor and homeobox-containing gene encoding a nuclear protein, was also found to work alongside Hox genes to pattern and segment the hindbrain. A loss-of-function experiment done on Pbx showed that the transcription factors produced from the nuclear protein were necessary to activate segmentation of Hox genes in the hindbrain ^{2, 7}. In the zebrafish embryo, loss-of-function of pbx2 and pbx4 protein corresponds to improper rhombomere identity (e.g. r2 forms as r1) and formation of rhombomeres 2 through 6 is entirely lost. This is due to lack of Pbx transcriptional activation of Hox genes involved in segmentation in the hindbrain²⁸. These experiments also pointed to the conclusion that the Hox/Pbx binding sites are vital to the auto- and cross-regulatory loops that regulate and initiate Hox genes in the hindbrain and spinal cord. As we have seen, auto- and cross-regulatory loops are what contribute to the colinear nature and nested expression of Hox genes and their derivatives, in the hindbrain and spinal cord while also contributing to their self-regulation and establishment in the primitive CNS of the embryo. Hox-1/Pbx binding sites are also directly involved in regulation of r4 in the zebrafish. To bring together the function of Pbx and the other proteins involved in the regulation of Hox gene expression, research²⁷ has also shown that loss of pbx2 and pbx4 in the zebrafish correlates directly and indirectly with down-regulation of FGF signaling in the hindbrain. Hox-3 paralogs in zebrafish are also regulated by an auto-regulatory loop dependent on Pbx expression;therefore, in order for proper expression of Hox genes to occur in each rhombomere of the hindbrain, Pbx must be functional and normally expressed to specify and determine each rhombomere in the hindbrain. Without Pbx, pharyngeal arch malformations, severe anemia, and other deformities are observed in mice and zebrafish embryos ²⁷. Additionally, lack of Pbx protein inhibits the expression of Hox genes and proper formation of primitive neuroectoderm and mesodermal tissues in the A-P and D-V axis ^{15, 26, 27}.

This work was supported by the Department of Science and Mathematics of Judson University (S. G. S. and J.O.H) and by funds from the William W. Brady Chair of Science endowment (J.O.H).