|Home | About | Journals | Submit | Contact Us | Français|
Significant progress has been made toward understanding the role of fgf8 in directing early embryonic patterning of the pharyngeal skeleton. Considerably less is known about the role this growth factor plays in the coordinated development, growth, and remodeling of the craniofacial skeleton beyond embryonic stages. To better understand the contributions of fgf8 in the formation of adult craniofacial architecture, we analyzed the skeletal anatomy of adult aceti282a/fgf8 heterozygous zebrafish. Our results revealed distinct skeletal defects including facial asymmetries, aberrant craniofacial geometry, irregular patterns of cranial suturing, and ectopic bone formation. These defects are similar in presentation to several human craniofacial disorders (e.g., craniosynostosis, hemifacial microsomia), and may be related to increased levels of bone metabolism observed in aceti282a/fgf8 heterozygotes. Moreover, skeletal defects observed in aceti282a/fgf8 heterozygotes are consistent with expression patterns of fgf8 in the mature craniofacial skeleton. These data reveal previously unrecognized roles for fgf8 during skeletogenesis, and provide a basis for future investigations into the mechanisms that regulate craniofacial development beyond the embryo.
Fibroblast growth factors (Fgfs) are intercellular signaling molecules that figure prominently in the network of signaling pathways that regulate early patterning and development of the pharyngeal skeleton. Specific requirements for Fgfs during craniofacial development include directing the migration, proliferation and survival of cranial neural crest (CNC) cells (Creuzet et al., 2004), formation of the pharyngeal arches (Crump et al., 2004), establishment of polarity in the mandibular arch (Tucker et al, 1999), and regulation of tooth morphogenesis (Jackman et al., 2004).
Fgfs have also been implicated in directing later stages of skeletal development, and in a variety of developmental disorders. In vitro models have demonstrated roles for Fgfs in regulating the differentiation and function of both osteoblasts (bone forming cells) and osteoclasts (bone resorbing cells), consistent with a role for Fgfs in bone growth, turnover, homeostasis, and remodeling (Shimoaka et al., 2002; Valta et al., 2006; Zuo et al., 2004). Research on mis-expression of various Fgf ligands and receptors in mice supports a critical role for Fgf signaling during the development and growth of the bony skeleton (Colvin et al., 1996; Deng et al., 1996; Govindarajan and Overbeek, 2006; Sobue et al., 2005; Valta et al., 2006; Yu et al., 2003). While certain Fgfs and Fgf receptors (Fgfrs) appear to function as positive regulators of bone development (Valta et al., 2006; Yu et al., 2003), others clearly function as negative regulators (Colvin et al., 1996; Deng et al., 1996). Fgf2 has opposite effects on bone formation during pre- and postnatal development – stimulating the production of osteoblasts during early development (Montero et al., 2000), and decreasing bone formation and density at later developmental stages (Sobue et al., 2005). Mutations in Fgfr1-3 have been implicated in both syndromic and non-syndromic craniosynostoses and skeletal dysplasia disorders including Aperts, Crouzon, Saethre-Chotzen and Pfeifer syndromes (reviewed by Nie et al., 2006). These data underscore dynamic and vital roles for Fgf signaling during skeletogenesis.
Here we examine the consequences of fgf8-haploinsufficeny on craniofacial architecture in the developmentally tractable zebrafish model. We hypothesized that if Fgf8 plays a critical role in bone formation and growth, then haploinsufficent adult zebrafish will exhibit subtle but distinct craniofacial defects. We found that aceti282a/fgf8 heterozygous adults exhibit discrete craniofacial asymmetries, altered craniofacial geometry, aberrant cranial suturing, and ectopic bone growth. Fgf8-haploinsufficency also resulted in increased levels of osteoblast and osteoclast activity, consistent with fgf8 functioning as a negative regulator of bone growth and remodeling in the zebrafish. These data provide novel insight into the functional roles of fgf8 during skeletogenesis, and provide a foundation for future investigations into the molecular mechanisms that regulate the coordinated formation, growth and remodeling of the craniofacial skeleton.
Zebrafish were maintained and bred at 28.5°C in a 14hr light/10hr dark cycle at the Forsyth Zebrafish Facility. Embryos were collected by natural mating of pair-wise crosses, and maintained at 28.5°C as described (Westerfield, 1995). For this study we used mutants that carried the aceti282a allele, which encodes a point mutation in the splice site following exon 2, causing a frame shift and introduction of a premature stop codon (Reifers et al., 1998). The aceti282a allele strongly attenuates Fgf8 signaling (Reifers et al., 1998), however some residual fgf8 message may be produced in aceti282a homozygous mutants (Draper et al., 2001). Heterogyous carriers of the aceti282a allele were genotyped by PCR analysis of tail clip DNA using dCAPS primers, bwd394 (5′ to 3′): GAGACGGACACATTTGGGAGTCGAG and bwd395 (5′ to 3′): AAGTCACAAAAGTGATGACTTTTTCAGATA (Invitrogen, Carlsbad, CA), followed by digestion with the restriction enzyme EcoRV (Neff et al., 1998).
The bony skeletons of adult zebrafish were stained with Alizarin red as described (Potthoff, 1984), with slight modifications. Adult fish between 8–12 months of age were fixed overnight in 4% paraformaldehyde (PFA) in phosphate buffered solution containing 0.1% Tween-20 (Sigma-Aldrich, St. Louis, MO). Fixed specimens were then enzymatically cleared using a trypsin solution containing 1g of trypsin powder (Sigma-Aldrich, St. Louis, MO) in 30ml of distilled H20 plus 30ml sodium borate saturated distilled H2O. After clearing, bone tissue was stained in 0.5% KOH with enough aqueous Alizarin red to turn the solution a deep purple. Pigment was then bleached in a 10% H2O2 solution, and specimens were stored in 80% glycerol. Bone was stained in living larval zebrafish using a 0.1mg/ml solution of the vital stain Quercetin (Sigma-Aldrich, St. Louis, MO), and visualized under UV illumination using a Zeiss M2 Bio stereomicroscope. All bone preparations were photographed using a Zeiss Axiocam digital imaging system mounted to an M2 Bio stereomicroscope (Zeiss), and processed in Adobe Photoshop 7.0.
All shape analyses where performed on cleared and stained adult (12 mos.) zebrafish from the same family (full siblings). Mandibular asymmetry was defined as the deviation (in μm) of the mandibular symphysis from the cranial midline (Fig. 2). Opercle asymmetry was defined as the difference in left and right opercle centroid size (CS) for each individual. CS is a geometric descriptor of size measured as the square root of the sum of squared distances of a set of landmarks to their centroid. Landmarks used to describe the opercle are illustrated in Fig. 3. The degree of aceti282a/+ mandibular and opercle asymmetry was compared to that in +/+ zebrafish using one-way ANOVA. The sidedness of asymmetry was assessed via Chi-square test and ANOVA on phenotypes that fell outside of the 95% confidence interval.
Cranial suture morphology was assessed by measuring the deviation (in μm) of the interfrontal and sagittal sutures from the cranial midline (Fig. 2). One-way ANOVA was used to compare suture morphology in aceti282a/+ and +/+ zebrafish. Patterning of the coronal (transverse) suture was also examined in aceti282a/+ zebrafish, but statistically significant differences were not observed, and these data are not presented.
Craniofacial shape was compared between aceti282a/+ animals and their homozygous wild-type siblings via landmark-based geometric morphometrics (GM) in tpsRelw 1.42 (Rohlf, 2005). Landmarks used to describe craniofacial geometry are presented in Figs. 3 and and4.4. Superimposition of landmark data was achieved using a Generalized Procrustes Analysis (GPA) (Gower, 1975; Rohlf and Slice, 1990), which superimposed landmark configurations such that the sum of squared distances between corresponding landmarks was minimized by scaling, translating, and rotating specimens with respect to a mean consensus configuration. We then performed a thin-plate spline (TPS) analysis, which models the form of an infinitely thin metal plate that is constrained at some combination of points (i.e., landmarks) but is otherwise free to adopt a target form in a way that minimizes bending energy (Bookstein, 1991). In morphometrics, this interpolation is applied to a Cartesian coordinate (x, y) system in which deformation grids are constructed from two landmark configurations (Bookstein, 1991). The total deformation of the thin-plate spline can be decomposed into geometrically orthogonal components based on scale (Rohlf and Marcus, 1993). These components (partial warps) can be localized to describe precisely what aspects of shape are different. A principal component analysis (PCA) was then performed on partial warp scores, formally referred to as relative warp analysis (RWA) (Bookstein 1996), to identify major axes of shape variation.
Osteoclast and osteoblast activities were examined in adult (12 mos.) aceti282a/+ zebafish and their homozygous wild-type siblings by assaying the presence of acid and alkaline phosphatase activity, respectfully, in the jaws of adult fish.
For detection of tartrate-resistant acid phosphatase (TRAP), adult fish were fixed over night at 4°C in 5% neutral formal-saline in 10% formalin. The following morning specimens were washed several times in distilled water, and eyes and viscera were removed, followed by decalcification in 10% EDTA in Tris buffer for 24 hrs. Specimens were then incubated in a tartrate buffer solution [50mM Na tartrate + 0.2M acetate buffer (0.2M acetic acid + 0.2M Na acetate), pH 5.5] for 2 hrs. TRAP activity was detected using a substrate solution containing 25.6% Michaelis veronal-acetate (1.94% sodium acetate · 3H20 + 2.94% sodium barbiturate), 61.3% distilled H20, 5.1% napthol AS-TR phosphate in N,N – dimethylforamide (10mg/ml), 4% pararosaniline-HCl (2% pararosaniline in 2N heated HCl), 4% sodium nitrite. The TRAP solution was made no more than 3 hrs prior to use, and specimens were incubated in this solution for 2 hrs at room temperature. Afterwards, specimens were washed several times in distilled H20. Tissue was enzymatically cleared using trypsin (1g trypsin powder in 30ml H20 + 30 ml sodium borate saturated H2O). Pigment was bleached in 10% H2O2, and specimens were stored in 80% glycerol.
For identification of osteoblasts, alkaline phosphatase (AP) activity was visualized. Specimens were fixed and eviscerated as described above, and incubated in 0.05M tris-maleate buffer pH 8.3 for 1 hr. Tris-maleate buffer was made by adding 2.42 grams of tris (hydroxymethyl) aminomethane + 2.32 grams maleic acid to 100ml H2O, adjusted to pH 8.3 with 0.2M NaOH. Specimens were then transferred to an AP substrate solution containing 0.5ml (10mg/ml) napthol AS-TR phosphate in N,N – dimethylforamide, 50ml tris-maleate buffer pH 8.3, plus 40mg diazonium salt (fast blue), and incubated for 1 hr. Specimens were cleared in trypsin and H2O2 as described above, and stored in 80% glycerol.
Whole-mount TRAP and AP staining provided a precise and generalized view of bone remodeling in zebrafish. TRAP activity was detected both in osteoclast cells and on bone surfaces that had undergone bone resorption. In particular, TRAP activity was observed in and around regions expected to undergo continuous remodeling, including cranial sutures, fins, scales and sites of muscle and ligamentous attachment (Fig. 1A–D). Similarly, AP staining labeled osteoblast cells with high precision, and was observed in regions predicted to be undergoing ossification and bone growth (Fig. 1E, F).
Specimens were mounted in 80% glycerol and photographed with a Zeiss Axiocam digital imaging system mounted to an M2 Bio stereomicroscope (Zeiss). The relative percent of TRAP and AP staining in the lower jaws of aceti282a/+ and homozygous wild-type individuals was measured using the histogram tool in Adobe Photoshop 7.0, and compared via one-way ANOVA.
WISH analyses were performed on staged embryos using an adapted protocol (Sagerstrom et al., 1996; Thisse et al., 1993). Juvenile and adult zebrafish were collected at different stages, fixed in 4% PFA in phosphate buffered solution containing 0.1% Tween 20, and dehydrated in methanol. Specimens were rehydrated, digested with 10μg/ml Proteinase K, and refixed in 4% PFA. Prehybridization was performed at 65°C for 3 hrs in 50% formamide, 5X SSC, 0.1% Tween-20, 4.6mM citric acid, 50μg/ml Heparin, and 500μg/ml tRNA. Anti-sense riboprobe was synthesized from fgf8 cDNA (Reifers et al., 1998) and labeled with digoxigenin (Roche Applied Science, Penzberg, Germany). Labeled riboprobe was added directly to pre-hybridization mix and allowed to incubate overnight at 65°C. Washes were performed the following day at 65°C in graded solutions from 100% hybridization mix to 100% 2X SSC, and then in 0.2X SSC. Specimens and α-dig antibody were pre-blocked at room temperature for at least 3 hrs in a solution containing 1 part 10% Boehringer blocking reagent dissolved in 1M Maleic Acid, 1 part lamb serum, and 3 parts filtered Maleic Acid Buffer (MAB: 100mM Maleic Acid, 150mM NaCl, 0.1% Tween 20, 7.9g/L NaOH, pH to 7.5). After pre-blocking, specimens were transferred to the α-dig antibody solution and blocked overnight at 4°C. The following morning specimens were washed first in MAB, and then in AP buffer solution (60mM Tris HCL pH to 9.5, 60mM NaCl, 30mM MgCl2, and 0.1% Tween-20). Staining was achieved with NBT and BCIP in AP buffer. After staining, specimens were refixed in 4% PFA, dehydrated in 100% methanol overnight to remove background staining, and stored in 80% glycerol.
Mandibular positioning relative to the midline was measured in aceti282a/+ adult zebrafish and their homozygous wild-type siblings. A significant difference was observed in the deviation of the mandibluar symphasis from the cranial midline in aceti282a/+ (n=52) and wild-type (n=45) zebrafish (F-ratio = 8.55, p = 0.0052). On average, deviation in aceti282a/+ zebrafish was twice as much as that observed in wild-type zebrafish (90 and 46 μm, respectively). Sidedness of mandibular asymmetry was determined by comparing heterozygous and homozygous phenotypes that fell outside the 95% confidence interval for each genotype. This comparison was used to control for the effects of extreme phenotypes. The average deviation from the midline was 9.2 μm to the left side (i.e., the left side was smaller than the right) in wild-type zebrafish (n=11), which was significantly different from the heterozygous phenotype where the average deviation of the mandibular symphasis was 128.8 μm to the left side (n=10) (F-ratio = 4.07, p = 0.05). Sidedness of mandibular asymmetry in aceti282a/+ zebrafish was confirmed using a Chi-square Goodness of Fit Test. Of 52 aceti282a/+ animals analyzed, 10 (19%) exhibited mandibular asymmetries outside the 95% confidence interval (Fig. 2A–C). Eight of the 10 (80%) asymmetric aceti282a/+ animals presented mandibles skewed to the left side, which is significantly more than what would be expected by chance (X2 = 3.6, df = 1, p = 0.05). These data are consistent with mandibular asymmetry having a directional (R>L) bias in aceti282a/+ zebrafish.
Opercle asymmetry was measured by comparing left and right opercle centroid sizes (CS). A significant difference was observed in opercle asymmetry in aceti282a/+ (n=50) and homozygous wild-type (n=48) zebrafish (F-ratio = 3.72, p = 0.058). Sidedness of opercle asymmetry was examined by comparing heterozygous and homozygous asymmetries that fell outside the 95% confidence interval for each genotype. The difference between the left and right sides of the opercle was skewed slightly to the right (R>L) in wild-type animals (n=18, L-R = −0.05), which is significantly different from the left-side bias observed in aceti282a/+ animals (n=14, L-R = 0.230, F-ratio = 3.49, p = 0.071). Sidedness of opercle asymmetry in aceti282a/+ zebrafish was confirmed using a Chi-square Goodness of Fit Test. Of 50 aceti282a/+ animals analyzed, 14 (28%) exhibited opercle asymmetries that exceeded at least one 95% confidence interval. Of these, 11 presented cases where the left opercle was larger than the right opercle, which is significantly more than what would be expected by chance (X2 = 4.6, df = 1, p = 0.05 (Fig. 3). The frequency (e.g., 28%) and bias (e.g., L>R) of opercle size asymmetry in aceti282a/+ zebrafish is consistent with observations in 7 day post fertilization (dpf) homozygous recessive aceti282a mutants (Albertson and Yelick, 2005), where 28% exhibited pharyngeal bone asymmetries characterized by skeletal elements (typically the opercle) missing from one side of the pharynx. Sixty percent of asymmetric aceti282a homozygous recessive mutants were missing bones from the right side of the head. Thus, whereas aceti282a mutants exhibited an asymmetry in the presence/absence of the opercle, aceti282a heterozygous zebrafish exhibited an asymmetry in the size of the opercle. Both asymmetries have a directional (L>R) bias. Notably, the sidedness of opercle and mandibular asymmetries was different in aceti282a/+ animals. Whether this observation speaks to an independent or inversely dependent origin of these defects is a question to be investigated in future studies.
Opercle defects in aceti282a/+ animals mainly affected the relative width of the element (Fig. 3D). This defect, however, was only observed in a percentage of aceti282a/+ animals (16/50 = 32%) and was often (but not always) asymmetrically expressed. Irregular opercle shapes observed in aceti282a/+ zebrafish were distinct from those in aceti282a homozygous recessive mutants (Fig. 3E–I). Hyoid arch dermal bones in aceti282a larvae were typically either missing, small or generally misshaped (n=194). A fraction (~10%) of mutants, however, exhibited more revealing bone phenotypes including fused dermal elements (Fig. 3F), enlarged “opercle-gain” morphologies (Fig. 3G) (Kimmel et al., 2003), and homeosis with both the opercle adopting the shape of the branchiostegal ray (Fig. 3H), and the branchiostegal ray adopting the shape of the opercle (Fig. 3I). These phenotypes are consistent with defects in DV patterning of the hyoid arch (Kimmel et al., 2003). Furthermore these bone defects were only observed in aceti282a homozygous recessive larvae - heterozygous larvae exhibited normal dermal bone shapes, suggesting separate roles for fgf8 during pre- and post-larval opercle development.
Cranial suture patterns were examined in aceti282a/+ and homozygous wild-type animals. Specifically, we measured the deviation of interfrontal and sagittal sutures from the cranial midline. In wild-type zebrafish, neither the interfrontal or sagittal sutures deviated significantly from the cranial midline (n=25). In aceti282a/+ zebrafish, however, these sutures exhibited considerable divergence from the cranial midline (n=52) (Fig. 2A, D, E). This difference in suture morphology was statically significant for both the interfrontal and sagittal sutures (F-ratio = 20.8, p <0.0001, and F-ratio = 8.36, p = 0.006, respectively). The coronal and lambdoid (transverse) sutures appeared to be unaffected in aceti282a/+ animals.
Aceti282a/+ zebrafish also exhibited aberrant craniofacial shape. Landmark-based GM analysis revealed significant differences in the length, height and overall geometry of the craniofacial skeleton between aceti282a/+ animals and their wild-type siblings. Fourteen landmarks on the oral jaws, suspensorium, pectoral girdle, and skull were used to describe craniofacial architecture (Fig. 4). A relative warp analysis was performed to identify major axes of shape variation. Aceti282a/+ zebrafish were significantly different from wild-type siblings along principal component (PC) axes 1 and 3 (F-ratio = 8.38, p = 0.006; F-ratio = 9.94, p = 0.003, respectfully), which collectively accounted for 37% of the total variation in craniofacial shape (Fig. 4A). Heterozygous and wild-type animals did not differ in their loadings on PC 2, which explained 17% of the variance and mainly defined the rotation of the upper jaw relative to the skull. This variation was artificially induced due to varying degrees of jaw adduction among samples, and is therefore not biologically meaningful. Deformation of shape along the axis that discriminates aceti282a/+ and wild-type siblings revealed specific differences in the shape of the craniofacial complex. Of particular interest is the deviation in the shape of the frontal bone in aceti282a/+ zebrafish (purple triangle, Fig. 4B), which is expanded both in length and height relative to the posterior portion of the skull (green line, Fig. 4B). Aceti282a/+ zebrafish also exhibited concomitant changes in the configuration of the upper jaws (blue triangle, Fig. 4B), and in the relative height of the skull (yellow line, Fig. 4B).
In addition to specific defects in craniofacial shape, aceti282a/+ zebrafish also exhibited more generalized bone defects, including irregular and ectopic bone formation. These defects seemed to be limited to the mandible (Fig. 5E), upper jaws, and anterior region of the skull, but until a careful dissection of the articulated craniofacial skeleton is performed we cannot rule out the possibility that other bones might also be affected. The appearance of ectopic bone in aceti282a/+ animals is consistent with our previous observations of homozygous recessive aceti282a mutants, where ectopic bones and cartilages were observed around the developing mandible in approximately 30% of mutants (Albertson, unpublished data).
The intricate configuration of the oral jaw apparatus in bony fishes is constrained by the mechanical requirements of respiration and feeding. These actions impose forces on the craniofacial skeleton that serve to continuously shape and remodel the oral jaw apparatus for optimal performance as the animal grows. We therefore hypothesized that the bone defects observed in aceti282a/+ zebrafish might relate to the mechanisms that underlie bone growth and remodeling.
Tartrate-resistant acid phosphatase (TRAP) and alkaline phosphatase (AP) activities were assessed using a whole-mount procedures to identify global patterns of bone remodeling in adult zebrafish (Fig. 1). We observed increased levels of both TRAP (n=20) and AP (n=20) in the jaws of aceti282a/+ zebrafish relative to their homozygous wild-type siblings (n=20 each TRAP and AP). TRAP levels were ubiquitously elevated in the mandibles of aceti282a/+ animals (Fig. 5A and B), whereas elevated AP levels were restricted to distal regions of the jaw (Fig. 5C and D). Elevated levels of TRAP and AP observed in the jaws aceti282a/+ animals were consistent with the frequent appearance of irregular and ectopic bones (Fig. 5E). TRAP and AP activities were, on average, two-fold greater in aceti282a/+ zebrafish than in wild-type zebrafish (Fig. 5F, F-ratio = 30.6, p < 0.001, and G, F-ratio = 7.24, p = 0.01). Similar increases in TRAP and AP levels were noted for the skull (data not shown). These data suggest that both osteoclast and osteoblast activities are elevated in aceti282a/+ animals, consistent with fgf8 functioning as a negative regulator of bone turnover in zebrafish.
Zebrafish development is characterized by extensive allometric growth, making it an excellent model with which to study skeletal remodeling (Witten et al., 2001). However, this attribute also means that skeletal development is particularly dynamic in zebrafish, characterized by shifts in the appearance and location of osteogenic cells depending on the size of the animal (Witten et al., 2001; Albertson, unpublished data). It is also clear that Fgfs play dynamic roles during skeletal development (Colvin et al., 1996; Deng et al., 1996; Govindarajan and Overbeek, 2006; Sobue et al., 2005; Valta et al., 2006; Yu et al., 2003). In this study, we present a characterization of bone remodeling activities in adult aceti282a/+ zebrafish, providing insights into Fgf8 functions at this developmental stage. Additional ongoing studies of zebrafish at multiple developmental stages are currently being performed to define the roles of Fgf8 signaling in bone formation and remodeling throughout development.
Fgf8 was expressed in discrete anatomical domains of the mature craniofacial skeleton. Fig. 6 illustrates fgf8 expression in the head of a 6 month-old zebrafish. Expression was clearly seen in the upper jaw apparatus (black arrow, Fig. 6A), opercular apparatus (white arrow, Figs. 6A, and 6B), cranial sutures (inset, Fig. 6A), and lower jaw (Fig. 6C). Fgf8 expression around the upper jaws was associated with the premaxilla, maxilla, kinethmoid, and the ethmoid region of the skull. This region of the craniofacial skeleton is a functional “hot spot” comprised of an intricate configuration of bones and ligaments that participate in action of upper jaw protrusion (Hernandez, 2000; Otten, 1983). Forces associated with this action, and imposed on the upper jaws, are likely to play important roles in the growth and remodeling of this region. Fgf8 expression associated with the opercle was most pronounced at the location where the opercle articulates with the skull and along the ventral/proximal edge of the bone (arrow, Fig. 6B). Fgf8 was also expressed in the sagittal cranial suture (inset, Fig. 6A), and along the ventral surface of the mandible (Fig. 6C). These expression patterns are consistent with defects observed in aceti282a/+ zebrafish, including upregulated osteoblast and osteoclast activities, suggesting a causal relationship between fgf8 deficiency and observed skeletal defects.
Recent efforts to characterize semi- and non-lethal craniofacial mutants in zebrafish have made important contributions to the study of skeletogenesis (Fisher and Halpern, 1999; Fisher et al., 2003; Draper et al., 2003; Elizondo et al., 2006; Albertson and Yelick, unpublished data). Fisher and colleagues (2003), for example, performed a large-scale ENU-mutagenesis screen for skeletal dysplasia in zebrafish. Using radiography to screen F1 adult zebrafish for dominant skeletal phenotypes, investigators identified the chihuahuadc124 (chi) mutant, which exhibited extensive skeletal defects and mapped to an interval that contained collagen type 1a (col1a1) (Fisher et al., 2003). Defects observed in chidc124 mutants were similar in presentation to humans with osteogenesis imperfecta (OI), a skeletal dysplasia typically caused by mutations in either COL1A1 or COL1A2 (Benusiene and Kucinskas, 2003). A more recent study by Elizondo and coworkers (2006) characterized the zebrafish nutriaj124e2 mutant, which was also identified in an ENU mutagenesis screen for postembryonic phenotypes. The nutriaj124e2 mutation mapped to the melastatin-type transient receptor potential (TRPM) channel, trpm-7 (Elizondo et al., 2006). During embryonic and early larval development nutriaj124e2 mutants were similar in size and appearance to wild-type zebrafish, but as adults they exhibited severe growth retardation, skewed body proportions, and a range of skeletal defects (Elizondo et al., 2006). Of particular interest is the altered sequence and timing of skeletal ossification in nutriaj124e2 mutants. An elegant statistical analysis, using a combination Wilcoxon test and logistic regression analysis, showed that nutriaj124e2 mutants exhibited significantly accelerated rates of endochondral ossification and correspondingly delayed rates of intramembranous ossification (Elizondo et al., 2006), revealing a complex and vital role for trpm-7 in the regulation of post-embryonic skeletogenesis.
The application of quantitative and statistical analyses to experimental embryology has also facilitated an understanding of the molecular mechanisms contributing to the development of skeletal shapes (Albertson and Yelick, 2004). For example, Kimmel et al. (2003) characterized a phenotypic series of opercle and branchiostegal ray morphologies associated with residual levels of the secreted peptide Endothelin-1 (Edn1). When Edn1 was severely reduced, both the opercle and branchiostegal rays were absent. However when Edn1 levels were only mildly reduced, the branchiostegal rays were absent but the opercle was expanded in size and altered in shape, resulting in what the authors referred to as an “opercle-gain” phenotype. These loss of function analyses, combined with the ventral pharyngeal arch expression domain of edn-1 mRNA (Miller et al., 2000), led the authors to posit an Edn1 morphogenic gradient model of hyoid bone patterning where levels of Edn1 are related to both the DV positioning and size of developing bones. The ventrally positioned branchiostegal rays are closer to the source of Edn1, and therefore more sensitive to its reduction. Lowering Edn1 levels decreases the slope of the gradient leading to a loss of the branchiostegal rays, ventral translocation of the opercle, and expansion of the region of opercle specification (Kimmel et al., 2003).
Hyoid arch bone phenotypes in aceti282a larvae were similar to those in edn-1 mutants as well as to other mutants that may function along the edn-1 genetic pathway (e.g., she, stu and hoo) (Miller and Kimmel, 2001). These hyoid bone morphologies can be grouped by their putative developmental origin, including fusion of the opercle and posterior branchiostegal ray (Fig. 3F), “opercle-gain” (Fig. 3G), and varying degrees of homeosis (Fig. 3H and I). These data are consistent with fgf8 functioning to pattern the hyoid arch along the DV axis. Whether fgf8 operates within or parallel to the edn-1 pathway remains to be investigated. Notably, opercle bone morphologies in aceti282a larvae were distinct from those observed in aceti282a/+ adults, supporting distinct roles for fgf8 in pre- and post-larval opercle development. During early larval development fgf8 appears to be necessary for proper DV patterning of the hyoid arch. During post-larval development fgf8 mRNA is expressed along the ventral-proximal surface of the opercle, and fgf8 haploinsufficiency results in defects specific to this region, implicating fgf8 as a regulator of opercle bone growth.
Zebrafish cranial sutures are similar in anatomy and development to those in mammals (Quarto and Longaker, 2005), making them a valuable model for the study of human diseases that affect suture formation. Craniosynostosis is a medical condition that involves the premature fusion of the cranial sutures, and affects 1 in 2500 live births (Morriss-Kay and Wilkie, 2005). Mutations have been identified in both syndromic and non-syndromic cases of craniosynostosis, the majority of which occur in fibroblast growth factor receptor (fgfr) types 1–3 (reviewed by, Morriss-Kay and Wilkie, 2005). These mutations also give rise to several related syndromes, characterized by additional defects in bone formation and growth. Given the genetic epidemiology of craniosynostoses in humans, it is not surprising that Fgf signaling figures prominently in the development and maintenance of cranial sutures. In mammals, as the skull vault grows, osteoblasts at the leading edges of skull bones secrete Fgf ligands, which activate receptors in adjacent undifferentiated mesenchyme. Low levels of Fgf lead to the activation of Fgfr2 and subsequent proliferation of osteoprogenitor cells. As Fgf levels increase, Fgfr2 is down regulated and Fgfr1 is up regulated, leading to a switch from cell proliferation to the overt differentiation of osteoblasts (Iseki et al., 1999). Proper growth of the cranial vault therefore depends on the balance between Fgf-mediated cell proliferation and differentiation along the cranial sutures (Morriss-Kay and Wilkie, 2005).
It remains unclear to what degree zebrafish share the molecular program used to regulate cranial development and growth in mammals. Indeed, the very homology of bones comprising the cranial vault across vertebrate classes is still debated (reviewed in Hanken and Gross, 2005). Nonetheless our results suggest that fgf8 plays a role in cranial suture development in zebrafish. Specifically, the aberrant suture patterning and the expanded frontal region of the skull observed in aceti282a/+ zebrafish are consistent with craniosynostosis phenotypes. If cranial suture patterning in zebrafish proves to be similar to that of mammals, continued work in this developmentally tractable system will greatly facilitate our efforts to understand the pathophysiology of craniosynostosis disorders, and the derivation of therapeutically relevant treatments for these syndromes.
Our data support a molecular genetic basis for directional craniofacial asymmetries, and fit into a larger body of literature seeking to understand the mechanisms that regulate laterality in vertebrates. Directional asymmetries differ from fluctuating asymmetries in that the former produces structures that are biased to either the right- or left-hand side within a population, whereas the later has no directional bias (right- or left-hand forms occur with equal frequency). Directional asymmetries are generally thought to have a strict genetic basis and develop as a consequence of an internal asymmetric developmental genetic mechanism (i.e., Kawakami et al., 2005). Conspicuous examples of directional asymmetries in vertebrates include the orientation of the heart and the positioning of the visceral organs. Fluctuating asymmetries, on the other hand, do not have a genetic basis and are thought to arise from random perturbations in developmental systems. A multitude of factors have been shown to influence levels of fluctuating asymmetries including environmental quality, hybridization among species, and inbreeding (reviewed by Palmer and Strobeck, 1997). While the mechanisms that regulate the asymmetric development of the brain, heart and viscera have been extensively studied (reviewed by Levin, 2005), we know comparatively little about how directional asymmetries have evolved in normally paired structures (i.e., the narwal’s ‘tusk’, the ears of owls, and the whole body in flatfish). Are the same asymmetric cues being co-opted for the asymmetric development of normally paired structures? Or have other asymmetric signals evolved?
We propose that the vertebrate body plan possesses a fundamental L-R handedness (Albertson and Yelick, 2005), and that the same molecular machinery that establishes the L-R axis during heart and visceral organ development also influences the development of a host of other structures. Our characterization of the zebrafish aceti282a mutant demonstrates that deficiency in this growth factor results in conspicuous asymmetric development of the pharyngeal skeleton (data presented above, Albertson and Yelick, 2005). It has also recently been shown that zebrafish lacking retinoic acid (RA) exhibit asymmetric somite development (Kawakami et al., 2005). In both fgf8 and RA deficient animals, asymmetric defects were linked to the presence and proper function of Kupffer’s vesicle, a ciliated node that mediates laterality information in zebrafish (Albertson and Yelick, 2005; Kawakami et al., 2005). These observations are consistent with a model where Fgf and RA signaling act to buffer the zebrafish embryo from the lateralizing effects of Kupffer’s vesicle, leading to bilaterally symmetric segmentation and development of the pharynx and somites (Albertson and Yelick, 2005; Kawakami et al., 2005). Under this model the wild-type function of a handful of factors are necessary for the proper symmetric development of paired structures in an otherwise asymmetric environment.
Additional evidence for a hypothesized asymmetric body plan comes from a range of observations in a variety of disciplines. From a clinical perspective, numerous human birth defects are characterized by craniodental asymmetries, including hemifacial microsomia, Treacher-Collins syndrome, unilateral clefting, Saethre-Chotzen syndrome, Russell-Silver syndrome, and hemifacial hypertrophy. In each of these, spontaneous or inherited defects in the normal developmental program lead to a break in symmetry. Experimentally induced developmental defects can also lead to pronounced asymmetric development of normally paired structures. As discussed above, experimental knock-down of Fgf and RA signaling in zebrafish leads to conspicuous directional asymmetries in the pharynx and somites, respectively. In addition, mice lacking the gene pitx1 exhibit an asymmetric reduction in hindlimb morphology with the right-side typically more affected than the left (Marcil et al., 2003). This asymmetric phenotype is similar in presentation to patterns of pelvic limb reduction in natural populations of threespine stickleback (Gasterosteus), ninespine stickleback (Pungitius) and Florida manatees (Trichechus manatus latirostris) (Shapiro et al., 2006). Moreover in both stickleback genera pelvic reduction mapped to pitx1 and was associated with alternate pitx1 expression patterns, suggesting a common genetic mechanism underlying the independent evolution of limb reduction (Shapiro et al., 2005; 2006). “Bookend” (mirror image) phenotypes observed among monozygotic twins, including hair whorl direction, hand preference, vision and hearing defects, and dental patterning, provide another line of evidence for a fundamentally asymmetric body plan that is established very early in development (Levin, 1999). Heritability estimates and mapping studies in humans and mice support a genetic basis for normal directional craniofacial asymmetries (Cassidy et al., 1998; Leamy et al., 2003). Asymmetries in skeletal traits are also a common feature among parents of children with unilateral clefting (AlEmran et al., 1999; McIntyre and Mossey, 2003; Yoon et al., 2003). In fact, the direction of certain facial asymmetries in parents was predictive of the direction of unilateral clefting in their children. This observation suggests that the same loci that underlie natural variation in facial asymmetry are also responsible for clinical asymmetries (Yoon et al., 2003). Finally, several vertebrate taxa have accentuated craniodental asymmetries along the left-right axis (i.e., flatfish, narwals and owls; reviewed by Albertson and Yelick, 2005), consistent with natural selection targeting an inherently asymmetric developmental program.
Taken together, these data suggest that the vertebrate body plan is only superficially bilaterally symmetrical, and that handedness is inherent to most if not all paired structures. Whether these latent asymmetries are a shared derived character of vertebrates, or an evolutionary consequence of our shared ancestry with echinoderms, which exhibit many peculiar left-right asymmetries that are regulated by many of the same mechanisms as in vertebrates (Cooke, 2004; Duboc et al., 2005, Hibino et al., 2006), remains a topic for debate. What is clear is that further elucidation of the molecular signaling pathways regulating these processes in the zebrafish model, are likely to facilitate our understanding of these basic developmental processes, and to provide a sound basis for the development of educated molecular-based strategies to predict, prevent and/or correct craniofacial asymmetries in humans.
We wish to thank members of the Albertson and Yelick laboratories for thoughtful comments on earlier versions of this manuscript, and Seija Cope for expert zebrafish husbandry. PCY is supported by NIDCR Grants DE12024 and DE12076.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.