Knockdown of hspg2 is associated with mandibular jaw joint fusion and neural crest cell dysfunction in zebrash

Background: Heparan sulfate proteoglycan 2 (HSPG2) encodes for perlecan, a large proteoglycan that plays an important role in cartilage formation, cell adhesion, and basement membrane stability. Mutations in HSPG2 have been associated with Schwartz-Jampel Syndrome (SJS) and Dyssegmental Dysplasia Silverman-Handmaker Type (DDSH), two disorders characterized by skeletal abnormalities. These data indicate a function for HSPG2 in cartilage development/maintenance. However, the mechanisms by which HSPG2 regulates cartilage development are not completely understood. Here, we explored the relationship between this gene and craniofacial development through morpholino-mediated knockdown of hspg2 using zebrash. Results: Knockdown of hspg2 resulted in abnormal development of the mandibular jaw joint at 5 days post fertilization (DPF). We surmised that defects in mandible development were a consequence of neural crest cell (NCC) dysfunction, as these multipotent progenitors produce the cartilage of the head. Early NCC development was normal in morphant animals as measured by distal-less homeobox 2a (dlx2a) and SRY-box transcription factor 10 (sox10) expression at 1 DPF. However, subsequent analysis at later stages of development (4 DPF) revealed a decrease in the number of Sox10 + and Collagen, type II, alpha 1a (Col2a1a) + cells within the mandibular jaw joint region of morphants. Concurrently, morphants showed a decreased expression of NK3 homeobox 2 (nkx3.2), a jaw joint molecular marker at 4 DPF. Conclusions: Collectively, these data suggest a complex role for hspg2 in jaw joint formation and late stage NCC differentiation. a Zeiss Stereo Microscope tted with Zen Software. Statistical analysis was performed using a Fisher’s exact test. For quantitative polymerase chain reaction (qPCR), RNA was isolated from a pool of embryos at the indicated time point using Trizol (Fisher) according to manufacturer’s protocol. Reverse transcription was performed using a Verso cDNA Synthesis Kit (Fisher) and total RNA was normalized across all samples. PCR was performed using an Applied Biosystem’s StepOne Plus machine with Applied Biosystem’s software. Sybr green (Fisher) based primer pairs for each gene analyzed are as follows: nkx3.2 sox10 ACGCTACAGGTCAGAGTCAC, sox10 rev ATGTTGGCCATCACGTCATG, rpl13a TCCCAGCTGCTCTCAAGATT, and rpl13a rev TTCTTGGAATAGCGCAGCTT. Analysis was performed using 2 ΔΔct indicating relative mRNA expression for each gene. For each biological replicate, RNA was isolated from a pool of injected embryos. The morphant pooled group gene expression is expressed relative to the expression from a pooled group of individuals injected with random control morpholinos. Each qPCR analysis was performed in biological replicate, which can be dened as two independent occasions with independent parents and an independent injection. For each biological replicate, the qPCR was performed in technical triplicates, however, the error bars on the graphs depicted are the standard error between biological replicates. Statistical analysis of messenger RNA (mRNA) expression was performed using a Student t-test on biological replicates.


Background
Mutation of the HSPG2 gene causes Schwartz-Jampel Syndrome (SJS) and Dyssegmental Dysplasia Silverman-Handmaker Type (DDSH) (1)(2)(3). SJS is a recessive disorder characterized by muscle stiffness (myotonia) and chondrodysplasia. DDSH is a less frequent but a more severe recessive disorder characterized by reduced joint mobility, severe limb shortening, and short stature (4). Both disorders have common clinical manifestations that include reduced stature, bowing of the long bones, and facial dimorphism (4,5). Interestingly, SJS is generally associated with loss of function mutations in HSPG2 that vary from missense mutations to splice site mutations. These mutations are spread throughout the HSPG2 protein, with those located closer to the C-terminus being more readily tolerated (5). Conversely, DDSH is caused by homozygous null mutations (frameshifts and point mutations) wherein there is an absence of functional protein, likely due to degradation (5). Despite mutation heterogeneity underlying SJS and DDSH, craniofacial abnormalities such as micrognathia are very common phenotypes in diagnosed patients (5). These manifestations of chondrodysplasia suggest that HSPG2 plays a role in regulating craniofacial development.
The function of HSPG2 during chondrogenesis is not completely understood, but is likely related to the various domains within the HSPG2 protein and the vast array of interactions between HSPG2 and the extracellular matrix (ECM) to promote signal transduction and stability (6,7). HSPG2 encodes for perlecan, a large proteoglycan that consists of ve domains, each of which have a unique function (8).
Four of these domains (domains II, III, IV, and V) have repeats homologous to low-density lipoprotein receptor (domain II), laminins (domains III and V), and immunoglobulins (domain IV), but domain I remains unique to perlecan (8). The N-terminal domain I possesses three ser-asp-gly motifs that serve as attachment sites for glycosaminoglycan (GAG) side chains like heparan sulfate (HS) and chondroitin sulfate (CS) (8-10). These side chains facilitate interaction with growth factors such as broblast growth factor-2 (FGF-2), vascular endothelial growth factors (VEGFs), and bone morphogenetic proteins (BMPs) (6,10,11) upon secretion of perlecan into the ECM. In addition, HS has been shown to speci cally interact with ECM proteins such as bronectin, laminin, and collagens I, II, III, IV, and V, suggesting perlecan and its side chains mediate cell adhesion and basement membrane stability (12).
As it relates to the chondrocyte phenotypes in SJS and DDSH, previous studies, both in vitro and in vivo, have established that disrupting HSPG2 expression leads to abnormal chondrocyte proliferation and disorganized columnar arrangement (13,14). This is signi cant as chondrocytes are the primary cells found in cartilage and are known to secrete a specialized ECM containing glycoproteins and proteoglycans, like perlecan, to maintain structural integrity (15). During craniofacial development, chondrocytes arise from neural crest cells (NCCs), a multipotent progenitor cell population that forms at the dorsal end of the neural tube upon neural tube closure. There are four populations of NCCs, but only cranial NCCs (CNCCs) migrate to the pharyngeal arches and develop into cartilage and bone, making them vital to proper craniofacial development (16-18). These CNCCs are regulated by growth factors like BMP, which are vital for mandibular morphogenesis, and FGFs, which are essential for CNCC differentiation (19)(20)(21).
The development of craniofacial bone is due to both endochondral and intramembranous ossi cation. In endochondral ossi cation, cells differentiate into chondrocytes and bones develop from a cartilaginous base. This form of ossi cation necessitates that chondrocytes are present in growth plates where they are arranged into zones of rest, proliferation, and hypertrophy (15,22). Perlecan is present in these cells and additionally in the synovial joints of the body, where it mediates differentiation through regulatory sox9 expression (23). Synovial joints possess uid-lled cavities that separate the surface of bones, likely developed by the upregulation of hyaluronic acid (or hyaluronan), which, in large quantities, promotes cellular separation, cavitation, and the formation of a joint (24)(25)(26). Studies focused on osteoarthritis have shown that perlecan side chains have the ability to bind to hyaluronan and can potentially mediate joint function through regulating the release of BMP-2 (27,28). Despite this, there are still considerable gaps in the literature on the mechanisms by which perlecan regulates joint development.
To begin to understand these mechanisms, we performed morpholino mediated knockdown using the zebra sh model. Apart from general advantages like a high fecundity rate, extrauterine development, and quick maturation, the zebra sh is an ideal organism to study craniofacial development due to the relative simplicity of the cartilaginous structures of the head and face, all of which are readily visible with stains like Alcian blue (29,30). Additionally, the zebra sh shares a high degree of genetic similarity to humans and successful studies have been performed using genetically manipulated sh (30). Thus, we performed transient knockdown of zebra sh hspg2 to determine the CNCC defects present during craniofacial development. Knockdown of hspg2 resulted in abnormal mandibular jaw joint formation and disrupted late stage differentiation of CNCCs, with little to no effect on early stage CNCC development. Collectively, our results suggest that hspg2 is essential for joint formation in the developing zebra sh.

Results
Morpholino-induced knockdown of hspg2 is associated with craniofacial phenotypes It has previously been reported that hspg2 mRNA is expressed ubiquitously throughout the head, eyes, and somites of the zebra sh. Additional immunohistochemistry staining revealed positive expression of perlecan in these regions from embryos at 2-3 hours post fertilization to 5 days post fertilization (DPF) (31,32). Based on this expression pattern and previous murine studies establishing that mutation of Hspg2 results in failure of the chondro-osseous junction of developing bones and craniofacial abnormalities (13), we hypothesized that morpholino mediated knockdown of the zebra sh hspg2 gene would cause craniofacial abnormalities.
We performed Alcian blue staining to detect craniofacial abnormalities in the developing cartilage. We measured the distance from the top of the ceratohyal to the tip of the Meckel's cartilage as a read out for mandibular truncation as previously described (33). Measurements of the distance between the top of the ceratohyal and Meckel's cartilage at 5 DPF ( Figure 1A, 1B, and 1C) showed that the injection of the hspg2 translation blocking morpholino caused a 7% truncation in the zebra sh mandible when compared to the random control group. The observed truncation was subtle in morphants and did not appear to be the consequence of a malformed Meckel's cartilage or defects in the development of the ceratohyal. However, upon higher magni cation, hspg2 morphants demonstrated an abnormal mandibular jaw joint between the Meckel's cartilage and palatoquadrate (the dorsal component of the mandibular arch), a phenotype that was not present in the random control group ( Figure 1A'-B', red arrowheads).
nkx3.2 expression is decreased in hspg2 morphants NK3 homeobox 2 (nkx3.2) was rst identi ed in the Drosophila melanogaster model (where it is known as bapx1) and is part of the NK family of homeobox genes (34). Homologues of the gene have been found in vertebrates and are expressed predominately in the rst pharyngeal arch (a developmental structure that gives rise to the mandible (arrow Figure 2A'), where the gene is essential for proper joint formation (34,35). Knockdown of nkx3.2 causes abnormal jaw joint phenotypes (primarily fusion phenotypes) in amphibians and zebra sh (34,36). Therefore, we utilized nkx3.2 expression as a marker of mandibular jaw joint development. In situ hybridization performed at 2 DPF (Figures 2A-B) demonstrated decreased expression of nkx3.2. nkx3.2 is primarily expressed in the pharyngeal arches, where we observed high expression in the 1 st and 5 th pharyngeal arches (Figure 2A'-B') (expressed posterior to the eye, annotated by the black arrow) and in the sclerotomal derivatives (36). Knockdown of hspg2 resulted in decreased expression of nkx3.2 in the pharyngeal arches when compared with the random control group (Figure 2A'-B'). Subsequent qPCR at 4 DPF con rmed a statistically signi cant decrease in nkx3.2 in morphants relative to the random control group ( Figure 2C).

Neural crest cells migrate normally in the absence of hspg2
Because defects in the number of and migration of NCCs are possible mechanisms by which craniofacial de cits may arise (33), we hypothesized that the craniofacial abnormalities present at 5 DPF might be due to early CNCC defects. To determine if hspg2 affects early CNCCs, we analyzed Tg(sox10:TagRFP) embryos at both the 18 somite (aligning with early NCC speci cation and migration) (37) and Prim-5 (corresponding with NCCs invading the pharyngeal arches) stages. Cells in the NCC lineage express SRYbox transcription factor 10 (sox10) at various stages of development where it maintains their survival, speci cation, and differentiation (38). Previous studies have used sox10 expression as a valid marker for visualizing NCCs during early developmental stages (33,39). Results revealed no discernable differences in the location or degree of RFP expression in hspg2 morphants relative to control ( Figure 3A-A' and 3B-B').

dlx2a expression in morphants is unaffected
We next analyzed the expression of distal-less homeobox 2a (dlx2a) at the Prim-5 stage in random control and hspg2 morpholino injected embryos to determine if CNCC speci cation occurs normally upon knockdown of hspg2. This homeobox gene is expressed in cranial neural crest cells migrating to the pharyngeal arches (40) and has been established as a marker of proper CNCC speci cation (33,39,41). In situ hybridization ( Figure 4A-B) revealed that there was no signi cant difference in the expression of dlx2a in morphants relative to random control injected. qPCR measurements performed at Prim-5 in both RC and MO groups validated the normal level of dlx2a expression in morphant animals ( Figure 4C). Collectively, these data suggest that early CNCC development was normal.
hspg2 knockdown affects cell numbers in jaw joint region Based on our results, which suggest that the early NCC lineage was not affected, we next hypothesized that hspg2 was mediating late stage CNCC differentiation and potentially mediating the defects observed at 5 DPF. To test this, we performed analysis of Sox10 + cells at 3 and 4 DPF using Tg(sox10:TagRFP) larvae. Chondrocytes across both random control and hspg2 morphant groups at 3 and 4 DPF had normal morphology and columnar arrangement. Chondrocytes to the left of the joint were more closely clustered together than those on the right and therefore, only cells from 3 rows to the left of the joint were easily discernable as opposed to 5 rows of identi able cells on the right of the joint.  Figure 5D). However, at 4 DPF ( Figure 6A-B), the number of Sox10 + cells were reduced relative to random control injected embryos ( Figure 6C-C') and the level of sox10 expression was approximately 50% of the control according to qPCR ( Figure 6D). Subsequent analysis of Col2a1a (collagen, type II, alpha 1a) + cells using the Tg (col2a1a:EGFP) transgenic reporter in the mandibular jaw joint region revealed similar chondrocyte morphology and arrangement as samples in the sox10 group. Cell counts ( Figure 7A-B) demonstrated a similar decline in numbers of EGFP+ cells in morphant animals at 4 DPF ( Figure 7C-C'). Collectively, these data show a progressive loss of differentiated NCCs between 3 and 4 DPF.

Discussion
Our analysis revealed a 7% mandibular truncation and an abnormal joint phenotype in animals with knockdown of hspg2. Deletion of Hspg2 in mice has been shown to cause truncated snouts, shorter and thicker mandibular structures, and at faces (13,42,43), but very little has been reported on joint phenotypes and how they pertain to hspg2 function. Although it is relatively novel, the idea of hspg2 mediating the mandibular jaw joint region is not unfounded. Similar to other synovial joints, the mandibular jaw joint contains a synovial capsule, which, in previous cell culture work using synovial cells, has been shown to express and require perlecan for proper development (23,44).
The mechanism by which perlecan mediates joint development is currently unknown. However, as discussed in the introduction, perlecan is a multi-domain protein with side chains that interact with various growth factors like BMP, WNT, and FGF; all of which are essential for neural crest development, chondrogenesis, and joint formation (43,44). De ciencies or abnormalities in the activation or level of such pathways may account for the decreased number of Sox10+ or Col2a1a+ cells at 4 DPF. Perlecan has also been found to bind to Ihh (Indian hedgehog) through its HS side chains, which in turn mediates the proliferation of chondrocytes (43). It should be noted however, that many of these effects and pathways have predominately been implicated in the joints of the appendicular skeleton and that development of the mandibular jaw joint could be different.
We further demonstrate that knockdown of hspg2 is associated with decreased numbers of Col2a1a + cells at 4 DPF. These data suggest that hspg2 has a function regulating CNCC differentiation, a nding that is supported by the number of Sox10 + cells at an equivalent time point. Our studies are supported by previous analysis in mice (Hspg2 -/-) that demonstrated abnormal arrangement and proliferation of chondrocytes in the appendicular skeleton (13,42). It must be noted however, that although these data support one another, the cells of the appendicular skeleton derive from a different germ layer (the mesoderm) than the cells of the craniofacial skeleton (the ectoderm). Both cell lineages give rise to cartilaginous structures, but the mechanisms by which each population differentiates are likely to be different, prompting further studies. One possible future direction of our work could be to determine the interplay between perlecan and FGF because perlecan binds to FGF-2, which increases the expression of sox9 in vitro (23,45). The protein output of Sox9 in turn is vital to chondrogenesis because it activates Col2a1 expression in mice (46). It is possible however, that there are various mechanisms underlying the function of hspg2 in joint development because the heparin side chains of perlecan are known to bind to collagen II (12) suggesting a direct function for perlecan in chondrogenesis. Interestingly, we also observed an initial increase in the number of Sox10+ cells at 3 DPF, which at the onset seems to counter the results observed at 4 DPF. However, this increase of cells could be due to a period of proliferation in chondrocytes followed by increased cell death between 3 and 4 DPF. Further studies in this area are warranted.
Knockdown of hspg2 was also associated with reduced expression of nkx3.2 at 4 DPF. These results, when understood in the context of the decrease of Col2a1a + cells found at 4 DPF, are supported by previous results performed in mesenchymal cell culture where nkx3.2 upregulates col2a1 by directly binding to the promoter (47). In this situation, diminished expression of nkx3.2 appears to be directly proportional to a decrease in Col2a1a+ cells and the differentiation of chondrocytes. It is not clear if hspg2 directly modulates nkx3.2 expression or if the decreased expression is simply the result of defects in the mandibular jaw joint, but studies performed in the chick have shown that Nkx3.2 and Sox9 cooperate to promote chondrogenic differentiation and serve as mediators of Sonic Hedgehog (Shh)induced chondrogenesis (48). This could be one of the mechanisms by which perlecan indirectly mediates the expression of nkx3.2 and it would prove to be a novel discovery. Recently, it was shown that nkx3.2 null animals are viable, making it possible to study this gene in relation to hspg2 without early lethality (49).
In this paper, we demonstrate zebra sh as an alternative animal model to study the role of hspg2 during craniofacial development. Induced knockouts in the murine model have resulted in embryonic lethality from mass hemorrhaging in the pericardial cavity and severe chondrodysplasia, both occurrences which can be temporarily circumvented in the developing zebra sh (31). To circumvent these limitations, three additional mouse models have been produced: the rst model lacks exon 3, causing loss of the 3 HS side chains (7), the second model is modeled after an SJS patient mutation where there is a G to an A substitution theorized to cause a misfolded protein (42,50), and the third is a model where early lethality is restored via tissue speci c expression of Hspg2 in chondrocytes (51).
While the rst two models are viable and can be used to examine adult skeletal phenotypes, the rst is centered around exploring the loss of only one domain and the second is mimicking more subtle phenotypes associated with SJS. Our project seeks to understand the role of perlecan in craniofacial development using a null phenotype, a feat not easily performed in a murine model. Zebra sh allow for this type of exploration because unlike mice, they are externally fertilized. This external fertilization enables the study of craniofacial development at early developmental stages, particularly with the use of transgenic sh to target speci c genes (52,53). The third model described restores early lethality using a chondrocyte promoter and consequently cannot be utilized to study chondrogenesis or craniofacial development (51). Zebra sh craniofacial development is conserved and the development of the viscerocranium, including the development of the pharyngeal arches, the migration and speci cation of NCCs, their differentiation, and signaling pathways involved have all been well characterized (29,54).
Additionally, because the zebra sh mandibular joint is a synovial joint which develops in a similar fashion to other vertebrate synovial joints, the mechanisms uncovered from this research could be translational to other areas (55).
Zebra sh are remarkably easy to manipulate genetically and have been used with great success in genetic studies. The zebra sh genome shares a high degree of genetic similarity with humans and thus, provides a manner in which to understand gene function and mechanisms (30). 70% of human genes have one zebra sh ortholog and 82% of the genes associated with disease also have at least one zebra sh ortholog (56). CRISPR mutagenesis has emerged as a manner of genetic manipulation readily adaptable to zebra sh (57) and future studies developing a germline non-sense mutant of hspg2 are a critical next step.
All work reported here has been completed by use of a single translation-blocking morpholino. While translation-blocking morpholinos are a simple and effective way in which to knockdown genes of interest, they have been associated with off target effects and non-speci c cell death. We did, however, utilize a random control morpholino to account for the possibility of morpholino-induced cell death, an endeavor that proved to be rather successful in previous studies (58,59). And while injection of heparin side chains is a potential rescue for the morphant phenotype we observed, there is the possibility that heparin sulfate co-injection would fail to rescue because a domain outside of domain I is essential for regulation of CNCCs. Therefore, such an experiment is unlikely to demonstrate a full rescue. Collectively, this points out a potential caveat to our work in that we study a morpholino derived phenotype. However, our data with one morpholino is supported by previous studies, including those completed using the murine model (13,42). In these studies, numbers of chondrocytes in the lateral skeleton are depleted, chondrocytes congregate abnormally, and mutant mice exhibit craniofacial abnormalities.
Our work contributes an additional angle to the role of hspg2 in skeletal development by examining the mandibular jaw joint region speci cally. We understand that additional morpholinos would help to substantiate our work, but our data is supported by previous studies, suggesting that what we observed is not a consequence of off-target effects. Nevertheless, future studies in a germline zebra sh mutant are required.

Conclusion
In summary, our results have created novel implications for the role of hspg2 in the development of the mandibular jaw joint, a region of the craniofacial skeleton for which development has not been well elucidated. Furthermore, connections between hspg2 and nkx3.2 have yet to be drawn in literature. Additional data concerning late CNCC differentiation raises pertinent questions about the fate of certain cells within different areas of the developing face. Although future studies to more cohesively understand the role of hspg2 in craniofacial development are needed, these data lay signi cant groundwork for future experiments in this eld and suggest that zebra sh are an acceptable model to study the function of hspg2 in craniofacial development.

Animal Care
For all experiments, embryos were obtained by crossing adult Tg(sox10:tagRFP), Tg(col2a1a:EGFP), or AB wildtype sh. Embryos were maintained in E3 embryo medium at 28˚C. All zebra sh were maintained at The University of Texas El Paso according to the Institutional Animal Care and Use Committee (IACUC) guidelines protocol 811689-5. All adult sh were obtained from the University of Colorado, Anschutz Medical Campus or the Zebra sh International Resource Center (ZIRC). Adult and larval zebra sh were euthanized and anesthetized according to guidelines from the American Veterinary Medical Association and approved IACUC protocols. For euthanasia, adults beyond the age of peak breeding age (>1.5 years old) were euthanized using a solution of 10g/L buffered solution of pharmaceutical grade MS 222. Fish were emerged in solution for 30 minutes at RT. All euthanized adults underwent secondary euthanasia with a cold ice bath (2-4 degrees C). Cessation of movement was indicative of euthanasia. Embryos (<7 days old) were euthanized using 1-10% sodium hypochlorite solution after being anesthetized in cold ice bath. For any genotyping and before xation, all sh, adults and larvae, were anesthetized using MS 222 (150mg/L for adults and 300mg/L for embryos). The degree of anesthesia was monitored by operculum movement of adults and cessation of movement for larvae.

Antisense oligonucleotide morpholino design and microinjection
Two antisense oligonucleotide morpholino sequences were designed in conjugation with Gene Tools LLC.
The rst was a translation blocking morpholino (MO) with the sequence 5'-TATCCTCGCCCCCATTTCTGCCAA-3', created to bind to the hspg2 translation start site and sterically knockdown perlecan translation in the developing larvae. The second was a random control morpholino with the sequence 5'-AAAAAAAAAAAAAAAAAAAAAAAA-3'. This random control morpholino was used to assure that the translation blocking MO microinjections were not causing any form of extraneous cell death as previously described (60).
The initial concentration gradient for the translation blocking morpholino was as follows: 0. For all experiments, the morphant experimental group is compared to either a random control group or wildtype non-injected. The randomized control morpholino has been shown through previous literature to have no associated phenotypes, indicating that it does not in uence nal results and therefore is the appropriate control group for comparison (58,59,61). Thus, for statistical analysis, comparisons were performed using a T-test between the random control group and the morpholino. MOs were injected into embryos at the one cell stage with a stock concentration of 0.10 µM (1.65 ng/embryo) and at a volume of 0.52 nL per embryo as explained above. Wildtype -non-injected larvae were used predominately as a baseline for fertilization rates.
Alcian Blue Staining and Imaging Zebra sh larvae (aged 5 DPF) were xed in 2% PFA in PBS, pH 7.5 for 1 hour at room temperature (RT). Samples were washed for 10 minutes with 100mM Tris pH 7.5/10mM MgCl2, stained with Alcian blue stain (pH 7.5: 0.4% Alcian blue (Anatech Ltd., MI) in 70% EtOH, Tris pH 7.5 (Fisher, MA), and 1 M MgCl2 (Fisher, MA)), and incubated overnight at RT. Samples were subsequently destained and rehydrated using an EtOH: Tris pH 7.5 gradient as previously described (39). Embryos were bleached (30% H2O2 (Sigma, St. Louis, MO), 20% KOH (Fisher, MA)) for 10 minutes at RT. Samples were washed twice for 10 minutes per wash in wash buffer (25% glycerol/0.1% KOH (Fisher, MA)) and stored at 4ºC in storage buffer (50% glycerol/0.1% KOH (Fisher, MA)) until imaged. The distance between the top of the Meckel's cartilage (the ventral component of the mandibular arch) and the top of the ceratohyal (a pharyngeal arch cartilage) was measured for each embryo as a method to analyze truncation of the Meckel's cartilage, which would be analogous to micrognathia as described previously (33). The nal measurement is the distance between the two structures and not the full length of the Meckel's cartilage. Distance measurements were performed using Zeiss software and all distances are in µM. For imaging, a representative sample of the two groups (hspg2 morphants and random control larvae) were dissected and viscerocranium was mounted on a glass slide with 100% glycerol. A Leica microscope was used to take high-resolution color images of each sample.
Whole mount in situ hybridization and quantitative real time polymerase chain reaction (PCR) Whole mount in situ hybridization was performed as described by Thisse and Thisse (62). Larvae were harvested and dechorionated at the indicated time point and xed in 4% paraformaldehyde (Electron Microscopy Sciences, PA) overnight at 4ºC. Larvae were then dehydrated using a methanol: PBS gradient and stored in 100% methanol overnight at -20°C. Embryos were rehydrated using a PBS:Methanol gradient, washed in PBS with 0.1% Tween 20 and permeabilized with proteinase K (10ug/ml) for the time indicated by Thisse and Thisse (62). Permeabilized larvae were prehybridized for 2 hours in hybridization buffer (HB) (50% deionized formamide (Fisher), 5X SSC (Fisher), 0.1% Tween 20 (Fisher), 50 µg/m heparin (Sigma), 500 µg/mL of RNase-free tRNA (Sigma), and 1M citric acid (Fisher). Larvae were then incubated overnight in fresh HB with probe (dlx2a and nkx3.2 at 127 ng) at 70°C. Samples were washed according to protocol, blocked in 2% sheep serum (Sigma) and 2 mg/ml bovine albumin serum (Sigma) for 2 hours at room temperature. Samples were then incubated with anti-DIG Fab fragments (1:10,000) (Sigma) overnight at 4°C. Samples were developed with BM purple AP substrate (Sigma) and imaged with a Zeiss Discovery Stereo Microscope tted with Zen Software. Statistical analysis was performed using a Fisher's exact test. For quantitative polymerase chain reaction (qPCR), RNA was isolated from a pool of embryos at the indicated time point using Trizol (Fisher) according to manufacturer's protocol.
Reverse transcription was performed using a Verso cDNA Synthesis Kit (Fisher) and total RNA was normalized across all samples. PCR was performed using an Applied Biosystem's StepOne Plus machine with Applied Biosystem's software. Sybr green (Fisher) based primer pairs for each gene analyzed are as follows: dlx2a fwd CCTCACGCAAACACAGGTTA, dlx2a rev TGTTCATTCTCTGGCTGTGC, nkx3.2 fwd GCAGATTTAGCGGACGAGAC, nkx3.2 rev GCTTCAACCACCAGCGTTAT, sox10 fwd ACGCTACAGGTCAGAGTCAC, sox10 rev ATGTTGGCCATCACGTCATG, rpl13a fwd TCCCAGCTGCTCTCAAGATT, and rpl13a rev TTCTTGGAATAGCGCAGCTT. Analysis was performed using 2 ΔΔct indicating relative mRNA expression for each gene. For each biological replicate, RNA was isolated from a pool of injected embryos. The morphant pooled group gene expression is expressed relative to the expression from a pooled group of individuals injected with random control morpholinos. Each qPCR analysis was performed in biological replicate, which can be de ned as two independent occasions with independent parents and an independent injection. For each biological replicate, the qPCR was performed in technical triplicates, however, the error bars on the graphs depicted are the standard error between biological replicates. Statistical analysis of messenger RNA (mRNA) expression was performed using a Student t-test on biological replicates.

Confocal Imaging and Transgenic Cell Counts
Transgenic larvae (Tg(sox10:tagRFP) and Tg(col2a1a:EGFP)) were xed at the stated time points using 4% paraformaldehyde. Fixed larvae were mounted in 0.6% low-melt agar in a glass bottom dish (Fisher).
Imaging was performed on a Zeiss LSM 700 at 20X and 40X Oil magni cation. Images were restricted to the larval craniofacial region. For each sh, a minimum of 20 to 30 z-stacks were collected. At 3 DPF, micron depth across the Tg(sox10:TagRFP) random control group ranged from 60-100 µm while micron depth ranged from 57-84 µm in morphants. Micron depth across the 4 DPF Tg(sox10:TagRFP) random control group ranged from 63-100 µm and from 66-93 µm in the morphant group. At 4 DPF Tg(col2a1a:EGFP) random control group micron depth ranged from 63-99 µm and from 72-90 µm in the morphant group.
The number of cells per z-stack (20-30 stacks/ sh) at both jaw joints were manually counted using the ImageJ cell counter, which marks each individual cell and keeps track of total cell numbers. Region of interest around the jaw joints were selected to be 3 rows of chondrocytes on the left side of the joint and 5 rows on the right. This region was selected because these cells were consistently visible across zstacks. Statistical signi cance was obtained by using a Student t-test with random control and morphant data (63).