Zebrafish Model for Spondylo-Megaepiphyseal-Metaphyseal Dysplasia Reveals Post-Embryonic Roles of Nkx3.2 in the Skeleton

Abstract

The regulated expansion of chondrocytes within growth plates and joints ensures proper skeletal development through adulthood. Mutations in the transcription factor NKX3.2 underlie Spondylo-megaepiphyseal-metaphyseal dysplasia (SMMD), which is characterized by skeletal defects including scoliosis, large epiphyses, wide growth plates, and supernumerary distal limb joints. Whereas nkx3.2 knockdown zebrafish and mouse Nkx3.2 mutants display embryonic lethal jaw joint fusions and skeletal reductions, respectively, they lack the skeletal overgrowth seen in SMMD patients. Here we report adult viable nkx3.2 mutant zebrafish displaying cartilage overgrowth in place of a missing jaw joint, as well as severe dysmorphologies of the facial skeleton, skullcap, and spine. In contrast, cartilage overgrowth and scoliosis are absent in rare viable nkx3.2 knockdown animals that lack jaw joints, supporting post-embryonic roles for Nkx3.2. Single-cell RNA sequencing and in vivo validation reveal increased proliferation and upregulation of stress-induced pathways, including prostaglandin synthases, in mutant chondrocytes. By generating a zebrafish model for the skeletal overgrowth defects of SMMD, we reveal post-embryonic roles for Nkx3.2 in dampening proliferation and buffering the stress response in joint-associated chondrocytes.

Full text

RESEARCH ARTICLE
Zebrafish model for spondylo-megaepiphyseal-metaphyseal
dysplasia reveals post-embryonic roles of Nkx3.2 in the skeleton
Joanna Smeeton
1,2,
*, Natasha Natarajan
1
, Arati Naveen Kumar
1
, Tetsuto Miyashita
3,4
, Pranidhi Baddam
5
,
Peter Fabian
1
, Daniel Graf
5,6
and J. Gage Crump
1,
*
ABSTRACT
The regulated expansion of chondrocytes within growth plates and joints
ensures proper skeletal development through adulthood. Mutations in
the transcription factor NKX3.2 underlie spondylo-megaepiphyseal-
metaphyseal dysplasia (SMMD), which is characterized by skeletal
defects including scoliosis, large epiphyses, wide growth plates and
supernumerary distal limb joints. Whereas nkx3.2 knockdown zebrafish
and mouse Nkx3.2 mutants display embryonic lethal jaw joint fusions
and skeletal reductions, respectively, they lack the skeletal overgrowth
seen in SMMD patients. Here, we report adult viable nkx3.2 mutant
zebrafish displaying cartilage overgrowth in place of a missing jaw joint,
as well as severe dysmorphologies of the facial skeleton, skullcap and
spine. In contrast, cartilage overgrowth and scoliosis are absent in rare
viable nkx3.2 knockdown animals that lack jaw joints, supporting post-
embryonic roles for Nkx3.2. Single-cell RNA-sequencing and in vivo
validation reveal increased proliferation and upregulation of stress-
induced pathways, including prostaglandin synthases, in mutant
chondrocytes. By generating a zebrafish model for the skeletal
overgrowth defects of SMMD, we reveal post-embryonic roles for
Nkx3.2 in dampening proliferation and buffering the stress response in
joint-associated chondrocytes.
KEY WORDS: Nkx3.2, Chondrocyte, Proliferation, Joint, Spine,
Zebrafish
INTRODUCTION
In much of the developing vertebrate body, cartilage is a transient
tissue that progressively remodels to bone through a process known as
endochondral ossification. In other contexts, such as cartilage of the
nose and ear and the articular cartilage lining the bony surfaces of
healthy joints, cartilage is permanent. The growth and function of both
transient and permanent cartilage relies on the stratification of
chondrocytes into distinct zones. In the growth plates of
endochondral bones, chondrocytes are arranged into a resting zone
containing stem cells (Mizuhashi et al., 2018; Newton et al., 2019), a
zone of round proliferative chondrocytes, a zone of proliferative
flattened chondrocytes that merges into a pre-hypertrophic zone, a
hypertrophic zone in which chondrocytes enlarge and calcify, and a
transitional zone in which chondrocytes undergo apoptosis or
transdifferentiate into osteoblasts (Giovannone et al., 2019; Jing
et al., 2015; Kronenberg, 2003; Yang et al., 2014; Zhou et al., 2014).
Joint cartilage is also zonally arranged: superficial chondrocytes of a
flattened morphology line the synovial cavity and produce specialized
molecules such as Prg4 (lubricin) (Askary et al., 2016; Kozhemyakina
et al., 2015), with deeper chondrocytes arranged in zones reminiscent
of growth plates and transitioning into the underlying subchondral
bone (Lui et al., 2015). Defects in the specification and maintenance of
these cartilage zones can lead to dwarfism and other skeletal dysplasias
in the context of the growth plate, and arthritis in the context of the
joint. The mechanisms for maintaining the correct proportions and
identities of cartilage layers at either growth plates or joints remain,
however, incompletely understood. In particular, we still know little
about how the proliferative expansion of cartilage is zonally regulated
to meet the differing demands of endochondral bone and joint growth.
Humans with spondylo-megaepiphyseal-metaphyseal dysplasia
(SMMD; OMIM #613330), a rare skeletal dysplasia linked to
homozygous frameshift mutations in the homeobox transcription
factor NKX3.2 (also called BAPX1 or NKX3-2), display skeletal
overgrowth in the wrists and digits that is accompanied by
supernumerary bones (pseudoepiphyses) as well as scoliosis
(Hellemans et al., 2009). In these patients, the cartilaginous
growth plates of long bones are abnormally wide. Whereas work
in animal models has revealed requirements for Nkx3.2 in regulating
chondrocyte development within both the growth plate and joints,
the skeletal overgrowth seen in SMMD has not been observed. In
chick and mouse, Nkx3.2 is expressed in the proliferating and pre-
hypertrophic chondrocytes of the growth plate but largely excluded
from the less proliferative hypertrophic zone and articular surfaces
of joints (Church et al., 2005; Provot et al., 2006; Tribioli et al.,
1997). Nkx3.2−/−mice have defects in their vertebrae and cranial
skeleton, which have been attributed to decreased cartilage
formation (Akazawa et al., 2000; Tribioli and Lufkin, 1999) and
is exacerbated by further mutation of the paralog Nkx3.1 (Herbrand
et al., 2002). However, despite expression of Nkx3.2 in the long
bones of the developing limbs, Nkx3.2−/−mice do not display the
elongated wrists and digits of SMMD patients at birth. As these
mice die shortly after birth, possibly because of asplenia, the lack of
limb overgrowth phenotypes could reflect postnatal roles for
Nkx3.2 in restricting cartilage growth. Nkx3.2 is thought to form
a positive regulatory loop with Sox9 (Yamashita et al., 2009; Zeng
et al., 2002), a master regulator of chondrogenesis (Bi et al., 1999),
consistent with their co-expression in growth plate chondrocytes
(Provot et al., 2006). Studies in the chick limb and mammalian cell
culture have suggested that Nkx3.2 may delay hypertrophic
maturation of growth plate chondrocytes through repression of
Handling Editor: Steve Wilson
Received 1 June 2020; Accepted 31 December 2020
1
Department of Stem Cell Biology and Regenerative Medicine, Keck School of
Medicine, University of Southern California, Los Angeles, CA 90033, USA.
2
Columbia Stem Cell Initiative, Department of Rehabilitation and Regenerative
Medicine, and Department of Genetics and Development, Columbia University
Irving Medical Center, Columbia University, New York, NY 10032, USA.
3
Department of Biological Sciences, University of Alberta, Edmonton, Alberta T6G
2E9, Canada.
4
Department of Organismal Biology and Anatomy, University of
Chicago, Chicago, IL 60637, USA.
5
Department of Dentistry, University of Alberta,
Edmonton, Alberta T6G 2R3, Canada.
6
Department of Medical Genetics, University
of Alberta, Edmonton, Alberta T6G 2R7, Canada.
*Authors for correspondence (gcrump@usc.edu; jms2504@cumc.columbia.edu)
J.S., 0000-0002-6126-2560; T.M., 0000-0003-0050-4594; P.B., 0000-0003-
0232-2022; P.F., 0000-0002-1096-6875; J.G.C., 0000-0002-3209-0026
1
© 2021. Published by The Company of Biologists Ltd
|
Development (2021) 148, dev193409. doi:10.1242/dev.193409
DEVELOPMENT

Col10a1 (Provot et al., 2006) and Runx2 (Kawato et al., 2011;
Yamashita et al., 2009), a transcription factor required for the
differentiation of cells into mineral-producing osteoblasts and
hypertrophic chondrocytes (Takeda et al., 2001).
In addition to roles in growth plate regulation, work in zebrafish
has revealed essential roles for Nkx3.2 in formation of the jaw joint
(Miller et al., 2003). Inhibition of Nkx3.2 function with an antisense
morpholino results in a complete loss of the zebrafish jaw joint,
including the bony retroarticular process that is a site of insertion for
the jaw adductor (adductor mandibulae). The role of Nkx3.2 in
specifying the jaw joint appears to be shared between zebrafish and
amphibians but not mammals, potentially due to positive regulation
of joint-promoting Gdf5/6 family members only in non-mammalian
species (Lukas and Olsson, 2018). Although the malleus-incus joint
(the evolutionary homolog of the zebrafish jaw joint) or the
temporomandibular joint (the mammalian jaw joint) are normal in
Nkx3.2−/−mice, the malleus is narrower and the associated gonial
bone is lost (Tucker et al., 2004).
Here, we characterize a zebrafish nkx3.2 mutant that lacks the jaw
joint (see also Miyashita et al., 2020 for a detailed analysis of adult
cranial defects). Similar to SMMD patients, mutants are adult viable
and develop cartilage outgrowths, albeit in the jaw joint region rather
than wrists and digits, as well as scoliosis. By comparing mutants to
animals with transient, early knockdown of nkx3.2 function, we find
that Nkx3.2 functions at post-embryonic stages to prevent cartilage
overgrowth and pattern the spine. Single-cell RNA-seq of facial
chondrocytes reveals a mutant-enriched population of chondrocytes
with high levels of cell cycle and stress-induced pathway genes,
including the prostaglandin D
2
synthase gene ptgdsb.1 and the
mTORC1 pathway genes sestrin1 (sesn1) (Budanov and Karin,
2008) and eukaryotic translation initiation factor 4E binding protein
3(eif4ebp3) (Tsukumo et al., 2016; Yogev et al., 2013). We validate
that, in place of the missing jaw joint, adult nkx3.2 mutants develop
abnormally proliferative chondrocytes with elevated levels of
ptgdsb.1 and sesn1, yet do not upregulate genes associated with
hypertrophic maturation. Rather than suppressing hypertrophic
maturation, our results indicate post-embryonic roles for Nkx3.2 in
restricting chondrocyte proliferation, possibly linked to prostaglandin
and mTORC1 regulation, to ensure proper skeletal proportions,
including in the limbs of SMMD patients.
RESULTS
Mutant nkx3.2 zebrafish lack a jaw joint and display skull
and spine defects as adults
Given concerns over efficacy and specificity of morpholino
knockdown (Eve et al., 2017; Gentsch et al., 2018; Joris et al.,
2017; Kok et al., 2015; Law and Sargent, 2014), we first aimed to
confirm requirements for nkx3.2 in zebrafish jaw joint development
(Miller et al., 2003) using genetic mutants. To do so, we generated
two mutant alleles, nkx3.2
el802
and nkx3.2
ua5011
using TALEN
(Huang et al., 2011; Sander et al., 2011) and CRISPR (Hwang et al.,
2013) mutagenesis, respectively (Fig. 1A). nkx3.2
el802
contains a
14 bp deletion that removes the ‘A’of the start codon and is
predicted to prevent translation of nkx3.2 mRNA, and nkx3.2
ua5011
introduces a 20 bp deletion and frameshift mutation predicted to
disrupt the DNA-binding homeobox domain. These are both
predicted to be severe loss-of-function alleles; lack of a zebrafish-
reactive Nkx3.2 antibody did not allow us to test this directly.
Consistent with the nkx3.2 morpholino knockdown phenotype
(Miller et al., 2003), animals homozygous for either nkx3.2 mutant
allele fail to develop a jaw joint, as assessed by Alcian Blue labeling
of cartilage at 7 days post-fertilization (dpf) (n=3/3; Fig. 1B;
Miyashita et al., 2020). In wild-type animals at 11 dpf, trps1:GFP
marks articular chondrocytes that are flanked by sox10:dsRed+
chondrocytes of the upper and lower jaw cartilages. Live imaging of
nkx3.2 mutant embryos revealed loss of trps1:GFP expression and a
continuous field of sox10:dsRed+ chondrocytes across the
presumptive jaw joint region (n=5/5; Fig. 1C). In contrast to the
Fig. 1. nkx3.2 mutant zebrafish are adult viable and develop craniofacial skeletal abnormalities. (A) Schematic of nkx3.2 mutant alleles: el802 –14 bp
deleted sequence including the ‘A’of the ATG at the nkx3.2 translation start codon; ua5011 –20 bp deleted sequence at the start of the homeobox DNA-binding
domain. Coding regions of exons shown in blue and non-coding regions in white. (B) Lateral views of pharyngeal skeletal preparations stained with Alcian Blue
(cartilage) and Alizarin Red (bone) show fusions of the jaw joint (arrowheads) in nkx3.2 mutants at 7 dpf. (C) Live imaging of transgenic juvenile zebrafish
demonstrates loss of sox10:DsRed+/trps1:GFP+jaw joint chondrocytes (arrowheads) in nkx3.2 mutants at 11 dpf. (D,E) Lateral views of the facial skeleton (D)
and ventral views of the neurocranium (E) in adult nkx3.2 mutants at 60 dpf show craniofacial skeletal abnormalities including gaping jaw, fused jaw joint and
displaced ceratohyal, and fusion of neurocranial orbitosphenoid and pterosphenoid bones (arrow in E). CH, ceratohyal; LJ, lower jaw; M, Meckel’s cartilage;
OS, orbitosphenoid; PQ, palatoquadrate; PTS, pterosphenoid; UJ, upper jaw; WT, wild type. Scale bars: 100 µm (B,C); 2 mm (D,E).
2
RESEARCH ARTICLE Development (2021) 148, dev193409. doi:10.1242/dev.193409
DEVELOPMENT

vast majority of nkx3.2 morpholino-treated embryos that die by
7 dpf, zebrafish homozygous for either mutant allele are viable and
fertile into adulthood, especially when raised separately from their
wild-type siblings, and display a fixed open jaw (n=6/6; Fig. 1D;
Miyashita et al., 2020).
Adult nkx3.2 mutant zebrafish display a number of other cranial
skeletal defects beyond jaw joint loss, as well as vertebral
abnormalities. Mutants display altered orientation of facial bones
(e.g. ceratohyal), possibly due to mechanical restriction from jaw joint
fusion (n=6/6; Fig. 1D). The paired pterosphenoid and orbitosphenoid
bones are articulating endochondral bones of the neurocranium
(Cubbage and Mabee, 1996) and these are abnormally fused in nkx3.2
mutants (n=6/6; Fig. 1E). Micro-computed tomography (µCT)
imaging further reveals that 11/11 nkx3.2 mutants have abnormal
ossification and a shortening of the rostral spine (Fig. 2B,F),
misalignment of ribs (Fig. 2C,G) and rotation and fusion of the
caudal vertebrae (Fig. 2D,H) at 60 dpf, with spinal defects apparent as
early as 30 dpf (n=11/11; Fig. S1). Overall, the combined rostral
defects and abnormal curvature are associated with shortening of the
spine (Fig. 2A,E). Spinal deformities are consistent with the reported
expression of nkx3.2 in the neural and hemal arches that contribute to
vertebrae (Crotwell et al., 2007), as well as the scoliosis and shortening
oftheneckinSMMDpatients.
Ectopic jaw cartilage overgrowth in nkx3.2 mutants
We next examined the long-term fate of nkx3.2 mutant cells that would
have normally formed the jaw joint. In whole-mount preparations of the
60 dpf skull stained with Alcian Blue (cartilage) and Alizarin Red
(bone), we observed greatly increased Alcian-positive tissue in the
region where the jaw joint would have been (n=6/6; Fig. 1D). This
cartilage overgrowth was particularly evident upon dissection of the
presumptive jaw joint tissue (insets in Fig. 3A,B). Hematoxylin and
Eosin (H&E) staining of frontal sections through the presumptive jaw
joint region further demonstrated a dramatic expansion of cartilage
tissue in mutants (n=3/3; Fig. 3C,D). We next assessed proliferation in
the presumptive jaw joint region at 21 dpf by subjecting animals to a 1 h
pulse of bromodeoxyuridine (BrdU), which is incorporated into newly
replicating DNA (Fig. 3E,F). In 3/3 wild types, we observed very few
BrdU+ cells near the jaw joint, and sparse BrdU+ chondrocytes in the
ceratohyal growth plate. In 3/3 mutants, we observed increased
numbers of BrdU+ chondrocytes in place of the missing jaw joint,
consistent with cartilage overgrowth. To validate the BrdU
incorporation analysis, we also assessed expression of Proliferating
Cell Nuclear Antigen (PCNA), a marker for DNA replication, in
chondrocytes (Fig. 3G-J). Around wild-type joints, we observed a few
PCNA+ cells within the lower jaw Meckel’s cartilage, but almost no
PCNA+ cells at the joint articular surface. PCNA+ chondrocytes were
more abundant in the proliferative zone of the ceratohyal growth plate.
In nkx3.2 mutants, numerous PCNA+ chondrocytes are evident in the
fused jaw joint region. Quantification of the percentage of chondrocytes
expressing PCNA (Fig. 3K) revealeda dramatic increase in proliferative
chondrocytes at the nkx3.2 mutant jaw joint region (identified by nkx3.2
expression in adjacent sections) compared with the wild-type jaw joint
(n=3 per genotype, P=0.009), similar to the levels of proliferative
chondrocytes within the wild-type ceratohyal growth plate. We also
observed a modest increase in proliferation rates within the mutant
ceratohyal growth plate (n=3 per genotype, P=0.07). These findings
indicate that chondrocytes in the presumptive jaw joint region, and
Fig. 2. nkx3.2 mutant zebrafish develop axial skeletal abnormalities. (A,E) µCT imaging of 60 dpf wild types (A,A′) and nkx3.2
ua5011/ua5011
mutants (E,E′)
shown with gradient 3D rendering in sagittal (A,E) and dorsal (A′,E′) views. (B-D,F-H) Magnifications of three spinal regions: anterior vertebrae (red box,
inclusive of the occiput and Weberian apparatus: B,B′,F,F′), precaudal vertebrae with rib articulations (tan box, white arrows indicate corresponding left and right
ribs: C,C′,G,G′), and caudal fin vertebrae (blue box, pink arrow indicates caudal-most vertebra: D,D′,H,H′) shown in 3D (B,C,D,F,G,H) and 2D (B′,C′,D′,F′,G′,H′)
slice. Scale bars: 1 mm (A,A′); 0.61 mm (E,E′).
3
RESEARCH ARTICLE Development (2021) 148, dev193409. doi:10.1242/dev.193409
DEVELOPMENT

likely to a lesser extent in the ceratohyal growth plate, become
increasingly proliferative in nkx3.2 mutants, likely contributing to the
cartilage overgrowth seen in juvenile and adult mutants.
Nkx3.2 function is required in post-embryonic stages
to suppress cartilage overgrowth
We next sought to determine whether the cartilage overgrowth seen in
adult nkx3.2 mutants was solely the consequence of not forming the
jaw joint,or alternatively reflected a continued requirement for nkx3.2
in chondrocyte regulation. To uncouple early versus continual
requirements for nkx3.2, we took advantage of the transient nature of
antisense morpholino knockdown –dilution during embryonic cell
divisions is thought to limit the window of efficacy of most
morpholinos to the first few days of embryogenesis (Eisen and Smith,
2008). As described earlier, the jaw joint loss seen in both nkx3.2
mutant alleles is nearly identical to that reported for nkx3.2
morpholino knockdown (Miller et al., 2003). We therefore injected
one-cell-stage embryos with the same dose of nkx3.2 morpholino
reported by Miller et al. (2003). Using sox10:dsRed to visualize
chondrocytes in live animals, we confirmed loss of the jaw joint in the
majority of morpholino-injected animals at 7 dpf and only raised
animals lacking jaw joints to adulthood. From 225 animals lacking
Fig. 3. Cartilage overgrowth at the jaw joint in nkx3.2 mutant zebrafish. (A,B) Ventral views of dissected pharyngeal skeletons stained with Alcian Blue
(cartilage) and Alizarin Red (bone) at 60 dpf. Insets with further dissections of the boxed regions in A and B demonstrate the wild-type retroarticular cartilage in
A and the cartilage overgrowth at the site of the fused jaw joint in nkx3.2 mutants in B. (C,D) Frontal H&E-stained sections through 60 dpf jaw joints show
fusion of the jaw joint and cartilage overgrowth in nkx3.2 mutants. (E-J) In wild types at 21 dpf (E,G,I), PCNA+ cells are present largely on the lower jaw side of the
jaw joint, and PCNA+ and BrdU+ chondrocytes are present in the proliferative zone of the ceratohyal. In nkx3.2
el802
mutants (F,H,J), proliferating chondrocytes
marked by BrdU (E,F) and PCNA (G-J) are present across the fused jaw joint region. (K) Quantification of PCNA+ cells per total nuclei stained with Hoechst.
Individual animals are plotted (dots) with mean±s.d. CH, ceratohyal; JJ, jaw joint; LJ, lower jaw;M, Meckel’s cartilage; PQ, palatoquadrate; UJ, upper jaw; WT, wild
type. Scale bars: 500 µm (A,B); 50 µm (C-J).
4
RESEARCH ARTICLE Development (2021) 148, dev193409. doi:10.1242/dev.193409
DEVELOPMENT

the embryonic jaw joint, only four inflated their swim bladders and
survived until adulthood (n=2 at 60 dpf, n=2 at 90 dpf ), in marked
contrast to the viability of jaw-joint-less nkx3.2 mutants. The four
viable morpholino-injected adults exhibited a similar open gape of
the jaw as seen in nkx3.2 mutants (Fig. 4A-F). In wild types,
permanent Meckel’s cartilage within the lower jaw is separated from
the cartilaginous growth plate of the upper jaw palatoquadrate bone
by the jaw joint and associated subchondral bone (Fig. 4D,G). In
mutants, we observed fusion of lower jaw cartilage to the
palatoquadrate cartilage growth plate (Fig. 4H; see also Fig. 3D). In
contrast, in morpholino-injected animals, bone continues to separate
lower jaw cartilage and the palatoquadrate cartilage growth plate
despite the lack of a jaw joint, and no cartilage overgrowth was
observed in any of the four adult-viable animals (Fig. 4I). Whereas
5/6 nkx3.2
el802
mutant animals displayed severe caudal tail curvature
defects (Fig. 4K), all four morpholino-injected adults had normal
spines and tails (Fig. 4J-L). These findings indicate post-embryonic
roles for Nkx3.2 in restricting joint-associated cartilage growth and
ensuring normal spine development.
Single-cell RNA-seq reveals upregulated mitotic and stress
response genes in nkx3.2 mutant chondrocytes
In order to understand how Nkx3.2 might regulate subtypes of joint-
associated chondrocytes, we used single-cell RNA-seq to profile
chondrocytes in wild-type and nkx3.2 mutant juvenile heads. In
zebrafish, fli1a:GFP and sox10:dsRed transgenes co-localize
predominantly in chondrocytes (Askary et al., 2017; Giovannone
et al., 2019). We therefore performed fluorescence activated cell
sorting (FACS) of fli1a:GFP+/sox10:DsRed+ cells (Fig. S2) from
pooled 21 dpf zebrafish heads (n=5 per genotype), followed by
single-cell barcoded cDNA synthesis using the 10x Genomics
platform and Illumina sequencing. After quality control, we obtained
1641 cells from wild-types and 1699 cells from nkx3.2
el802
mutants
(Fig. 5A). We used Seurat (Butler et al., 2018) on aggregated data and
Uniform Manifold Approximation and Projection (UMAP) for
dimension reduction and visualization (McInnes et al., 2018) to
identify 11 distinct cell clusters (Fig. 5B). Five clusters are defined by
high levels of cartilage collagen-encoding genes, col2a1a and
col9a1a (Col2
hi
/Col9
hi
), with two of these representing chondrocytes
undergoing S-phase DNA replication ( pcna+) or mitosis (ube2c+/
pcna+) (Fig. 5C,E). We also observed clusters with hypertrophic
chondrocyte features (col10a1a+/spp1+/col2a1a
lo
), articular
chondrocyte features ( f13a1b+/prg4b+/col2a1a
lo
), perichondrial
features ( foxp4+/col2a1a
lo
), two with mesenchymal/fibroblast
connective tissue features (ifitm1+/col5a1+/col2a1a
lo
) and one
with an osteoblast signature (col1a1a+/ifitm5+) (Askary et al.,
2016; Moffatt et al., 2008; Zhao et al., 2015).
All of the cell clusters present in wild types are also present in
nkx3.2 mutants. However, we observed two clusters comprised
almost entirely of mutant cells (Fig. 5D). The first represents mitotic
chondrocytes (94% mutant cells), consistent with the abnormal
chondrocyte proliferation seen in place of the missing jaw joint in
Fig. 4. Adult nkx3.2 morphants lack the
cartilage overgrowth and spine defects
of nkx3.2 mutants. (A-I) Lateral views of
whole-mount (A-C) and dissected
(D-F) facial skeletons at 60 dpf. Alcian
Blue (cartilage) and Alizarin Red (bone)
staining show similar gaping jaws in
mutant and morphant animals. High
magnification views of the jaw joint
(G-I) show ectopic cartilage across the
fused jaw joint domain (arrowheads) in
mutants (H) but not morphants (I). Arrows
indicate distinct palatoquadrate growth
plates present in wild types and
morphants but not in mutants. (J-L) Lateral
views of the caudal spine stained with
Alcian Blue and Alizarin Red show spinal
curvature in mutants but not in morphants.
CH, ceratohyal; LJ, lower jaw; M, Meckel’s
cartilage; UJ, upper jaw. Scale bars: 2 mm
(A-C; J-L); 500 µm (D-F); 50 µm (G-I).
5
RESEARCH ARTICLE Development (2021) 148, dev193409. doi:10.1242/dev.193409
DEVELOPMENT

Fig. 5. See next page for legend.
6
RESEARCH ARTICLE Development (2021) 148, dev193409. doi:10.1242/dev.193409
DEVELOPMENT

mutants (Fig. 3E-K). A second chondrocyte cluster (98% mutant cells)
was characterized by higher levels of sesn1 and eif4ebp3 (Fig. 5C, E),
which encode inhibitors and repressed targets, respectively, of
mTORC1-driven cellular growth and translation (Budanov and
Karin, 2008; Tsukumo et al., 2016; Yogev et al., 2013). Both genes
are induced by cellular stress in other systems (Budanov and Karin,
2008; Sukarieh et al., 2009; Taba et al., 2000). Compared with other
chondrocytes, this mutant chondrocyte cluster also displayed higher
levels of the prostaglandin D
2
synthase gene ptgdsb.1, another stress-
response pathway (Fig. 5E). However, many other regulators of
chondrocyte biology, such as sox9a and hypertrophic genes runx2b
and col10a1a, were unchanged in mutants.
Upregulation of ptgdsb.1 and sesn1 in nkx3.2 mutant
chondrocytes
We next sought to understand the in vivo location of the mutant
chondrocyte cluster identified in the single-cell analysis. In wild types at
21 dpf, we observed nkx3.2 expression in chondrocytes on either side of
the jaw joint but not in the superficial zone (Fig. 6A,G,I). Expression of
nkx3.2 was also seen in pre-hypertrophic chondrocytes of the
bidirectional ceratohyal growth plates, in a largely non-overlapping
pattern to expression of the hypertrophic marker col10a1a (Fig. 6C,K,
M), similar to what has been observed in chick and mouse (Church
et al., 2005; Provot et al., 2006; Tribioli et al., 1997). In nkx3.2
el802
mutants, nkx3.2 mRNA transcripts were present in chondrocytes
spanning the fused jaw joint (Fig. 6B,H,J), consistent with single-cell
RNA-seq analysis showing nkx3.2 expression in the mutant-enriched
chondrocyte cluster (Fig. 5E), yet col10a1a was not expressed
(Fig.6D).Inwildtypes,ptgdsb.1 was expressed in the late
hypertrophic zone in the upper jaw palatoquadrate and ceratohyal
cartilages, in the proliferative zone of the ceratohyal growth plate, at
modest levels in the lower jaw Meckel’s cartilage and at lower levels in
jaw joint articular cartilage, particularly on the upper jaw side of the
joint (Fig. 6E,G,K,M). Although sesn1 was expressed on either side of
the wild-type jaw joint, it was expressed at much lower levels at the
articular surface, with only rare sesn1+/nkx3.2+ double-positive cells
detected; in the ceratohyal growth plate sesn1 was expressed at high
levels in all regions except the proliferative zone (Fig. 6I,O). In nkx3.2
mutants, ptgdsb.1 and sesn1 were upregulated across most
chondrocytes within the fused jaw joint region, which we identified
by co-expression of nkx3.2 (Fig. 6E-J). In the mutant ceratohyal growth
plates, we observed largely normal zones of ptgdsb.1,col10a1a,sesn1
and nkx3.2 expression, with a potential modest upregulation of sesn1 in
proliferative zones (Fig. 6K-P). Quantification of fluorescence intensity
in the nkx3.2-positive jaw joint region confirmed upregulation of
ptgdsb.1 and sesn1 in mutants (Fig. 6Q; ptgdsb.1:n=6, P=0.003; sesn1:
n=4, P=0.03). These results show upregulation of stress response and
mTORC1 pathway genes in the nkx3.2 mutant chondrocytes that fail to
form a jaw joint and contribute to cartilage overgrowth (Fig. 6R).
DISCUSSION
Our resultssupport temporally distinct roles for Nkx3.2 in embryonic
specification of the zebrafish jaw joint and post-embryonic regulation
of joint-associated cartilage growth. By comparing embryonic-only
loss of nkx3.2 (morpholino knockdown) to constitutive loss
(mutants), we uncovered post-embryonic requirements for Nkx3.2
in restricting subarticular cartilage growth and patterning the spine. In
contrast to Nkx3.2−/−mice that die shortly after birth and do not
display skeletal overgrowth (Akazawa et al., 2000; Tribioli and
Lufkin, 1999), the cartilage overgrowthseen in the jaw region of adult
viable nkx3.2 mutant zebrafish is a closer model to the supernumerary
bones and lengthening of the wrist and digits in SMMD patients. As
the number of cartilage segments in the digits appears to be
proportional to cartilage length, with joint-derived signals such as
Wnt9a inhibiting the formation of another joint within a certain
distance (Hartmann and Tabin, 2001), the increased numbers of digit
bones could be explained by cartilage overgrowth analogous to that
seen in the zebrafish nkx3.2 mutant jaw. Spinal curvature defects are
also more pronounced in zebrafish nkx3.2 than mouse Nkx3.2
mutants, again more closely aligning to the scoliosis seen in SMMD
individuals (Hellemans et al., 2009; Tribioli and Lufkin, 1999).
Previous studies have focused on a potential role for Nkx3.2 in
inhibiting hypertrophic maturation of chondrocytes. As in chick
(Provot et al., 2006), we observed that nkx3.2 expression largely
anti-correlates with col10a1a expression in hypertrophic
chondrocytes. However, we observed no changes in hypertrophic
cartilage maturation, or expansion of hypertrophic gene expression,
in zebrafish nkx3.2 mutants, similar to what has been reported for
mouse Nkx3.2 mutants (Akazawa et al., 2000; Tribioli and Lufkin,
1999). Instead, we observed a marked increase in chondrocyte
proliferation within the mutant jaw joint region, and a more modest
trend toward increased proliferation within the mutant growth plate
of the ceratohyal endochondral bone. We did not, however, observe
apparent cartilage overgrowth in the cranium and spine that are also
affected in zebrafish nkx3.2 mutants, suggesting that Nkx3.2 has
roles beyond restricting chondrocyte proliferation.
Our expression analysis also points to important differences in
zones of chondrocytes between growth plates and joints. In the
bidirectional growth plates of the ceratohyal (Fig. 6S), we observed a
central proliferative zone of nkx3.2
lo
,sesn1
lo
,ptgdsb.1+
chondrocytes (Fig. 3E,I; Giovannone et al., 2019). Flanking these
proliferative chondrocytes, we observed nkx3.2+, ptgdsb.1−zones,
which we consider pre-hypertrophic based on the absence of
col10a1a expression. Toward the middle of the ceratohyal we
observed successive zones of col10a1a+ hypertrophic chondrocytes
and then ptgdsb.1+; col10a1a
lo
late hypertrophic chondrocytes. In
contrast to the growth plate, we did not observe a prominent nkx3.2
lo
,
ptgdsb.1+ proliferative zone close to the jaw joint. Relative to its
expression in the pre-hypertrophic zones of the growth plates, nkx3.2
expression is also higher in subarticular chondrocytes of the juvenile
jaw joint, consistent with a greater role in restricting chondrocyte
proliferation. In nkx3.2 mutants, upregulation of ptgdsb.1 but not the
hypertrophic marker col10a1a in the fused jaw joint region suggests
that chondrocytes are adopting a partial ptgdsb.1+proliferative
identity rather than a late hypertrophic identity. This is supported by
single-cell transcriptomic analysis showing that mitotic chondrocytes
are greatly increased in abundance in nkx3.2 mutants, and our
observations of increased proliferation and cartilage overgrowth at the
fused jaw joint region. However, upregulation of sesn1 in the fused
jaw joint region suggests that mutant chondrocytes are not equivalent
to growth plate proliferative chondrocytes that normally lack sesn1.
This is borne out by mutant chondrocytes forming a distinct cluster in
our single-cell analysis. Our data therefore support a role for Nkx3.2
in limiting chondrocyte proliferation, particularly at the jaw joint
where it is most abundantly expressed.
Fig. 5. Single-cell analysis reveals altered facial chondrocyte subtypes in
nkx3.2 mutants. (A,B) UMAP projections of the aggregate datasets colored
according to dataset (A) or cluster identity (B). (C) Heatmap of the top 4-6
enriched genes per cluster. Yellow indicates high expression and magenta
minimal. (D) Percent contribution within each cluster by control (WT) and
mutant cells. (E) UMAP projections of gene expression split by genotype. Red
indicates high expression and greyminimal. Pink circles indicate position of the
mitotic chondrocytes cluster and blue circles indicate the chondrocytes-2
cluster.
7
RESEARCH ARTICLE Development (2021) 148, dev193409. doi:10.1242/dev.193409
DEVELOPMENT

Fig. 6. See next page for legend.
8
RESEARCH ARTICLE Development (2021) 148, dev193409. doi:10.1242/dev.193409
DEVELOPMENT

The higher levels of nkx3.2 expression at the wild-type jaw joint
compared with the pre-hypertrophic zone of the growth plates may
reflect slower growth ofthe subarticular zone relative to the expansive
growth of the endochondral cartilage template; this is reflected by a
fivefold lower proliferation rate at the jaw joint (Fig. 3K). These
findings align with recent single-cell transcriptome analysis in mouse
showing that chondrocyte zones in joints and growth plates are related
but not identical (Lui et al., 2015). It will be interesting to test how
Bmp, Fgf, and Shh signaling pathways regulate nkx3.2 expression at
joints versus growth plates, given previous studies showing important
roles for each of these pathways in regulating Nkx3.2 expression
(Wilson and Tucker, 2004; Zeng et al., 2002). It will also be
importanttodeletenkx3.2 only at post-embryonic stages to further
confirm its requirements independent from initial specification of the
fish jaw joint. Whereas we detected a modest increase in chondrocyte
proliferation at the ceratohyal growth plate of nkx3.2 mutant
zebrafish, we did not detect prominent cartilage overgrowth. This
cannot be attributed to compensation by nkx3.1 (which we failed to
detect in our single-cell RNA-seq analysis) or a broader genetic
compensation (as we saw no evidence of the nonsense-mediated
decay of nkx3.2 transcripts required for transcriptional adaptation).
One possibility is that the preferential requirement for nkx3.2 at the
zebrafish jaw joint reflects the lower levels of baseline proliferation
there, and hence the greater magnitude of cartilage overgrowth when
nkx3.2 is missing.
The increased proliferation of joint-associated chondrocytes in
nkx3.2 mutants was linked to increased expression of stress-induced
genes such as prostaglandin synthase and mTORC1 inhibitors.
The presence of stress-induced pathways in late hypertrophic
chondrocytes could reflect the hypoxic stress linked with their
eventual apoptosis. However, strong expression of prostaglandin D
2
synthase ptgdsb.1 in the proliferative zone of the wild-type growth
plate was surprising, as it is not clear what types of stress these cells
would experience. Expression of Ptgdsb proteins in the adult
zebrafish skeleton has also been observed by proteomic analysis
(Kessels et al., 2014). In mouse, single-cell transcriptomic analysis of
the growth plate revealed expression of the prostaglandin E
2
synthase
Ptges3 in both the early proliferative zone and the late hypertrophic
zone (Li et al., 2016), similar to ptgdsb.1 in zebrafish. The functions
of prostaglandins in regulating chondrocyte biology, whether they are
linked to proliferation, and why zebrafish would have D
2
and mouse
E
2
subtypes remain outstanding questions. In the zebrafish testes,
treatment with the D
2
analogue BW-245C from 15 to 40 days of life
resulted in upregulated expression of sox9a (Pradhan and Olsson,
2014), a master regulator of chondrogenesis in the skeletal system
(Bi et al., 1999; Yan et al., 2005), though whether similar regulation
occurs in chondrocytes has not been examined. In nkx3.2 mutants, it
will be interesting to determine whether the upregulation of stress-
induced pathways such as prostaglandin production, as well as
modulators of mTORC1 signaling such as sesn1, are linked to the
observed ectopic cartilage outgrowth, and possibly scoliosis and
other skeletal defects.In the postnatal mouse, the proliferative zone of
the growth plate undergoes an mTORC1-regulated transition to
become a source of clonal stem cells that fuel cartilage growth
(Mizuhashi et al., 2018; Newton et al., 2019). One possibility is that
Nkx3.2 may function to normally regulate such an mTORC1-
regulated transition below the joint surface, and to a certain extent in
the pre-hypertrophic zone, to prevent cartilage overgrowth. The
supernumerary bones and lengthening of the wrists and digits in
SMMD patients may reflect an analogous role for Nkx3.2 in fine-
tuning the rate of stem cell-mediated expansion of cartilage. In the
future, it will be informative to determine how prostaglandins and
other stress-induced pathways interact with mTORC1 signaling to
regulate chondrocyte proliferation, and how Nkx3.2-mediated
regulation of these pathways differentially fine-tunes stem cell-
mediated chondrocyte expansion in joints versus growth plates.
MATERIALS AND METHODS
Zebrafish lines
The Institutional Animal Care and Use Committees of the University of
Southern California, Columbia University, and the University of Alberta
approved all use of zebrafish in this study. Previously reported zebrafish lines
used in this study include: Tg(sox10:dsRED)
el10
(Das and Crump, 2012),
trps1
j127aGt
(RRID:ZFIN_ZBD-GENO-100809-11) (Talbot et al., 2010),
Tg(fli1a:EGFP)
y1
(ZIRC Cat# ZL1085 ZDB-ALT-011017-8) (Lawson and
Weinstein, 2002). nkx3.2 mutant lines were generated by TALEN (el802)and
CRISPR (ua5011) mutagenesis. TALEN constructs were generated using
TALEN targets L: TAATGCTCCGCGACCCGGTC and R: GACATGGC-
TGTGCGCAGTAA. For genotyping, fin clips were performed at 14 or 90
dpf and genotyped by PCR using GoTaq DNA polymerase (Promega).
Genotyping for allele el802 was performed using nkx32-F: TAACCCTAA-
TGCTCCGCGAC and nkx32-R: TGGCGAGACTCCTCTTTTCG primers
to generate 118 bp WT and 104 bp MUT bands. nkx3.2 mutants inflate their
swim bladders and are recovered at expected rates for analyses at 7-21 dpf (39/
161: 24.2% mutant). When raised with wild types, mutant adults are
recovered at lower rates (6/90: 6.6% mutant) perhaps owing to food
competition with phenotypically normal clutch-mates. To raise sufficient
numbers of mutants to adulthood for experimental purposes, genotyping was
performed at 14 dpf. Mutants and wild types were then raised at similar
densities in different tanks and size matched for downstream analyses at the
indicated ages with no survival defects noted.
µCT
Wild-t ype (AB; sox10:GFP) and nkx3.2
ua5011/ua5011
zebrafish were scanned
using MILabs μCTat the School of Dentistry, University of Alberta, Canada.
For µCT scanning, 30 and 60 dpf individuals from two distinct F
2
populations
for each genotype (wild type: n=10; nkx3.2
ua5011
:n=11 at each age) were
fixed in 4% paraformaldehyde (PFA) for 24 h then dehydrated in a graded
ethanol series. Scanning parameters were as follows: voxel size=10 µm;
voltage=50 kV; current=0.24 mA; exposure time=75 ms. Skeletal
reconstruction was performed using AMIRA (Milabs) by selecting and
delineating regions of high tissue densities at a voxel size of 10 μm.
Histology and proliferation analysis
Anaesthetized fish were transferred into system water containing 4.5 mg/ml
BrdU (B5002, Sigma-Aldrich) and immersed for 1 h before euthanasia.
Tissue was fixed overnight in 4% PFA, transferred to 20% EDTA at room
temperature for decalcification for 3 days and then processed for paraffin
embedding. Immunohistochemistry was performed on 5 µm sections with
antigen retrieval by steaming for 20 min in citrate buffer ( pH 6.0). Primary
antibodies used were rat anti-BrdU (1:100, MCA2060GA, Bio-Rad), mouse
Fig. 6. Increased expression of ptgdsb.1 and sesn1 across the fused jaw
joint of nkx3.2 mutant zebrafish. (A,B) Colorimetric in situ hybridization
demonstrates nkx3.2 expression in the sub-articular zone of the jaw joint and
across the fused jaw joint in nkx3.2
el802
mutants at 21 dpf. (C-F) Colorimetric in
situ hybridization shows expansion of ptgdsb.1 but not col10a1a across the
fused jaw joint region of nkx3.2 mutants. (G-J′) RNAScope in situ hybridization
shows ectopic expression of ptgdsb.1 (G-H′) and sesn1 (I-J′) across the fused
jaw joint region of mutants (identified by nkx3.2 expression in red). DAPI labels
nuclei in blue. Dashed lines in H,H′indicate fused jaw joint region. (K-P)
RNAscope in situ hybridizations show expression of col10a1a,ptgdsb.1,
nkx3.2 and sesn1 in the growth plates of the ceratohyal. (Q) Quantification of
fluorescence intensity at the jaw joint demonstrates upregulation of ptgdsb.1
and sesn1 in nkx3.2 mutants. Individual animals are plotted (dots) with mean
±s.d. (R,S) Model of chondrocyte zones at the wild-type and fused mutant jaw
joint (R) and within the bidirectional ceratohyal growth plate (S). AC, articular
surface; Hyp, hypertrophic; JJ, jaw joint; LH, late hypertrophic; M, Meckel’s
cartilage; PH, pre-hypertrophic; PQ, palatoquadrate; PZ, proliferative zone;
SA, sub-articular zone; WT, wild type. Scale bars: 50 µm.
9
RESEARCH ARTICLE Development (2021) 148, dev193409. doi:10.1242/dev.193409
DEVELOPMENT

anti-PCNA (1:1000, P8825, Sigma-Aldrich) diluted in serum-free antibody
diluent (Dako, Agilent) overnight at 4°C. Primary antibodies were detected
by incubating slides in secondary AlexaFluor antibodies (A21094, A11001,
Invitrogen) diluted at 1:500 in antibody diluent for 1 h at room temperature
in the dark with Hoechst 33342 nuclear stain. H&E staining was performed
on 5 µm sections. Briefly, sections were de-paraffinized, stained for 2 min in
Hematoxylin solution (Harris Modified, HHS16, Sigma-Aldrich), followed
by a brief acid rinse and water wash and incubation for 2 min in Blueing
Reagent (Scott’s Tap Water Substitute, 11160, EK Industries). Slides were
then stained in Eosin Y (E6003, Sigma-Aldrich) for 30 s followed by
mounting using Cytoseal 60 (8310, Richard-Allan Scientific) for imaging.
In situ hybridization and RNAscope
In situ hybridization labeled probes for col10a1a (Askary et al., 2016) and
ptgdsb.1 were generated by PCR. ptgdsb.1 primers: Fwd, CTGCAAACAT-
GACGAGTGTC; Rev, AATGCTTTGCCATTCTTTCCC. Digoxigenin
(DIG)-labeled antisense probes were synthesized using SP6 or T7 polymerase
(Roche). In situ hybridization was performed as previously described (Askary
et al., 2016). Briefly, after deparaffinization, 5 µm sections were digested in
7.5 µg/ml proteinase K for 5 min and post-fixed in 4% PFA/0.2% glutaral-
dehyde for 20 min. Then, 1 µg of DIG-labeled probe per slide was incubated
overnight at 62°C. Following hybridization, slides were washed 3× in 50%
Formamide, 1× saline sodium citrate (SSC), 0.1% Tween-20 Wash buffer and
3× in MABT [0.1M maleic acid, 0.15M NaCl, 0.1% Tween-20 ( pH 7.5)].
Slides were blocked in 2% Roche Blocking buffer followed by 1 h incubation
in anti-DIG-AP (Roche). Color reaction was developed with NBT/BCIP
(Roche). Nuclei were counterstained with Fast Red and slides were mounted
using Cytoseal for imaging. RNAscope staining was performed using the
RNAscope multiplex fluorescent assay v2 using the manufacturer’sprotocol
for formalin-fixed paraffin-embedded 5 µm sections (323100, Advanced Cell
Diagnostics). RNAscope probes for col10a1a (C1), nkx3.2 (C2), ptgdsb.1
(C3) and sesn1 (C3) were synthesized by Advanced Cell Diagnostics. For
RNAscope analysis, nuclei were stained with DAPI. For all in situ hybridi-
zations and RNAscope stains we saw similar patterns of expression in at least
three individuals for each experiment.
Morpholino injections
As per Miller et al. (2003), knockdown of nkx3.2 (bapx1) expression during
early embryonic development was performed by injecting antisense
morpholino oligos. bapx1-MO1 (5′-GCGCACAGCCATGTCGAGCAGC-
ACT-3 ′; ATG start complementarysequence underlined) was purchased from
GeneTools, diluted to 3 mg/ml and injected into the yolk of one-cell-stage
Tg(sox10:DsRed)
el10
embryos. Morpholino-injected animals to be raised to
adulthood were screened for jaw joint fusion by imaging sox10:DsRed+
craniofacial cartilages at 7 dpf.
Skeletal staining
Adult specimens were fixed overnight at 4°C in 4% PFA. Following a 1 h
wash in Tris/MgCl
2
, cartilage was stained overnight in 0.01% Alcian Blue in
10 mM MgCl
2
. Samples were re-hydrated through an ethanol/100 mM Tris
(pH 7.5) series into 0.5% KOH and incubated overnight at 4°C. Samples
were bleached in 3% H
2
O
2
/0.5% KOH for 6-8 h and then neutralized
overnight in 35% NaBO
4
. Tissue was cleared using 1% Trypsin in 35%
NaBO
4
for 3-4 h followed by staining in 0.02% Alizarin Red (pH 7.5)
overnight. Samples were washed in 50% glycerol/0.5% KOH to remove
residual stain and stored in 100% glycerol before imaging using a Leica S8
APO stereomicroscope.
Single-cell preparation and sequencing
For single-cell profiling of juvenile chondrocytes, zebrafish double-positive
for fli1a:GFP; sox10:DsRed were selected and screened for mutant joint
fusion phenotype. PCR analysis to confirm genotypes was performed at
14 dpf. At 21 dpf, wild-type and nkx3.2
el802
mutant craniofacial skeletons
were micro-dissected and processed into a single-cell suspension using
mechanical and chemical dissociation for 30 min at 28°C in protease solution
[0.25% Trypsin, 1 mM EDTA ( pH 8.0), PBS] containing collagenase-D
(Roche Life Sciences) (n=5 animals/genotype). GFP+/DsRed+ live cells were
sorted into suspension medium (1% calf serum, 0.8 mM CaCl
2
,50U/ml
penicillin, 0.05 mg/ml streptomycin) and immediately loaded onto a 10x
Genomics microfluidic chip. Barcoded cDNA libraries were generated using
10x Genomics Chromium Single Cell 3′Library and Gel Bead Kit v.2
according to the manufacturer’s instructions. Library sequencing on Illumina
NextSeq was performed to a depth of at least 1,000,000 reads/cell for each
library. Cell Ranger software (v. 3.1.0, 10x Genomics) was used for barcode
recovery, genome alignment (Ensembl GRCz11) and to generate gene-by-cell
count matrices with default parameters for each library. RNA-seq files have
been deposited in the NCBI Gene Expression Omnibus and are available
under accession number GSE151354.
Data analysis
R software with Seurat Version 3 was used for downstream analysis. Cells
expressing <200 and >2500 unique RNA Features and >5% mitochondrial
RNA were excluded. Datasets were aggregated using the merge function.
Aggregate data was log normalized, centered and scaled using default
settings of NormalizeData and ScaleData functions. Linear dimensional
reduction was performed with principal component analysis using the
RunPCA function. Cell clusters were determined using Find Neighbors and
FindClusters functions. Cluster marker genes were identified using
FindMarkers function. In order to focus the analysis on the craniofacial
chondrocytes affected in nkx3.2 mutants, cells of the gill were identified
based on the expression of ucmaa and ppp1r1c (P.F. and J.G.C.,
unpublished) and fin cells based on expression of posterior hox genes
(hox9-13) (Fig. S3) (Ahn and Ho, 2008). We then excluded gill and fin cells
from downstream analysis using the subset function and re-clustered the
craniofacial-specific cells as above. UMAP non-linear dimensional
reduction was used to visualize clusters and gene expression.
Imaging and statistical analysis
Fluorescence imaging of live zebrafish and sections was performed using a
Zeiss LSM800 confocal microscope and Zen software or Leica SP8
confocal and Leica LAS software. Skeletal preparations were imaged using
Leica S8APO and DM2500 microscopes. Image processing was performed
with the same settings used among all images from each experiment using
Fiji (Schindelin et al., 2012) and Adobe Photoshop. Proliferating PCNA+
cells and total Hoechst+ nuclei at the jaw joint and ceratohyal growth plates
were counted using the cell counter function of Fiji and expressed as a %
PCNA+/total cells. The regions selected for proliferation analysis were
identified by wild-type joint morphology and/or by performing RNAscope
staining for nkx3.2 in adjacent 5 µm sections (e.g. Fig. 3G,H PCNA staining
versus Fig. 6I,J nkx3.2 expression). Fluorescence intensity of sesn1 and
ptgdsb.1 transcripts was measured in at least two sections per animal in Fiji
using the ‘Measure’feature of a region of interest defined by co-expression
with nkx3.2 and shown as mean intensity/area. Statistical analysis was
performed in GraphPad Prism software using unpaired two-tailed t-test
function (proliferation analysis) or unpaired two-tailed t-test with Welch’s
correction (fluorescence intensity analysis) and represented in the scatter dot
plot as individual data points with mean±s.d.
Acknowledgements
We thank Megan Matsutani and Jennifer DeKoeyer Crump for fish care; Jeffrey Boyd
at the University of Southern California Stem Cell Flow Cytometry Core for FACS;
Nellie Nelson for single-cell RNA-seq library preparation; the University of Southern
California’s Center for High-Performance Computing (HPC) for single-cell RNA-seq
data alignments; and Ted Allison for his helpful feedback and facilitating our
collaboration.
Competing interests
The authors declare no competing or financial interests.
Author Contributions
Conceptualization: J.S., J.G.C.; Methodology: J.S., N.N., T.M., P.B.; Software: J.S.,
P.F.; Validation: J.S., N.N., A.N.K., P.F.; Formal analysis: J.S., N.N., A.N.K., T.M.,
P.B., D.G., J.G.C.; Investigation: J.S., N.N., A.N.K., T.M.,P.B., P.F.; Resources: P.F.,
D.G., J.G.C.; Data curation: J.S., P.F.; Writing - original draft: J.S., J.G.C.; Writing -
review & editing: J.S., P.F., J.G.C.; Visualization: J.S., P.B.; Supervision: D.G.,
J.G.C.; Project administration: J.G.C.; Funding acquisition: J.S., D.G., J.G.C.
10
RESEARCH ARTICLE Development (2021) 148, dev193409. doi:10.1242/dev.193409
DEVELOPMENT

Funding
This work was supported by National Institutes of Health grants R00DE027218 (to
J.S.) and R35DE027550 (to J.G.C), and Natural Sciences and Engineering
Research Council of Canada grant RGPIN-2014-06311 (to D.G.). Deposited in PMC
for release after 12 months.
Data availability
RNA-seq files have been deposited in GEO under accession number GSE151354.
Supplementary information
Supplementary information available online at
https://dev.biologists.org/lookup/doi/10.1242/dev.193409.supplemental
Peer review history
The peer review history is available online at
https://dev.biologists.org/lookup/doi/10.1242/dev.193409.reviewer-comments.pdf
References
Ahn, D. and Ho, R. K. (2008). Tri-phasic expression of posterior Hox genes during
development of pectoral fins in zebrafish: implications for the evolution of
vertebrate paired appendages. Dev. Biol. 322, 220-233. doi:10.1016/j.ydbio.
2008.06.032
Akazawa, H., Komuro, I., Sugitani, Y., Yazaki, Y., Nagai, R. and Noda, T. (2000).
Targeted disruption of the homeobox transcription factor Bapx1 results in lethal
skeletal dysplasia with asplenia and gastroduodenal malformation. Genes Cells 5,
499-513. doi:10.1046/j.1365-2443.2000.00339.x
Askary, A., Smeeton, J., Paul, S., Schindler, S., Braasch, I., Ellis, N. A.,
Postlethwait, J., Miller, C. T. and Crump, J. G. (2016). Ancient origin of
lubricated joints in bony vertebrates. eLife 5, e16415. doi:10.7554/eLife.16415
Askary, A., Xu, P., Barske, L., Bay, M., Bump, P., Balczerski, B., Bonaguidi,
M. A. and Crump, J. G. (2017). Genome-wide analysis of facial skeletal
regionalization in zebrafish. Development 144, 2994-3005. doi:10.1242/dev.
151712
Bi, W., Deng, J. M., Zhang, Z., Behringer, R. R. and de Crombrugghe, B. (1999).
Sox9 is required for cartilage formation. Nat. Genet. 22, 85-89. doi:10.1038/8792
Budanov, A. V. and Karin, M. (2008). p53 target genes sestrin1 and sestrin2
connect genotoxic stress and mTOR signaling. Cell 134, 451-460. doi:10.1016/j.
cell.2008.06.028
Butler, A., Hoffman, P., Smibert, P., Papalexi, E. and Satija, R. (2018). Integrating
single-cell transcriptomic data across different conditions, technologies, and
species. Nat. Biotechnol. 36, 411-420. doi:10.1038/nbt.4096
Church, V., Yamaguchi, K., Tsang, P., Akita, K., Logan, C. and Francis-West, P.
(2005). Expression and function of Bapx1 during chick limb development. Anat.
Embryol. 209, 461-469. doi:10.1007/s00429-005-0464-z
Crotwell, P. L. and Mabee, P. M. (2007). Gene expression patterns underlying
proximal-distal skeletal segmentation in late-stage zebrafish, Danio rerio.Dev.
Dyn. 236, 3111-3128. doi:10.1002/dvdy.21352
Cubbage, C. C. and Mabee, P. M. (1996). Development of the cranium and paired
fins in the zebrafish Danio rerio (Ostariophysi, Cyprinidae). J. Morphol. 229, 121-
160. doi:10.1002/(SICI)1097-4687(199608)229:2<121::AID-JMOR1>3.0.CO;2-4
Das, A. and Crump, J. G. (2012). Bmps and id2a act upstream of Twist1 to restrict
ectomesenchyme potential of the cranial neural crest. PLoS Genet. 8, e1002710.
doi:10.1371/journal.pgen.1002710
Eisen, J. S. and Smith, J. C. (2008). Controlling morpholino experiments: don’tstop
making antisense. Development 135, 1735-1743. doi:10.1242/dev.001115
Eve, A. M. J., Place, E. S. and Smith, J. C. (2017). Comparison of Zebrafish
tmem88a mutant and morpholino knockdown phenotypes. PLoS ONE 12,
e0172227. doi:10.1371/journal.pone.0172227
Gentsch, G. E., Spruce, T., Monteiro, R. S., Owens, N. D. L., Martin, S. R. and
Smith, J. C. (2018). Innate immune response and off-target mis-splicing are
common morpholino-induced side effects in Xenopus. Dev. Cell 44, 597-610.e10.
doi:10.1016/j.devcel.2018.01.022
Giovannone, D., Paul, S., Schindler, S., Arata, C., Farmer, D. T., Patel, P.,
Smeeton, J. and Crump, J. G. (2019). Programmed conversion of hypertrophic
chondrocytes into osteoblasts and marrow adipocytes within zebrafish bones.
eLife 8, e42736. doi:10.7554/eLife.42736
Hartmann, C. and Tabin, C. J. (2001). Wnt-14 plays a pivotal role in inducing
synovial joint formation in the developing appendicular skeleton. Cell 104,
341-351. doi:10.1016/S0092-8674(01)00222-7
Hellemans, J., Simon, M., Dheedene, A., Alanay, Y., Mihci, E., Rifai, L., Sefiani,
A., van Bever, Y., Meradji, M., Superti-Furga, A. et al. (2009). Homozygous
inactivating mutations in the NKX3-2 gene result in spondylo-megaepiphyseal-
metaphyseal dysplasia. Am. J. Hum. Genet. 85, 916-922. doi:10.1016/j.ajhg.
2009.11.005
Herbrand, H., Pabst, O., Hill, R. and Arnold, H.-H. (2002). Transcription factors
Nkx3.1 and Nkx3.2 (Bapx1) play an overlapping role in sclerotomal development
of the mouse. Mech. Dev. 117, 217-224. doi:10.1016/S0925-4773(02)00207-1
Huang, P., Xiao, A., Zhou, M., Zhu, Z., Lin, S. and Zhang, B. (2011). Heritable
gene targeting in zebrafish using customized TALENs. Nat. Biotechnol. 29,
699-700. doi:10.1038/nbt.1939
Hwang, W. Y., Fu, Y., Reyon, D., Maeder, M. L., Tsai, S. Q., Sander, J. D.,
Peterson, R. T., Yeh,J.-R. J. and Joung, J. K. (2013). Efficient genome editing in
zebrafish using a CRISPR-Cas system. Nat. Biotechnol. 31, 227-229. doi:10.
1038/nbt.2501
Jing, Y., Zhou, X., Han, X., Jing, J., von der Mark, K., Wang, J., de Crombrugghe,
B., Hinton, R. J. and Feng, J. Q. (2015). Chondrocytes directly transform into
bone cells in mandibular condyle growth. J. Dent. Res. 94, 1668-1675. doi:10.
1177/0022034515598135
Joris, M., Schloesser, M., Baurain, D., Hanikenne, M., Muller, M. and Motte, P.
(2017). Number of inadvertent RNA targets for morpholino knockdown in Danio
rerio is largely underestimated: evidence from the study of Ser/Arg-rich splicing
factors. Nucleic Acids Res. 45, 9547-9557. doi:10.1093/nar/gkx638
Kawato, Y.,Hirao, M., Ebina, K., Tamai, N., Shi, K., Hashimoto, J., Yoshikawa, H.
and Myoui, A. (2011). Nkx3.2-induced suppression of Runx2 is a crucial mediator
of hypoxia-dependent maintenance of chondrocyte phenotypes. Biochem.
Biophys. Res. Commun. 416, 205-210. doi:10.1016/j.bbrc.2011.11.026
Kessels, M. Y., Huitema, L. F. A., Boeren, S., Kranenbarg, S., Schulte-Merker,
S., van Leeuwen, J. L. and de Vries, S. C. (2014). Proteomics analysis of the
zebrafish skeletal extracellular matrix. PLoS ONE 9, e90568. doi:10.1371/journal.
pone.0090568
Kok, F. O., Shin, M., Ni, C.-W., Gupta, A., Grosse, A. S., van Impel, A.,
Kirchmaier, B. C., Peterson-Maduro, J., Kourkoulis, G., Male, I. et al. (2015).
Reverse genetic screening reveals poor correlation between morpholino-induced
and mutant phenotypes in zebrafish. Dev. Cell 32, 97-108. doi:10.1016/j.devcel.
2014.11.018
Kozhemyakina, E., Zhang, M., Ionescu, A., Ayturk, U. M., Ono, N., Kobayashi,
A., Kronenberg, H., Warman, M. L. and Lassar, A. B. (2015). Identification of a
prg4- expressing articular cartilage progenitor cell population in mice. Arthritis
Rheumatol. 67, 1261-1273. doi:10.1002/art.39030
Kronenberg, H. M. (2003). Developmental regulation of the growth plate. Nature
423, 332-336. doi:10.1038/nature01657
Law, S. H. W. and Sargent, T. D. (2014). The serine-threonine protein kinase PAK4
is dispensable in zebrafish: identification of a morpholino-generated
pseudophenotype. PLoS ONE 9, e100268. doi:10.1371/journal.pone.0100268
Lawson, N. D. and Weinstein, B. M. (2002). In vivo imaging of embryonic vascular
development using transgenic zebrafish. Dev. Biol. 248, 307-318. doi:10.1006/
dbio.2002.0711
Li, J., Luo, H., Wang, R., Lang, J., Zhu, S., Zhang, Z., Fang, J., Qu, K., Lin, Y.,
Long, H. et al. (2016). Systematic reconstruction of molecular cascades
regulating GP development using single-cell RNA-seq. Cell Rep. 15,
1467-1480. doi:10.1016/j.celrep.2016.04.043
Lui, J. C., Chau, M., Chen, W., Cheung, C. S. F., Hanson, J., Rodriguez-Canales,
J., Nilsson, O. and Baron, J. (2015). Spatial regulation of gene expression during
growth of articular cartilage in juvenile mice. Pediatr. Res. 77, 406-415. doi:10.
1038/pr.2014.208
Lukas, P. and Olsson, L. (2018). Bapx1 is required for jaw joint development in
amphibians. Evol. Dev. 20, 192-206. doi:10.1111/ede.12267
McInnes, L., Healy, J., Saul, N. and Großberger, L. (2018). UMAP: uniform
manifold approximation and projection. J. Open Source Software 3, 861. doi:10.
21105/joss.00861
Miller, C. T., Yelon, D., Stainier, D. Y. R. and Kimmel, C. B. (2003). Two endothelin
1 effectors, hand2 and bapx1, pattern ventral pharyngeal cartilage and the jaw
joint. Development 130, 1353-1365. doi:10.1242/dev.00339
Miyashita, T., Baddam, P., Smeeton, J., Oel, A. P., Natarajan, N., Gordon, B.,
Palmer, A. R., Crump, J. G., Graf, D. and Allison, W. T. (2020). nkx3.2 mutant
zebrafish accommodate jaw joint loss through a phenocopy of the head shapes of
Paleozoic jawless fish. J. Exp. Biol. 223, jeb216945. doi:10.1242/jeb.216945
Mizuhashi, K., Ono, W., Matsushita, Y., Sakagami, N., Takahashi, A., Saunders,
T. L., Nagasawa, T., Kronenberg, H. M. and Ono, N. (2018). Resting zone of the
growth plate houses a unique class of skeletal stem cells. Nature 563, 254-258.
doi:10.1038/s41586-018-0662-5
Moffatt, P., Gaumond, M.-H., Salois, P., Sellin, K., Bessette, M.-C., Godin, E., de
Oliveira, P. T., Atkins, G. J., Nanci, A. and Thomas, G. (2008). Bril: a novel
bone-specific modulator of mineralization. J. Bone Miner. Res. 23, 1497-1508.
doi:10.1359/jbmr.080412
Newton, P. T., Li, L., Zhou, B., Schweingruber, C., Hovorakova, M., Xie, M., Sun,
X., Sandhow, L., Artemov, A. V., Ivashkin, E. et al. (2019). A radical switch in
clonality reveals a stem cell niche in the epiphyseal growth plate. Nature 567,
234-238. doi:10.1038/s41586-019-0989-6
Pradhan, A. and Olsson, P.-E. (2014). Juvenile ovary to testis transition in
Zebrafish involves inhibition of Ptges. Biol. Reprod. 91, 33. doi:10.1095/
biolreprod.114.119016
Provot, S., Kempf, H., Murtaugh, L. C., Chung, U.-I., Kim, D.-W., Chyung, J.,
Kronenberg, H. M. and Lassar, A. B. (2006). Nkx3.2/Bapx1 acts as a negative
regulator of chondrocyte maturation. Development 133, 651-662. doi:10.1242/
dev.02258
11
RESEARCH ARTICLE Development (2021) 148, dev193409. doi:10.1242/dev.193409
DEVELOPMENT

Sander, J. D., Cade, L., Khayter, C., Reyon, D., Peterson, R. T., Joung, J. K. and
Yeh, J.-R. J. (2011). Targeted gene disruption in somatic zebrafish cells using
engineered TALENs. Nat. Biotechnol. 29, 697-698. doi:10.1038/nbt.1934
Schindelin, J., Arganda-Carreras, I., Frise, E., Kaynig, V., Longair, M., Pietzsch,
T., Preibisch, S., Rueden, C., Saalfeld, S., Schmid, B. et al. (2012). Fiji: an
open-source platform for biological-image analysis. Nat. Methods 9, 676-682.
doi:10.1038/nmeth.2019
Sukarieh, R., Sonenberg, N. and Pelletier, J. (2009). The eIF4E-binding proteins
are modifiers of cytoplasmic eIF4E relocalization during the heat shock response.
Am. J. Physiol. Cell Physiol. 296, C1207-C1217. doi:10.1152/ajpcell.00511.2008
Taba, Y., Sasaguri, T., Miyagi, M., Abumiya, T., Miwa, Y., Ikeda, T. and
Mitsumata, M. (2000). Fluid shear stress induces lipocalin-type prostaglandin D
2
synthase expression in vascular endothelial cells. Circ. Res. 86, 967-973. doi:10.
1161/01.RES.86.9.967
Takeda, S., Bonnamy, J.-P., Owen, M. J., Ducy, P. and Karsenty, G. (2001).
Continuous expression of Cbfa1 in nonhypertrophic chondrocytes uncovers its
ability to induce hypertrophic chondrocyte differentiation and partially rescues
Cbfa1-deficient mice. Genes Dev. 15, 467-481. doi:10.1101/gad.845101
Talbot, J. C., Johnson, S. L. and Kimmel, C. B. (2010). hand2 and Dlx genes
specify dorsal, intermediate and ventral domains within zebrafish pharyngeal
arches. Development 137, 2507-2517. doi:10.1242/dev.049700
Tribioli, C., Frasch, M. and Lufkin, T. (1997). Bapxl: an evolutionary conserved
homologue of the Drosophila bagpipe homeobox gene is expressed in splanchnic
mesoderm and the embryonic skeleton. Mech. Dev. 65, 145-162. doi:10.1016/
S0925-4773(97)00067-1
Tribioli, C. and Lufkin, T. (1999). The murine Bapx1 homeobox gene plays a critical
role in embryonic development of the axial skeleton and spleen. Development
126, 5699-5711.
Tsukumo, Y., Alain, T., Fonseca, B. D., Nadon, R. and Sonenberg, N. (2016).
Translation control during prolonged mTORC1 inhibition mediated by 4E-BP3.
Nat. Commun. 7, 11776. doi:10.1038/ncomms11776
Tucker, A. S., Watson, R. P., Lettice, L. A., Yamada, G. and Hill, R. E. (2004).
Bapx1 regulates patterning in the middle ear: altered regulatory role in the
transition from the proximal jaw during vertebrate evolution. Development 131,
1235-1245. doi:10.1242/dev.01017
Wilson, J. and Tucker, A. S. (2004). Fgf and Bmp signals repress the expression of
Bapx1 in the mandibular mesenchyme and control the position of the developing
jaw joint. Dev. Biol. 266, 138-150. doi:10.1016/j.ydbio.2003.10.012
Yamashita, S., Andoh, M., Ueno-Kudoh, H., Sato, T., Miyaki, S. and Asahara, H.
(2009). Sox9 directly promotes Bapx1 gene expression to repress Runx2 in
chondrocytes. Exp. Cell Res. 315, 2231-2240. doi:10.1016/j.yexcr.2009.03.008
Yan, Y.-L., Willoughby, J., Liu, D., Crump, J. G., Wilson, C., Miller, C. T., Singer,
A., Kimmel, C., Westerfield, M. and Postlethwait, J. H. (2005). A pair of Sox:
distinct and overlapping functions of zebrafish sox9 co-orthologs in craniofacial
and pectoral fin development. Development 132, 1069-1083. doi:10.1242/dev.
01674
Yang, L., Tsang, K. Y., Tang, H. C., Chan, D. and Cheah, K. S. E. (2014).
Hypertrophic chondrocytes can become osteoblasts and osteocytes in
endochondral bone formation. Proc. Natl. Acad. Sci. USA 111, 12097-12102.
doi:10.1073/pnas.1302703111
Yogev, O., Williams, V. C., Hinits, Y. and Hughes, S. M. (2013). eIF4EBP3L acts
as a gatekeeper of TORC1 in activity-dependent muscle growth by specifically
regulating Mef2ca translational initiation. PLoS Biol. 11, e1001679. doi:10.1371/
journal.pbio.1001679
Zeng, L., Kempf, H., Murtaugh, L. C., Sato, M. E. and Lassar, A. B. (2002). Shh
establishes an Nkx3.2/Sox9 autoregulatory loop that is maintained by BMP
signals to induce somitic chondrogenesis. Genes Dev. 16, 1990-2005. doi:10.
1101/gad.1008002
Zhao, H., Zhou, W., Yao, Z., Wan, Y., Cao, J., Zhang, L., Zhao, J., Li, H., Zhou, R.,
Li, B. et al. (2015). Foxp1/2/4 regulate endochondral ossification as a suppresser
complex. Dev. Biol. 398, 242-254. doi:10.1016/j.ydbio.2014.12.007
Zhou, X., von der Mark, K., Henry, S., Norton, W., Adams, H. and de
Crombrugghe, B. (2014). Chondrocytes transdifferentiate into osteoblasts in
endochondral bone during development, postnatal growth and fracture healing in
mice. PLoS Genet. 10, e1004820. doi:10.1371/journal.pgen.1004820
12
RESEARCH ARTICLE Development (2021) 148, dev193409. doi:10.1242/dev.193409
DEVELOPMENT

Supplemental Figure 1
Figure S1. nkx3.2 mutant zebrafish develop spinal abnormalities at 30 dpf. Lateral and
dorsal views of µCT imaging of 30 dpf wild-type (WT, A) and nkx3.2ua5011/ua5011 mutant (B) fish
reveals spinal abnormalities in mutants, including a kink in the caudal spine. Scale bar = 2mm.
Development: doi:10.1242/dev.193409: Supplementary information
Development • Supplementary information

Supplemental Figure 2
Figure S2. Fluorescence activated cell sorting (FACS) for single-cell isolation of fli1a:GFP
+;sox10:DsRed+ populations. Confocal images to the left show fli1a:GFP+;sox10:DsRed+
chondrocytes in the jaw region at 21 dpf. FACS plots to the right show gating for GFP (x-axis)
and DsRed (y-axis), with cells in the upper right quadrant (boxed) used for single-cell analysis.
Development: doi:10.1242/dev.193409: Supplementary information
Development • Supplementary information

Supplemental Figure 3
Figure S3. Identification of fin and gill populations in single-cell RNA-seq analysis. Fin
and gill cells are present in unique clusters in the UMAP visualization of all profiled cells. Gill
cells are marked by high levels of expression of ucmaa and ppp1r1c. Fin cells are marked by
expression of posterior hox9-13 genes. Although we attempted to isolate only head
chondrocytes, this analysis reveals some contamination by pectoral fin chondrocytes during the
dissection process.
Development: doi:10.1242/dev.193409: Supplementary information
Development • Supplementary information