Nubp2 is required for cranial neural crest survival in the mouse.

Abstract

The N-ethyl-N-nitrosourea (ENU) forward genetic screen is a useful tool for the unbiased discovery of novel mechanisms regulating developmental processes. We recovered the dorothy mutation in such a screen designed to recover recessive mutations affecting craniofacial development in the mouse. Dorothy embryos die prenatally and exhibit many striking phenotypes commonly associated with ciliopathies, including a severe midfacial clefting phenotype. We used exome sequencing to discover a missense mutation in nucleotide binding protein 2 (Nubp2) to be causative. This finding was confirmed by a complementation assay with the dorothy allele and an independent Nubp2 null allele (Nubp2^null). We demonstrated that Nubp2 is indispensable for embryogenesis. NUBP2 is implicated in both the cytosolic iron/sulfur cluster assembly pathway and negative regulation of ciliogenesis. Conditional ablation of Nubp2 in the neural crest lineage with Wnt1-cre recapitulates the dorothy craniofacial phenotype. Using this model, we found that the proportion of ciliated cells in the craniofacial mesenchyme was unchanged, and that markers of the SHH, FGF, and BMP signaling pathways are unaltered. Finally, we show evidence that the phenotype results from a marked increase in apoptosis within the craniofacial mesenchyme.

Summary Statement

An ENU screen identifies a novel allele of Nubp2 and a requirement in cranial neural crest survival and proper midfacial development.

Introduction

Vertebrate craniofacial development is a dynamic process highly dependent on a fascinating cell population known as the neural crest. Craniofacial neural crest cells (CNCC) are a multipotent and migratory cell type which delaminate from the dorsal aspect of the closing neural tube and travel ventrolaterally into the frontonasal prominence and pharyngeal arches where they become the mesenchyme of the emerging facial primordia [1, 2]. On each side of the mouse face, proliferation of the post-migratory neural crest causes medial and lateral nasal prominences to swell from the frontonasal prominence around embryonic day (E) 10, resulting in paired horseshoe-shaped outgrowths with visible depressed nasal pits in the center. The outgrowth of these prominences is coupled with that of the maxillary prominences to bring the three primordia into contact [3, 4]. This contact and the subsequent fusion of the medial nasal prominence, lateral nasal prominence, and maxillary prominence create the nostrils and upper lip [5]. As development continues past E10.5, further growth brings the paired medial nasal prominences into contact at the midface, where they fuse and give rise to the nasal septum, philtrum, and premaxilla.

During these early stages of craniofacial development, the CNCC rely on signaling cues from surrounding tissues to regulate their survival and the proliferation. These include Sonic Hedgehog (SHH) and Fibroblast Growth Factor 8 (FGF8) secreted by the surface ectoderm [6-9]. Loss of these signals, or an inability of CNCC to properly transduce ligand binding within the cell, impairs the outgrowth and subsequent fusion of the facial primordia, often ultimately leading to midline defects of the face [10]. The CNCC are also a particularly vulnerable cell population as evidenced by their marked susceptibility to ribosome biogenesis insults, sensitivity to the teratogenic effects of alcohol and many examples of genetic mutations leading to elevated CNCC apoptosis [11-17].

Despite significant advances in understanding the signaling pathways and transcriptional networks involved in craniofacial developmental disorders, their etiology often remains unclear. Unbiased forward genetics approaches in model organisms allow new insights and investigations into previously unexamined roles for genes and pathways in embryogenesis [18]. Here we report a new allele from an N-ethyl-N-nitrosourea (ENU) screen in the mouse that is integral to craniofacial development. We found the causal mutation to be in the gene nucleotide binding protein 2 (Nubp2). Nubp2 is a highly conserved P-loop NTPase first described in early sequencing and mapping of cDNA isolated from E7.5 mouse embryos [19]. In Saccharomyces cerevisiae it was identified as cytosolic Fe-S cluster deficient (Cfd1) in a screen for genes able to convert Iron regulatory protein 1 into c-aconitase [20] [21]. These and subsequent studies in yeast and animal cell lines established that NUBP2 is an integral component of the cytosolic iron-sulfur cluster assembly pathway where it acts along with its homologous binding partner NUBP1 as a scaffold for transfer of the iron-sulfur cofactor to non-mitochondrial apoproteins [22-24]. Later studies revealed that knockdown of Nubp1, or both Nubp1 and Nubp2, led to excessive centrosome duplication in vitro [25]. Further work indicated that silencing of Nubp1 and Nubp2 in cell culture led to supernumerary cilia formation [26]. An ENU screen isolated a mouse allele of Nubp1 with interrupted lung budding and centriole over-duplication [27].

This implication of Nubp2 in regulation of centriole duplication and ciliary dynamics, and the importance of the primary cilia to intercellular signaling, led us to hypothesize aberrant ciliogenesis would disrupt crucial CNCC signaling pathways, resulting in loss of mitogenic and/or survival signals and subsequent midfacial fusion defects. On the contrary, here we demonstrate that loss of Nubp2 from the CNCC does not lead to ciliogenesis or signaling defects, but causes a rapid onset of apoptosis throughout the craniofacial mesenchyme following CNCC migration which underlies the distinctive craniofacial presentation of the dorothy mutant. This represents the first example of a role for iron-sulfur cluster assembly pathway components in midfacial development.

Materials and Methods

Animal husbandry

All animals were housed and cared for in accordance with the Guide for the Care and Use of Laboratory Animals (National Institutes of Health), and all animal-associated activity was approved by the Institutional Animal Care and Use Committee of the Cincinnati Children’s Hospital Medical Center (Protocol # IACUC2016-0098). All animals were monitored daily by registered veterinary technicians and additional care provided where necessary. Euthanasia was performed via Isoflurane sedation followed by cervical dislocation and secondary euthanasia. Mice were housed in ventilated cages provided with water and chow ad libitum. During timed mating, noon of the day a copulation plug was detected was considered embryonic day (E) 0.5 (E0.5).

ENU mutagenesis

ENU mutagenesis was performed as previously published [18]. Wildtype C57BL/6J (B6; Jackson Laboratory, Bar Harbor ME) 6- to 8-week-old males were given a fractionated dose of ENU (three weekly dosages of 90-100 mg/kg ENU, Sigma). Mutagenized males were bred with Tg(Etv1-EGFP)^{B2192 GSat} females (FVB/N TaC background) as part of a larger effort to identify potential mutants in cortical neuron patterning [28]. Following a three-generation backcross breeding scheme [18], G3 litters were dissected at late embryonic stages to detect gross developmental anomalies. The allele was propagated with test crosses onto the FVB background (to facilitate mapping if needed) until the causal allele was identified.

Variant detection

Exome sequencing was carried out in the CCHMC Genetic Variation and Gene Discovery Core Facility (Cincinnati, OH, USA). An exome library was prepared from genomic DNA using the NimbleGen EZ Exome V2 kit (Madison, WI, USA). Sequencing of the resulting exome library was done with an Illumina Hi Seq 2000 (San Diego, CA, USA). The Broad Institute’s GATK algorithm was used for alignment and variant detection. Variants were then filtered as described in Table 3. Mutations were validated by Sanger sequencing.

Table 3:

Explanation of exome sequencing result filtering

Total variants	176,315
Filter out non-homozygous variants	36,590
Filter out non-SNP variants	30,733
Filter out variants listed in dbSNP	5,800
Filter out "Low impact" variants	267
Filter out genes with multiple variants	116
Filter out known polymorphisms	4
Filter out variants seen in CCHMC control exomes	3
Filter out genes with no null phenotype	2
Filter out variant that doesn't segregate with phenotype	1

Alleles and genotyping

Nubp2^Dor (Nubp2 c.626T>A; Nubp2^V209D; dorothy) was discovered in an ENU mutagenesis screen as described above and maintained on a mixed background. Dorothy genotyping was performed using the TaqMan™ Sample-to-SNP™ Kit using proprietary primers designed by the vendor targeting Nubp2:c:626T>A or Zbtb12:c:1061C>A (Thermo Fisher Scientific, Waltham, MA, USA). The Nubp2^tm1a (Nubp2^null) allele was imported from the European Conditional Mouse Mutagenesis Program (EUCOMM, EM:08954) [29] using standard in vitro fertilization protocols. We excised the reporter gene trap cassette to create a traditional floxed allele Nubp2^tm1c (Nubp2^fx) by crossing the Nubp2^tm1a with carriers of the Flippase enzyme [30]. PCR primers amplified a 470bp amplicon unique to the Nubp2^tm1a allele (Nubp2^tm1a F: TCATGCTGGAGTTCTTCGCC, Nubp2^tm1a R: ATTCTTCCGCCTACTGCGAC). A separate primer pair amplify a 263bp sequence in the wildtype Nubp2 allele and 457bp band in the Nubp2^fx allele (Nubp2WtFxF: GGATCCCAGTGCTGAGCTTT, Nubp2WtFxR: ACCACCTGCTAGCACTCAAC). For conditional ablations, we used the Wnt1-cre allele (B6.Cg-H2afv^{Tg(Wnt1-cre)11RthTg(Wnt1-GAL4)11Rth}/J) [31] which was genotyped using a standard Cre detection PCR reaction (CreF: GCGGTCTGGCAGTAAAAACTATC, CreR: GTGAAACAGCATTGCTGTCACTT). Visualization of neural crest cells targeted by Wnt1-cre was accomplished using the mT/mG double-fluorescent Cre reporter allele developed by Muzumdar et al. [32].

Cell culture

Human Embryonic 293T (HEK293T) cells were cultured in Dulbecco’s modified eagle medium (DMEM, Sigma-Aldrich) supplemented with 10% Fetal Bovine Serum and 1% Penicillin-Streptomycin (Sigma-Aldrich). C- and N- terminal GFP-tagged Nubp1 and Nubp2 expression plasmids [33] were transfected using Lipofectamine® 3000 (Thermo Fisher Scientific, Waltham, MA, USA) and imaged using a Nikon C2 Confocal microscope.

Histology and Immunohistochemistry

Embryos designated for hematoxylin and eosin (H&E) staining were dissected and fixed overnight in 10% neutral buffered formalin solution (Sigma-Aldrich) at room temperature (RT) and then dehydrated in 70% ethanol for 48h. Following fixation and dehydration, samples were submitted to the CCHMC Pathology Research Core (Cincinnati, OH, USA) for paraffin embedding. Embedded samples were sectioned at a thickness of 10-14μM, stained with hematoxylin and eosin, and sealed with Cytoseal Mounting Medium (Thermo-Scientific). Samples designated for Immunohistochemical (IHC) analysis were dissected and fixed overnight in 4% paraformaldehyde at 4°C followed by immersion in 30% sucrose solution at 4°C for 24-48 hours until sufficiently dehydrated as observed by changes in sample buoyancy. Embryos were then embedded sagittally (E9.5) or coronally (E10.5) in OCT compound (Tissue-Tek) and cryosectioned at a thickness of 10-14μM. Sections were dried on a slide warmer at 42°C for 30-45 minutes for general IHC or dried at room temperature for 15-30 minutes when cilia were to be imaged. Samples were then soaked in boiling citrate buffer for 45-60 minutes for antigen retrieval followed by three PBS washes and blocked for 1 hour at RT in 4% normal goat serum in PBST. After blocking, sections were incubated overnight at 4°C with the following primary antibodies as indicated: Cleaved Caspase 3 (1:400, Cell Signaling 9661), PHH3 (1:500, Sigma H0412), AP2 (1:20, Developmental Studies Hybridoma Bank), Laminin (1:250, Sigma L9393), γ-Tubulin (1:1000, Sigma T6557), Arl13b (1:250, Proteintech 17711-1-AP). The following morning, sections were washed three times in PBS and incubated at RT in AlexaFluor 488 goat anti-mouse and AlexaFluor 594 goat anti-rabbit secondary (1:500, Life Technologies) for 1 hour at RT, washed 3 times in PBS and incubated in DAPI (1:1000, Sigma) for 20 minutes at RT. After 3 more PBS washes slides were sealed with ProLong Gold Antifade Reagent (Life Technologies) and imaged using a Nikon C2 confocal microscope. IHC images depicting cell death and proliferation were quantified using ImageJ (Rasband, W.S., ImageJ, U. S. National Institutes of Health, Bethesda, Maryland, USA, https://imagej.nih.gov/ij/, 1997-2018). Quantification of primary cilia was performed using NIS elements software (Nikon Instruments, Melville, NY, USA).

In Situ Hybridization

Whole mount in situ hybridization was performed as described previously [34] with a hybridization temperature of 65°C. Probes were transcribed from linearized plasmids using T7 RNA polymerase: Etv5 (Xin Sun/ Gail Martin), Ptc1 [35], Msx1 [36], FGF8 [37], and Sox10 [38]

RNA Sequencing

3 WT and 3 Wnt1-cre; Nubp2^cKO mutant E9.5 embryos were dissected from the yolk sac, and heads were snap frozen on dry ice. RNA was isolated and the individual samples were used for paired-end bulk-RNA sequencing (BGI-Americas, Cambridge, MA). Results for one of the 3 mutants were of low quality and were discarded at the start of analysis. RNA-Seq analysis pipeline steps were performed using CSBB [Computational Suite for Bioinformaticians and Biologists: https://github.com/csbbcompbio/CSBB-v3.0]. CSBB has multiple modules, RNA-Seq module is focused on carrying out analysis steps on sequencing data, which comprises of quality check, alignment, quantification and generating mapped read visualization files. Quality check of the sequencing reads was performed using FASTQC (http://www.bioinformatics.bbsrc.ac.uk/projects/fastqc). Reads were mapped (to mm10 version of Mouse genome) and quantified using RSEM-v1.3.0 [39].

RESULTS

Dorothy mutants were recovered in an ENU mutagenesis experiment.

We performed a forward Ethyl-N-Nitrosourea (ENU) genetic screen in the mouse to discover new alleles important for craniofacial development [18]. Dorothy (dor) mutants were isolated from this screen and readily identified by a characteristic craniofacial phenotype (Figure 1, Tables 1 and 2). At E16.5, dorothy mutants displayed a striking midfacial cleft phenotype in which the rostral-most nasal cartilage appears to be absent while the remaining nasal and maxillary structures were shifted laterally (Figure 1: A-F). We also noted micromelia and oligodactyly in each mutant embryo and commonly observed blood pooling in the distal hindlimbs reminiscent of the eponymous ruby slippers [40] (Table 2). While all surviving embryos (6/6) collected at E16.5 had midline clefting and truncated limb phenotypes, mandibular development was variably affected and ranged from apparently normal development in 3/6 to complete agenesis in 1/6 non-necrotic E16.5 embryos observed. Curiously, despite the dysmorphic nature of the dorothy face, histological analysis revealed the presence of tooth buds in relatively appropriate positions, albeit with slightly delayed development (Figure 1: C, F, white dotted outlines). We also noted a significant portion of dorothy mutants did not survive beyond organogenesis stages. These commonly featured extensive hematomas and appeared morphologically to have arrested around E9.0-E9.5 (Figure 1: G-N, Table 1,2). During maintenance of the dorothy colony, living embryos past E10.5 appropriate for phenotypic analysis became increasingly rare. We hypothesize this was due to the increasingly inbred genetic background as the propagation of line was initially onto an FVB background in the event genetic mapping would be needed to clone the causal gene. The initial B6/FVB hybrid environment in the first matings may have led to increased dorothy mutant survival which declined with subsequent homogeneity on the B6 background.

Figure 1: The Dorothy mutants have variable craniofacial phenotypes.

Whole-mount and histological analysis of E16.5 control (A-C) and dorothy (D-F) mutant embryos. Mutants consistently present with midfacial clefting, micromelia and oligodactyly. The nasal cartilage (black arrows) is underdeveloped and ectopically located. Accumulation of blood or other fluid under the epithelium are also often noted (red asterisk). (F) Coronal sections also reveal cleft palate (black asterisks). Tooth buds are present but tooth development appears delayed (yellow dotted outlines). (G-N) Dorothy mutants with more severe phenotypes are developmentally arrested at the start of organogenesis (G-M), and often show signs of pervasive hematoma (red arrows). (Scale bars in B,C,E,F = 1mm, G-N = 200μm).

Table 1:

Recovery of dorothy mutants at different stages of embryogenesis

	E9.5		E10.5		E11.5		E16.5
Genotype	Observed	Expected	Observed	Expected	Observed	Expected	Observed	Expected
Nubp2+/+	23	20.25	34	30.5	6	5	51	45.25
Nubp2dor/+	40	40.5	69	61	10	10	120	90.5
Nubp2dor/dor	18	20.25	19	30.5	4	5	10	45.25
	p= 0.730		p= 0.055		p= 0.819		p= 6.172E-09

Table 2:

dorothy phenotyping at different stages of embryogenesis

E9.5
Term	MGI ID	(n)	(%)
hematoma	MP: 0008817	13	72.22%
abnormal first pharyngeal arch morphology	MP: 0006337	11	61.11%
small frontonasal prominence	MP: 0030249	7	38.89%
embryonic growth retardation	MP: 0003984	1	5.56%
E10.5
abnormal lateral nasal prominence morphology	MP: 0009902	19	100.00%
abnormal medial nasal prominence morphology	MP: 0009903	19	100.00%
hematoma	MP: 0008817	15	78.95%
embryonic growth retardation	MP: 0003984	15	78.95%
small mandibular prominence	MP: 0030346	15	78.95%
small maxillary prominence	MP: 0030342	8	42.11%
absent maxillary prominence	MP: 0030344	7	36.84%
E11.5
small mandibular prominence	MP: 0030346	4	100%
hematoma	MP: 0008817	3	75%
abnormal lateral nasal prominence morphology	MP: 0009902	3	75%
abnormal medial nasal prominence morphology	MP: 0009903	3	75%
abnormal eye development	MP: 0001286	3	75%
absent maxillary prominence	MP: 0030344	2	50%
E16.5
embryonic lethality during organogenesis	MP: 0006207	4	40%
Remaining phenotypes scored from 6/10 embryos not clearly necrotic
abnormal digit development	MP: 0006280	6	100%
micromelia	MP: 0008736	6	100%
oligodactyly	MP: 0000565	6	100%
midline facial cleft	MP: 0000108	4	66%
embryonic lethality during organogenesis	MP: 0006207	4	66%
hematoma	MP: 0008817	3	50%
protruding tongue	MP: 0009908	3	50%
absent snout	MP: 0000450	2	33%
absent mandible	MP: 0000087	2	33%
embryonic growth arrest	MP: 0001730	1	16%
encephalomeningocele	MP: 0012260	1	16%

The dorothy allele is a missense mutation in nucleotide binding protein 2 (Nupb2).

We performed exome sequencing to identify the causal allele in dorothy mutants. We sequenced three mutants and initially identified approximately 176,000 variants after quality control and read depth filters were applied. We first filtered out variants that were heterozygous in one or more mutants or that didn’t result from single nucleotide changes working under the assumption the dorothy allele is a recessive SNP, bringing the list of potential variants from ~176,000 to ~30,000 (Table 3). After filtering out SNPs known to be strain polymorphisms and variants predicted to have “low” impact, along with genes containing multiple variants (assumed to be highly polymorphic genes), 3 variants remained. One of these was in Pbx2 which has been deleted in mouse with no resulting phenotype [41]. The remaining variants were in nucleotide binding protein 2 (Nubp2) and zinc finger and BTB domain containing 12 (Zbtb12). We further genotyped ten mutants and only homozygosity for the Nubp2 was completely concordant with the phenotype as the Zbtb12 mutation was also homozygous in 9 unaffected embryos (Table 3). Thus, Nubp2:c.626T>A; p.Val209Asp was identified as a candidate causal variant.

The Val209Asp missense mutation changes the coding for a highly conserved nonpolar valine into a charged aspartic acid residue five amino acids downstream from a pair of iron-interacting cysteine residues [23] (Figure 2A). Sanger sequencing confirmed the nucleotide change c626T>A in dorothy embryos (Figure 2B). Transfection of HEK293T cells with GFP-tagged hNUBP2^wt and hNUBP2^Dor expression constructs [33] did not identify a change in translation or cellular localization of hNUBP2^V209D (Figure 2C,D).

Figure 2: The dorothy mutation is an allele of Nubp2.

(A) Amino acid sequence of NUPB2 in multiple model organisms. Nonpolar side chains (valine and isoleucine) at the position of the affected residue (red box, highlighted in yellow) are heavily conserved and located in close proximity to known functional Iron-interaction sites (highlighted in purple). In the dorothy mutant (p.Val209Asp; NUBP2^V209D), this is replaced with an aspartate residue (red) harboring an electrically charged side chain. (B) Sanger sequencing confirms that Nubp2:c.626T>A is homozygous in embryos with the dorothy phenotype. (C,D) transfection of HEK293T cells with pAcGFP-C1-hNUBP2 (C) and pAcGFP-C1-hNUBP2^DOR (D) expression constructs revealed that the corresponding fusion proteins were both expressed and localized in roughly the same pattern, suggesting that the Nubp2^dor mutation does not preclude translation of NUBP2 protein.

While the dorothy phenotype was completely concordant with homozygosity for the NUBP2^V209D allele, we sought to further confirm that the causal allele was in Nubp2 via a complementation test with an independently derived null allele of Nubp2. We imported the Nubp2^tm1a (Nubp2^null) allele from the international mouse phenotyping consortium [29, 42]. We first demonstrated that the Nubp2^null/null embryos were not viable and were never recovered at early organogenesis stages (Table 4). Consistent with this, we noted frequent resorptions during these experiments, consistent with an early developmental lethality phenotype. We then intercrossed Nubp2^dor/wt with Nubp2^null/wt and never recovered Nubp2^null/dor embryos at E16.5 (Table 5). This failure of the null allele to complement the ENU allele is the strongest possible genetic evidence that dorothy is an allele of Nupb2.

Table 4:

Embryonic lethality of Nubp2 ablation

E7.5-E10.5 embryo recovery
Genotype	Expected	Observed
Nubp2+/+	8.75	20
Nubp2Null/+	17.5	15
Nubo2Null/Nul1	8.75	0
	p= 7.613E-06

Table 5:

Complementation test of Nubp2^dor as the causative allele

Embryo recovery
	E16.5		E10.5
Genotype	Expected	Observed	Expected	Observed
Nubp2+/+	4.25	6	7.25	14
Nubp2Null/+	4.25	3	7.25	8
Nubp2dor/+	4.25	8	7.25	7
Nubp2Null/dor	4.25	0	7.25	0
	p= 0.0344		p= 0.00347

Nubp2 deletion from the neural crest lineage recapitulates the dorothy craniofacial phenotypes.

Given the dorothy craniofacial phenotype, we hypothesized a unique requirement for Nubp2 in the CNCC lineage. In order to test the hypothesis of cell-autonomous requirement for Nubp2 in the CNCC and circumvent the early lethality and phenotypic variability of the dorothy allele, we used Wnt1-cre to ablate Nubp2 in just the neural crest [31]. We first generated the Nubp2^tm1c (Nubp2^fx) allele by crossing the Nubp2^tm1a with a FLP transgenic [30] . We then mated Nubp2^null/wt with Wnt1-cre to generate Wnt1-cre Nubp2^null/wt. These were again mated to Nubp2^fx/wt or Nubp2^fx/fx to create the Wnt1-cre; Nubp2^cKO embryos. We also performed similar crosses to generate Wnt1-cre; Nubp^fx/fx. The two genotypes did not result in significantly different phenotypes and are collectively referred to here as Wnt1-cre; Nupb2^cKO (Supplemental Table 1). The resulting Wnt1-cre; Nupb2^cKO embryos had phenotypes remarkably similar to the dorothy homozygous mutants (Figure 3A-H). At E15.5, embryos were recovered with midfacial clefts and undersized nasal, mandibular, and maxillary structures. Histological analysis revealed severe hypoplasia of the cartilage primordia of the nasal septum along with agenesis of the lateral cartilage primordia of the nasal capsule (Figure 3C,D,G,H, black and yellow dotted outlines, respectively). Combined with the cleft palate, this seems to lead to a situation where the olfactory epithelium is continuous with the oral cavity and, as in the dorothy mutant, the brain is shifted rostrally. We next set out to determine the developmental stage at which mutants are first phenotypically distinct from littermates. Virtually all Wnt1-cre; Nubp2^cKO embryos (34/35) were anatomically indistinguishable from littermate controls at E9.5 (Figure 3: I-L; Table 6). In contrast, a majority (14/17) of conditional mutants at E10.5 are easily identified by their conspicuously undersized facial primordia (Figure 3: M-P; Table 6). Thus, the Wnt1-cre; Nubp2^cKO phenotype appears to emerge after CNCC migration, during early expansion stages in the pharyngeal arch, and results in a loss of craniofacial tissue.

Figure 3: Conditional ablation of Nubp2 in the neural crest lineage causes midfacial defects and recapitulates the craniofacial phenotype of the dorothy mutant.

(A-H) Wnt1-cre; Nubp2^cKO embryos recovered at E15.5 have severe craniofacial abnormalities including midline clefting and mandibular hypoplasia. (C, G) H&E stained coronal sections demonstrate palatal shelves fail to elevate. (D, H) higher magnification reveals that the cartilage primordium of the nasal septum is severely truncated (black dotted outline), while cartilage primordium of the nasal capsule lateral to the septum (yellow dotted outline) appears to be completely absent in the mutant. (I-L) At E9.5 Wnt1-cre; Nubp2^cKO mutants are indistinguishable from wild-type (M-P) but at E10.5 the phenotype becomes apparent with hypoplastic nasal prominences, maxillary and mandibular prominences). (Scale bars in C, G = 500μm; D,H=50μm; I-P 100μm.)

Table 6:

Phenotypic description of Wnt1-cre; Nubp2^cKO embryos

E9.5
Term	MGI ID	(n)	(%)
no abnormal phenotype detected	MP: 0002169	34/35	97%
abnormal first pharyngeal arch morphology	MP: 0006337	1/35	3%
E10.5
Term	MGI ID	(n)	(%)
abnormal lateral nasal prominence morphology	MP: 0009902	14/17	82.35%
abnormal medial nasal prominence morphology	MP: 0009903	14/17	82.35%
small maxillary prominence	MP: 0030342	12/17	70.59%
small mandibular prominence	MP: 0030346	4/17	28.57%
no abnormal phenotype detected	MP: 0002169	3/17	17.65%
embryonic growth retardation	MP: 0003984	1/17	5.88%
E11.5
Term	MGI ID	(n)	(%)
abnormal lateral nasal prominence morphology	MP: 0009902	10/12	83.33%
abnormal medial nasal prominence morphology	MP: 0009903	10/12	83.33%
small maxillary prominence	MP: 0030342	10/12	83.33%
absent mandible	MP: 0000087	9/12	75.00%
hematoma	MP: 0008817	6/12	33.33%
no abnormal phenotype detected	MP: 0002169	2/12	16.67%

Deletion of Nubp2 in the CNCC does not affect centriole duplication or ciliogenesis.

Nubp2 has previously been implicated in the regulation of centriole duplication and ciliogenesis [25, 26]. We therefore hypothesized that similar disruptions in primary cilia biology may compromise the CNCC-derived mesenchyme. We used immunohistochemistry with confocal fluorescence microscopy to image and quantify centrosomes and primary cilia within the E9.5 craniofacial mesenchyme (Figure 4). We used antibodies for γ-tubulin to mark centrioles and ARL13B to highlight the ciliary axoneme. The cilia in Wnt1-cre; Nubp2^cKO mutants were qualitatively indistinct from those in littermates (Figure 4, A-H). We then quantified the number of cilia and centrioles per nucleus for conditional mutants and controls. We observed slight increase in average cilia/cell (Figure 4I, p=0.009) but no difference in centriole number (Figure 4J; p=0.422).

Figure 4: Craniofacial mesenchymal cells have normal primary cilia and centrioles.

(A-H) representitive fields showing immunostaining with fluorescent antibodies against γ-Tubulin and Arl13b in E9.5 craniofacial mesenchyme. No gross abnormalities are observed in mutant cilia (white arrows) or centrioles (white arrowheads). Mitotic cells can be observed with two centrioles (white M). (I) a minor, but statistically significant increase in the proportion of cilia per cell was measured in mutants. (J) There was no observed difference in the proportion of centrioles per cell in mutants. Scale bars: 5μm.

Loss of Nupb2 does not alter SHH, FGF or BMP signaling in craniofacial neural crest cells.

The midfacial defects found in the Wnt1-cre; Nubp2^cKO mutants suggested a number of developmental signaling pathways may be adversely affected. Appropriate levels of SHH, FGF and BMP signaling are all known to be crucial for proper CNCC development. We addressed each of these individually with whole mount in situ hybridization on selected markers as canonical transcriptional targets of each signaling pathway.

Hedgehog signaling is well known to be a survival signal for the CNCC [43]. Genetic ablation of smoothened in the neural crest using the same Wnt1-cre allele used here led to a phenotype highly reminiscent of the dorothy phenotype [6]. However, Ptch1 expression was qualitatively unaffected in Wnt1-cre; Nubp2^cKO mutants at E9.5 (Figure 5, A-D). FGF signaling is required from the surface ectoderm to the neural crest-derived facial mesenchyme during early craniofacial development [8, 9]. Etv5 as a measure of FGF signaling was expressed at normal levels (Figure 5, E-H). Finally, disturbances in BMP signaling during craniofacial development have been associated with defective fusion of the facial primordia [44], and mutation of the downstream transcription factor Msx1 causes cleft lip/palate. However, Msx1 levels also appeared unchanged in conditional mutants (Figure 5, I-L).

Figure 5: Major developmental signaling pathways are unaffected in Wnt1-cre; Nubp2^cKO embryos.

(A-L) Whole mount in situ hybridizations of E9.5 embryos and (M) normalized TPM values derived from RNA-sequencing of dissected E9.5 heads show that the SHH (A-D), FGF (E-H) and BMP (I-L) signaling pathways are unaltered at E9.5 in Wnt1-cre; Nubp2^cKO embryos (Scale bars =100μm).

In order to further assess these canonical pathways and perform an unbiased transcriptional analysis of mutants, we dissected E9.5 heads from three controls and two mutants and performed bulk RNA sequencing. We first measured levels of SHH signaling (Ptch1, Gli1, Gli2, Gli3), FGF signaling (Fgf8, Etv5, Spry1, Spry2), and BMP signaling (Msx1, Lef1) and saw no significant changes between wild-type and mutant embryos, consistent with the whole mount in situ results (Figure 5, M, Table S2). Nubp2 transcripts were reduced by approximately 60% in Wnt1-cre; Nubp2^cKO heads. This residual Nubp2 expression can likely be attributed to non-CNCC tissues which don’t express the Wnt1-cre recombinase. Additionally, mutants used for this experiment had a copy of the Nubp2^null (Nubp2^tm1a) allele, a conditional gene trap allele which could easily produce some sequence reads mapped to Nubp2 in the RNA-Seq analysis. After selecting for genes with a log fold change greater than ± 0.5 and false discovery rate ≤ 0.1, 61 genes were found to be differentially expressed (Table S2). Multiple analyses failed to identify a disrupted core molecular pathway or process from this list of differentially expressed genes.

Neural crest-specific ablation of Nubp2 does not significantly alter migration or proliferation, but increases apoptosis.

In the absence of intercellular signaling defects, we hypothesized that the failure of facial primordia to sufficiently expand may be due to differences in the ability of mutant neural crest cells to migrate, proliferate, and/or survive. To highlight migrating neural crest cells, we performed whole mount in situ hybridization on E9.5 (n=4) and E10.5 (n=2) Wnt1-cre; Nubp2^cKO embryos using a Sox10 riboprobe (supplemental Figure 1 A-D). Sox10 is expressed by migratory NCC streams at E9.5 and in the developing cranial ganglia and dorsal root ganglia at E10.5. There were no significant changes in Sox10 expression patterns between mutants and littermate controls. We also made use of the mT/mG cre reporter allele [32] to visualize the NCC distribution in Nubp2^cKO mutants at E10.5 (n=1, supplemental Figure 1 E-F). Again, the distribution of GFP signal in peripheral CNCC-derived structures Wnt1-cre; Nubp2^cKO; mT/mG was similar to littermate controls. However, loss of GFP signal is obvious in the nasal prominences and pharyngeal arches which are the affected tissues in mutants populated by post-migratory CNCC.

Following migration of the CNCC, the craniofacial mesenchyme must be highly proliferative in order to produce the maxillary, mandibular, and nasal prominences and to ensure that the paired prominences meet at the midline to form the intermaxillary segment. At E9.5 the ratio of proliferating cells in the craniofacial mesenchyme as measured by a mitotic index of Phosphorylated Histone H3 to DAPI-stained nuclei is unchanged in Wnt1-cre; Nubp2^cKO mutants (Figure 6, A-C; p=0.453). This ratio is mildly elevated at E10.5, when the lateral nasal prominences and medial nasal prominences are noticeably truncated in comparison to littermate controls (Figure 6, D-F; p=0.060).

Figure 6: Loss of Nubp2 leads to decreased survival of craniofacial neural crest.

(A-E’) immunofluorescent staining for Phosphorylated Histone 3 (PHH3) at E9.5 (A-C; p=0.453) and E10.5 (D-F; p=0.060) shows no significant change in the proportion of proliferating Wnt1-cre; Nubp2^cKO craniofacial mesenchymal cells at either stage. (G-L) in contrast, staining for CC3 at E9.5 (G-I; p=0.040) shows a slight increase in apoptotic mesenchyme and by E10.5 (J-L; p<0.0001) there is a striking and statistically significant increase. Scale bars: 100μm.

We then assayed levels of apoptosis by calculating the ratio of Cleaved Caspase 3 to DAPI-stained nuclei. We observed increased apoptosis at E9.5 (Figure 6, G-I; p=0.040) and a drastic increase at E10.5 (Figure 6, J-L; p<0.0001). This increase was more substantial in the MNPs than the LNPs (Fig. 6K). We do note that there is a wide range of rates of apoptosis in the mutants. This is consistent with the phenotypic variability where some embryos appear normal for a day or so longer in development than most littermates. In fact, we further quantified the rates of apoptosis and proliferation in “mild” and “severe” mutants and do note a striking difference in the apoptotic index (Supplemental Figure 2). Taken together, these data show CNCC lacking Nubp2 are capable of migrating into the frontonasal prominence and pharyngeal arches, correctly regulating ciliogenesis, responding properly to intercellular signaling, and proliferating at a normal rate, but undergo abrupt apoptosis between E9.5 and E10.5.

DISCUSSION

In this report, we describe our identification of the dorothy mutation and determine it to be an allele of Nubp2. The dorothy mutant has a variable phenotype with highly penetrant micromelia, syndactyly, dysmorphic nasal cartilage, and midfacial clefting (Figure 1, Tables 1, 2). Exome sequencing identified a variant in Nubp2 as a candidate causal mutation (Figure 3, Table 3). We used a null allele to show Nubp2 is required for early development (Table 4) and a complementation test to further demonstrate dorothy is a hypomorphic allele of Nubp2 (Table 5). We then determined that the dorothy craniofacial phenotype is due to a cell-autonomous requirement for Nubp2 in the neural crest lineage (Wnt1-cre; Nubp2^cKO), with a phenotype emerging between E9.5 and E10.5 (Figure 3). We used these Wnt1-cre; Nupb2^cKO embryos to show that ciliogenesis is only mildly affected in the craniofacial mesenchyme (Figure 4), despite previously demonstrated roles for Nubp2 as a regulator of ciliogenesis [26]. Several canonical signaling pathways required for craniofacial development and the proportion of proliferating cells are unaltered in Wnt1-cre; Nupb2^cKO embryos (Figure 5, 6). Ultimately, we observed a conspicuous increase in apoptosis among the post-migratory CNCC mesenchyme of the craniofacial primordia starting around E9.5 (Figure 6, G-L). We hypothesize that it is this ectopic cell death which prevents the craniofacial primordia from growing sufficiently to bring the opposing medial nasal prominences into contact, resulting in severe midfacial clefting and hypoplasia of the mandible, nasal septum and nasal capsule. This study is the first to demonstrate a requirement for Nubp2 in development with a more specific role in craniofacial development.

The discovery of this role for Nubp2 illustrates the enduring power of unbiased forward genetic screens, as it is unlikely the gene would have otherwise been selected for study in craniofacial development. To our knowledge, no clinical disorders have been associated with NUBP2 variants. Moreover, this report is the first to show a CNCC defect resulting from mutations in a member of the iron-sulfur cluster assembly pathway. During the course of this study, the International Mouse Phenotyping Consortium characterized the Nubp2^tm1b/tm1b mouse as “preweaning lethal with complete penetrance” (Nubp2^tm1b is derived from the Nubp2^null allele we described above) [42]. ENU mutagenesis creates missense mutations that are often hypo-morphic, providing more insight into how essential genes function in specific developmental processes. For instance, ENU-induced mutations of Hsd17b7 and Grhl2 lead to informative craniofacial and neural phenotypes despite the fact that total deletion of either gene is lethal before completion of organogenesis [45-48]. In many such cases the effects of point mutations cannot be inferred from knockout studies alone, and forward screens can provide a genetic model more relevant to the natural population. For example, patients with mutations in both alleles of POLR1C present with Treacher Collins Syndrome, while Plr1c^null/null mice are embryonic lethal [42, 49, 50].

NUBP1 and NUBP2 participate in an early step of the iron-sulfur cluster assembly pathway as a scaffold for the nascent cluster, which is transferred to target proteins via a process involving ATP hydrolysis [22, 23]. Depletion of Nubp2 in human cell lines compromises iron-sulfur cluster assembly pathway function [51]. The NUBP1^NISW mutant is compromised in its ability to bind NUBP2, thus preventing formation of the canonical heteromeric scaffold complex [27]. This may explain some of the phenotypic similarities between the Nubp1^Nisw/Nisw and dorothy alleles, such as micromelia, syndactyly, and defects in ocular development. Cataracts are also associated with mutations in Xpd/Ercc2 and Spry2, both of which are extramitochondrial iron-sulfur proteins [52-55]. Loss of Iop1 in the mouse, which operates downstream of NUBP1/NUBP2 in the pathway, causes lethality prior to E10.5 as does loss of Nubp2 [56, 57]. These data taken together suggest that aspects of the dorothy and Nubp1^Nisw/Nisw phenotypes may be attributed to decreased iron-sulfur cluster assembly pathway activity, but also that complete loss of the pathway may be lethal prior to organogenesis. This is consistent with our inability to recover Nubp2^null/nul1 embryos. Unfortunately, we were unable to directly assay iron sulfur cluster assembly pathway function during this study as MEFs derived from dorothy mutants did not expand in vitro. It is possible that NUBP2 homocomplexes are more important for iron-sulfur cluster maturation in the craniofacial mesenchyme, but recent in vitro studies using yeast proteins indicate that the homodimer is unable to catalyze ATP hydrolysis and therefore is presumably unable to transfer mature clusters to target proteins [58, 59]. It’s possible that in order to become catalytically active the NUBP2 homodimer requires interaction with a component not included in the cited studies or that the complex functions differently in mammals.

Another possibility to consider is if NUBP2 has acquired a secondary role during mammalian evolution. Such cases of “protein moonlighting” can be established when proteins acquire mutations or new post-translational modifications during evolution and new interactions arise [60]. One relevant example is the case of Iron regulatory protein 1. When the cytosolic iron-sulfur cluster pathway is able to act on the protein, it becomes enzymatically active as an aconitase. However, in the absence of cofactor, Iron regulatory protein 1 becomes an active RNA-binding protein modifying the turnover of target mRNAs [61]. Presumably these functions evolved as a way for the cell to couple the activity of an important iron-sulfur protein to regulation of transcripts relevant to iron homeostasis. Could Nubp2 have another function outside of the cytosolic iron-sulfur cluster assembly pathway that explains the presence of midfacial clefting in Nubp2, but not Nubp1 mutants? Our RNAseq experiment showed that pleckstrin homology like domain family A member 3 (Phlda3) transcripts were significantly upregulated in E9.5 Wnt1-cre; Nubp2^cKO heads, while transcripts related to the iron-sulfur cluster assembly pathway, ciliary function, and all major signaling pathways remained unchanged (Supplemental Table 2). PHLDA3 is a proapoptotic repressor of the oncogene Akt [62], which is consistent with the increased apoptotic index of Wnt1-cre; Nubp2^cKO craniofacial mesenchyme. The activation of a pro-apoptotic transcriptional landscape in CNCC at an embryonic stage before any other significant changes are detectable could indicate a more direct role for NUBP2 than previously considered. NUBP2 may interact with transcriptional regulators to promote CNCC survival. A similar function for Nubp1 in different contexts could explain why an increased expression of NUBP1 protein is associated with poor prognosis in certain types of cancer [63]. Future studies focused on Nubp1 and Nubp2 could attempt to identify novel binding partners for the proteins in mammalian tissue, which would not only address this alternate hypothesis but might help shed more mechanistic light on their roles in centrosome duplication and ciliogenesis and show whether these latter functions involve other iron-sulfur cluster assembly pathway components.

Supplementary Material

1

Supplemental Figure 1: Loss of Nubp2 does not interfere with neural crest migration. (A-D) Sox10 whole mount in situ hybridization at E9.5 (A-B) and E10.5 (C-D) shows normal organization of NCC migratory streams and peripheral nervous system derivatives in Wnt1-cre; Nubp2^cKO mutants. (E-F) Fluorescent imaging of Wnt1-cre; Nubp2^cKO; mT/mG and Wnt1-cre; mT/mG littermate control E10.5 embryos shows intact and normally organized peripheral NCC and dorsal root ganglia, while the frontonasal prominence and pharyngeal arches (populated by post-migratory CNCC mesenchyme) appear truncated and have diminished GFP signal in the mutant.

2

Supplemental Figure 2: Phenotypic variation among Wnt1-cre; Nubp2^cKO E10.5 embryos corresponds to variation in the apoptotic, but not proliferative index. Sorting proliferation (A) and apoptosis (B) data based on severity of mutant phenotypes at E10.5 reveals that mutants with no abnormal phenotype detectable by sight (mild mutant) are indisinct from controls. On the other hand, mutants which can be easily distinguished by hypoplastic craniofacial primordia (severe mutants) have significantly higher apoptotic indices.

• The ENU mutant dorothy exhibits midfacial clefting, micromelia, and syndactyly

• Dorothy is caused by Nubp2:c:626T>A which codes NUBP2:p:Val209Asp (Nubp2^dor)

• Nubp2 is required for organogenesis, Nubp2^null/null and Nubp2^null/dor are lethal

• Midfacial cleft in Wnt1-cre; Nubp2^cKO embryos is caused by neural crest apoptosis