Recruitment of pedigrees

This study included two families, family 1 from China (Fig. 1a) and family 2 from Germany (Fig. 1b) who matched from GeneMatcher based on a common candidate gene [17]. Family 1 from China consisted of three individuals across two generations, with the nine-year-old proband diagnosed by orthodontists from the Affiliated Stomatology Hospital of Nanjing Medical University using panoramic radiography and clinical examination. All participants in family 1 have no syndromes and supernumerary teeth. Family 2 from Germany underwent expert clinical examination at the University Medical Centre of Mainz.

Fig. 1

Pedigrees and phenotypes in the families. a Pedigree of family 1. The proband (II:1) is a 9-year-old boy with NSTA. b Pedigree of family 2. The proband (III:3) is an 11-year-old boy with agenesis of eight permanent teeth, auricular dysplasia, hearing impairment and further findings of olfactory dysfunction. Square, male; circle, female; black, patient; arrow, proband. c–e Intra-oral photographs and panoramic radiographs of individuals from family 1. Schematic of congenitally missing teeth of proband and his mother. Asterisks and solid squares indicate the congenitally missing teeth. Max, maxillary; Mand, mandibular. f Photographs of the proband and father in family 2. (i) Frontal view, (ii) right profile and ear and (iii) left profile and ear. First- and third-degree microtia of the right and left ear, respectively. (iv) Mature cataract in the right eye of the father. g Pure-tone audiometry of the proband in family 2 following bone-conduction implantation, note normal hearing on the right ear and mild hearing loss on the left ear

Molecular genetic testing

In family 1, genomic DNAs from all family members were extracted using a Qiagen Blood Kit (Qiagen, Hilden, Germany) and whole-exome capture was performed with the Agilent SureSelect Human All Exon V6 followed by next-generation sequencing on an Illumina HiSeq sequencer. Variant analysis was performed with the Genome Analysis Toolkit (GATK, version 3.3.0). Multi-sample variant calling was performed by HaplotypeCaller, and variants were filtered by Variant Quality Score Recalibration for both single nucleotide polymorphisms (SNPs) and insertions/deletions (InDels) with the following filters: (1) removal of Exome Aggregation Consortium (ExAC), 1,000 Genomes Project and Exome Sequencing Project (ESP6500) browser variants with a minor allele frequency (MAF) > 0.01, (2) retaining variants located in the exon and splicing regions, (3) retaining SNPs predicted to be harmful by at least two tools (SIFT, PolyPhen-2, MutationTaster and CADD), (4) keeping variants common in all patients but not in normal subjects, (5) identifying the potential causative variant related to TA which was predicted by Phenolyzer (Fig. 2a).

Fig. 2

a Flow chart outlining selection of the causative variant. b Schematic diagram of the gene location of the damaging allele. c Sanger sequencing of the heterozygous c.103G > A (I:2 and II:1) and wildtype (I:1) alleles in the FGFR1 gene. Red dotted frames indicate the positions of causative variants. d Conservation of each amino acid residue across species is shown. The red arrow indicates the mutated amino acid. Glycine at position 35 is conserved. e Sanger sequencing of the heterozygous c.1859G > A (II:5 and III:3) and wildtype (II:6 and III:4) alleles in the FGFR1 gene. Red dotted frame indicates the position of causative variants

For family 2, genomic DNAs from the proband, his parents and younger brother were extracted from whole blood using an automated standard procedure. A custom-designed targeted genomic panel including 151 syndromic and non-syndromic deafness genes was performed as previously described [18]. All variants were mapped to the human reference sequence GRCh37/hg19.

Sanger sequencing

The variants in both families were validated using Sanger sequencing. Primers of FGFR1 were designed (family 1: forward: 5’-AAACATTGACGGAGAAGTAGGTG-3’; reverse: 5’-TTCCTAACTTTGCCTCTTTCTTC-3’, family 2: forward: 5’-CTAGTTGCATGGGTGGCG-3’; reverse: 5’-GTTCTCAGCCCACCCCAC-3’) with Primer-BLAST (NCBI). In family 1, the Sanger sequencing data were analyzed using Chromas (version 1.0.0.1, Technelysium Pty Ltd., Australia). In family 2, the Sanger sequencing data were analyzed using Mutation Surveyor. The variants co-segregated with disease in the two families.

Cell culture, lentiviral construction and transfection

Human embryonic kidney 293 (HEK293; ATCCCRL-1573) were purchased from ATCC and human dental pulp stem cells (hDPSCs) were isolated from tooth extraction. Pulp tissues were minced, digested with collagenase type I (Item#: 1904MG100, BioFroxx, Germany) and trypsin with alpha-modified minimum essential medium Eagle (α-MEM) in a centrifuge tube with shaking every 5 min for 4 times and collected in a medium-sized dish. The third generation of dental pulp stem cells (DPSCs) were harvested and incubated with the antibodies CD29-APC, CD90-FITC, CD73-PE and CD45-PE (BD Pharmingen, England) for 1 h in the dark and washed twice with PBS. The specific fluorescence of the samples was examined with a flow cytometer (BD Biosciences, San Jose, CA, USA). HEK293 and hDPSCs were cultured in Eagle's minimum essential medium (EMEM) and α-MEM supplemented with 10% fetal bovine serum, 1% penicillin–streptomycin solution at 37 °C in 5% CO2.

Lentiviruses (Lenti-FGFR1-MT-G103A-3FLAG-OE and Lenti-FGFR1-3FLAG-OE) were prepared by transfection of plasmids containing the open reading frame of wildtype or mutant human FGFR1 into HEK293. Lentivirus (FV115-mCMV-ZsGreen) was used as a control. When HEK293 and hDPSCs reached 40–50% confluency, they were transfected with three lentiviruses (Lenti-FGFR1-MT-G103A-3FLAG-OE; Lenti-FGFR1-3FLAG-OE and FV115-mCMV-ZsGreen) at a multiplicity of transfection of 5 pfu/HEK293 and 100 pfu/hDPSC in the medium containing 5 μg/mL polybrene.

RNA extraction and quantitative real-time polymerase chain reaction (qRT-PCR)

To compare gene expression of cells which were transfected by lentiviruses, total RNA was extracted from the cells by FastPure® Cell/Tissue Total RNA Isolation Kit V2 (Vazyme, Nanjing, China) and reverse-transcribed to cDNA (RR036A, Takara Bio, Shiga, Japan). The mRNA expression was evaluated using SYBR Mastermix (Q712-02, Vazyme, Nanjing, China) on QuantStudio7 qRT-PCR System (Applied Biosystems) and normalized against the endogenous GAPDH RNA control. The primers were listed as follows: FGFR1 (forward: 5’-ACGCAGGATGGTCCCTT-3’, reverse: 5’-GTTGTGGCTGGGGTTGTAG-3’) and GAPDH (forward: 5’GGACCTGACCTGCCGTCTAG-3’, reverse: 5’-GTAGCCCAGGATGCCCTTGA-3’).

Immunofluorescence (IF) staining

The cells which were transfected with lentiviruses were fixed with 4% paraformaldehyde, infiltrated with Triton X-100 solution (Beyotime, China) for 12 min and blocked with goat serum. Then, cells were washed by PBS twice and incubated with anti-FGFR1 antibody (diluted 1:1000, CST, #9740) overnight, followed by incubation with a mixture of secondary antibody with fluorochrome for 1.5 h in the dark. The nuclei were stained with DAPI (Beyotime, China). Finally, the result was observed under a fluorescence microscope (Leica, Germany).

Western blot

To evaluate the expression of target proteins of transfected cells, we collected the cells for lysis in RIPA buffer (Beyotime, Shanghai, China) on ice. Protein samples of the same amount were loaded in 10% SDS-PAGE for electrophoresis separation and transferred to 0.22 μm polyvinylidene difluoride (PVDF) membranes (Millipore, Massachusetts, USA). While blocked with 5% non-fat milk for 2 h at room temperature, the membranes were incubated at 4 °C overnight with primary antibodies including FGFR1 (diluted 1:1000, CST, #9740), E-cadherin (diluted 1:1000, CST, #3195), N-cadherin (diluted 1:1000, CST, #13,116), Vimentin (diluted 1:1000, CST, #5741), ID4 (diluted 1:1000, Abcam, ab220166), SMAD6 (diluted 1:1000, ab273106), SMAD7 (diluted 1:1000, ab216428) and GAPDH (diluted 1:1000, Beyotime, AG019). When washed with Tris-buffered saline containing 0.05% Tween-20 (TBST buffer) for three times and incubated with horseradish peroxidase-conjugated secondary antibodies (1:10,000), the protein bands were visualized by chemiluminescence reagents (P10100, NcmECL Ultra).

Cell apoptosis and proliferation assay

For determination of apoptosis, transfected cells were seeded in six-well plates, treated with trypsin (Gibco, Grand Island, USA) and resuspended as a single-cell suspension after incubating 48 h. We used Annexin V-PE Apoptosis Detection Kit (BD Biosciences, San Jose, CA, USA) to stain cells and analyzed using a Fluorescence Activated Cell Sorting (FACS) System by BD Biosciences (San Jose, CA, USA). Data were analyzed with FlowJo software (TreeStar, Ashland, OR, USA). Cell proliferation was assessed by absorbance using a Cell Counting Kit-8 assay (CCK8, Dojindo, Kumamoto, Japan) according to the manufacturer’s instructions. Cells were seeded in 96-well plates at a density of 3 × 103 cells per well. We added 10 μl CCK8 solution into each well for 2 h incubation at 37 °C. The absorbance at a wavelength of 450 nm was measured on a spectrophotometer microplate reader (Multiskan MK3, Thermo).

RNA sequencing (RNA-seq)

Three pairs of biological replicates (transfected with overexpression lentivirus or control vector as 1 sample) of the RNA sample were collected by 1 mL TRIzol reagent (Invitrogen Corporation). The library preparation was carried out according to the instructions provided with the Trio RNA-Seq Library Preparation Kit (Nugen Technologies, USA) followed by sequencing using an Illumina HiSeq sequencing platform. Skewer software was employed to filter low-quality reads and obtain high-quality clean reads. FastQC software (v0.11.5, http://www.bioinformatics.babraham.ac.uk/projects/fastqc) was employed for quality control analysis. For alignment, STAR software (2.5.3a, https://github.com/alexdobin/STAR) was used to compare the clean reads and the reference gene sequence. Compared against the reference genome, the number of sequences of each chromosome were counted, and then the average depth was calculated within each 5 kb of the reference genome and was taken log2 to complete the reference genome density distribution statistics. For all samples, StringTie software (v1.2.1c, http://ccb.jhu.edu/software/stringtie) was used to count the original sequence counts of known genes, and the expression of known genes was calculated using fragments per kilobase million to calculate the metrics.

Differentially expressed transcripts were identified using DEGseq package in BioConductor (https://bioconductor.org/packages/release/bioc/html/DESeq2.html). The absolute value of log2fold changes ≥ 1 and p ≤ 0.05 of differential genes were integrated to create a volcano plot and heat map. The Kyoto Encyclopedia of Genes and Genomes (KEGG) terms were identified in differentially expressed genes by R Cluster Profiler package (4.0.5).

Statistical analysis

The GO and KEGG analyses were performed using R software (version 4.0.5), and the false discovery rate was used to control for multiple testing. For all graphs, statistical analyses were performed using a two-tailed, unpaired Student’s t-test (GraphPad Prism-6 software; San Diego, CA, USA). Data were considered statistically significant at p-values < 0.05.