Proband I: A 46-year-old female presented with mild short stature, standing at 149 cm (1.46 standard deviations, SDs), and weighing 50 kg. The patient reported a history of limping since childhood, accompanied by unequal leg lengths. Over the past 2 years, hip pain emerged, with both pain and limping intensifying in the last 6 months. Physical examination revealed pelvic tilt, with the affected lower limb shortened by approximately 2 cm compared to the contralateral side. No webbed neck, hyperpigmentation, or facial dysmorphisms were observed upon physical examination. Electrocardiography (ECG), cardiac ultrasound, and abdominal ultrasound showed no abnormalities. Pelvic X-rays indicated bilateral hip joint dislocation, with a complete dislocation on the left side (Figure 1B). The patient and her parent denied any history of swaddling during infancy. No family history related to DDH was found in family lineage tracing (Figure 1A).

Proband II: A 49-year-old female with a height of 162 cm and weight of 57 kg. The patient noticed unequal length of lower limbs at 4 years old and experienced right hip joint pain following physical exertion 20 years ago. Examination revealed congenital dysplasia of the right hip joint. Over the past 4 years, pain has intensified following physical labor. Pelvic tilt is observed, with the affected lower limb shortened by approximately 2 cm compared to the contralateral side. While her ECG revealed sinus tachycardia, ultrasound examinations and physical assessments did not show any significant abnormalities. Pelvic X-rays displayed a total dislocation of the right hip joint (Figure 1B). Similarly, this patient also denied any history of swaddling during her early years. Also, no family history was found in this family lineage (Figure 1A).

To clarify the pathogenic mutations in two patients, we initiated our investigation with Whole Genome Sequencing (WGS) of the DNA from the two patients. According to the guidelines set by the American College of Medical Genetics and Genomics (ACMG), we assessed single nucleotide variants (SNVs) and copy number variants (CNVs). Unfortunately, no pathogenic or likely pathogenic genomic variations were identified. Subsequently, we analyzed all variants of uncertain significance (VUS) and discerned that both patients harbored an identical heterozygous missense mutation (NM_002880.4: c.193A>G [p. Lys65Glu]) in the RAF1 gene, which was confirmed by Sanger sequencing (Figure 2A). Both parents of two patients were without this mutation, the RAF1 c.193A>G mutation was shown to be de novo (Supplementary Figure S1).

To investigate the suspected pathogenicity of the detected mutation, a range of mutation prediction tools were employed to analyze its potential harm. Bioinformatic analyses revealed a substantial likelihood of pathogenicity, as reflected by the scores from several prediction instruments (SIFT = 0.06, PolyPhen = 0.816, CADD = 24.9, RawScore = 3.57). Additionally, this mutation was not observed in various population-based databases, including the 1000 Genomes Project, Exome Aggregation Consortium, and Genome Aggregation Database. Similarly, this mutation has not been reported in the Catalogue Of Somatic Mutations In Cancer database. Significantly, the mutation is situated in a region of the gene known for its conservation (Figures 2B, C).

To determine the impact of the mutation on the structure of the RAF1 protein and the formation of its corresponding complexes, we first simulated the structural effects of the c.193A>G mutation on the functional domain of the RAF1 protein. The mutation of lysine to glutamate at residue 65 on Ras binding domain (RBD) impedes the dynamic of RBD-cysteine-rich domain (CRD) in the absence of KRas-binding. Located at the critical junction of β-strand 1 and β-strand 2 in the RBD, residue 65 features a side-chain structure that is pivotal for the structural integrity of both the β-sheet and the linkage loop between the RBD and the CRD (Figure 3A). Comparative molecular dynamics simulations (MDS) of the wild type and the missense mutation c.193A>G (p.Lys65Glu) were conducted, with the root mean square deviation (RMSD) results indicating enhanced stability in the mutant K65E variant (Figure 3B). In the wild-type RBD-CRD structure, a tendency for the inward clustering of RBD and CRD regions was observed, leading to a more pronounced morphology of the linkage loop (Supplementary Figure S2). Conversely, the missense mutation c.193A>G (p.Lys65Glu) exhibited marked alterations in the β-sheets and loops within the RBD and CRD, but not the linkage loop (Supplementary Figure S3). It seems to indicate that, in the absence of ligand binding, the longer side-chain structure of lysine makes it more spatially flexible, while the shorter side-chain structure of glutamate suggests that it is relatively stable (Supplementary Figure S4 and Supplemental Movie).

Further, we explored the changes in the binding between mutant RAF1 and Kras proteins. The mutation of lysine to glutamate at residue 65 on RBD attenuates RBD-KRas interaction. In the wild-type configuration, C-C and C-N interactions between lysine and arginine are predominant, with distances exceeding 4 Å, while the primary interaction in the missense c.193A>G (p.Lys65Glu) is characterized by an O-N contact within a closer range of 3.5 Å (Figure 3C; Supplementary Figure S5). This indicates a transition in the RBD-Kras interaction from van der Waals forces to hydrogen bonding. However, the results of molecular dynamics simulations showed that the weak interaction force between residue 41 arginine of Kras and residue 65 lysine of RBD in the wild type persisted, whereas the hydrogen bond formed in the missense c.193A>G (p.Lys65Glu) was broken upon completion of the MDS preprocess (Figure 3D). Over the 50 ns MDS duration, the interaction forces in the wild type remained constant, while in the missense c.193A>G (p.Lys65Glu), only intermittent van der Waals forces were observed between the two residues (Supplementary Figure S6).

Our initial investigation focused on the effects of the missense mutation c.193A>G (p.Lys65Glu) on RAF1 mRNA and protein expression. In the absence of EGF stimulation, the expression levels of mutant RAF1 mRNA were significantly lower than those of the wild-type RAF1 mRNA (Figure 4B). Notably, under EGF stimulation, the expression levels of the mutant RAF1 mRNA exhibited a significant increase in comparison to the wild type (Figure 4G). Complementing this, Western blot (WB) analysis further confirmed that the missense mutation c.193A>G (p.Lys65Glu) results in enhanced RAF1 protein expression (Figure 4J). Subsequent immunofluorescence assays showed no significant differences in the intracellular localization of the mutant versus wild-type RAF1 proteins (Figure 5). Drawing from in vitro expression experiments and bioinformatic predictions, we postulate that the RAF1 missense mutation c.193A>G (p.Lys65Glu) might contribute to disease progression by influencing RAF1 protein expression levels or by modifying its interaction with other molecules, potentially leading to dysregulation in downstream signaling pathways.

To elucidate the effect of RAF1 expression alterations on downstream signaling, we examined ERK and MEK expression, as well as their phosphorylated forms, within the Ras/ERK pathway under EGF stimulation conditions. In the absence of EGF stimulation, phosphorylation levels of ERK and MEK showed no significant difference between the mutant and wild-type groups (Figures 4C–E), suggesting that the mutated RAF1 protein does not substantially influence the inactive state of the Ras/ERK pathway. Conversely, under EGF stimulation, the transducer activity of RAF1 is amplified, leading to the formation of a complex with Ras molecules and, consequently, elevated phosphorylation of downstream ERK and MEK. Notably, when compared to the wild-type RAF1 under EGF stimulation, the mutated RAF1 protein exhibits a significant increase in expression (Figures 4F, J), indirectly indicating a potential deviation from the expected complex formation with Ras. Additionally, after EGF stimulation, the phosphorylation levels of MEK and ERK in the RAF1 pathway are considerably lower in the mutant group compared to the non-mutated group (Figures 4H–J), further supporting the hypothesis that the missense mutation may impair phosphorylation of downstream molecules by altering interactions with Ras.

Two unrelated probands and their family members were included in this study. After obtaining informed consent, clinical data and venous blood samples were collected from participating families. The collection of venous blood samples adhered to the Helsinki Declaration and was conducted in accordance with the ethics approved by the Institutional Review Boards (IRB ID: SWYXNO. 2023-013).

Demographic information of the probands was extracted from medical records. A comprehensive and systematic physical examination was conducted on the probands, including ultrasonography of the heart and abdomen. Additionally, full-length X-rays of lower limbs and an anteroposterior view of the pelvis were taken to determine any combined skeletal deformities. The three-dimensional computed tomography (CT) reconstructions were taken to reveal dysplastic acetabula. After collecting venous blood samples from the probands and their parents, DNA was extracted following the instructions of the DNA extraction kit (TIANamp, DP348-02).

Using the DNBSEQ-T7 (BGI, Shenzhen) sequencing platform, the DNA samples underwent 30X PCR-free whole-genome sequencing (Supplemental Methods). The raw data were analyzed and annotated using a previously published analytical pipeline, the Peking Union Medical College Hospital Pipeline (PUMP) (Wang et al., 2018; Chen et al., 2021; Zhao et al., 2021) (Supplemental Methods).

We performed an agnostic analysis of WGS data for causal variants potentially contributing to individual patients’ clinically observed phenotypes. (Supplemental Methods). The interpretation of causal variants was based on ACMG guidelines (Richards et al., 2015). Sanger sequencing was performed independently on available subjects and parental samples to validate variant interpretation by an orthogonal sequencing method and to investigate segregation according to Mendelian expectations for the identified variant allele(s) (Supplementary Methods, Supplementary_Table_S1).

Hras-RAF1 complex (PDB ID: 4G0N) and Kras-RAF1 complex (PDB ID: 6XGU) containing the changed residue were used as protein templates for before and after mutation comparisons (Fetics et al., 2015; Tran et al., 2021). All the 3-dimensional structures were visualized and preliminary analyzed using PYMOL (Version 2.5.4, Schrodinger: https://www.schrodinger.com/pymol). The UCSF Chimera (Version 1.17.3, https://www.cgl.ucsf.edu/chimera) was used for the p.Lys65Glu residue change exploration (Pettersen et al., 2004). The side chain positions were selected as the highest probability positions for none-clashes with neighboring residues obtained from the Dunbrack backbone-dependent rotamer library (Dunbrack, 2002).

GROMACS (2023.2) was used to preprocess the structure before and after mutation of the RAF protein and Kras-RAF complex and perform 50 ns MD (Abraham et al., 2015). Incomplete structures in some residues were repaired by complementary patch using PDBfixer (Version 1.9, https://github.com/pandegroup/pdbfixer) (Eastman et al., 2013). The OPLS-AA/L all-atom force field was adopted for MDS (Kaminski et al., 2001). Energy minimization, NVT equilibration, and NPT equilibration were both performed for 100 ps, respectively, using the STEEP algorithm and the leap-frog algorithm (Lemkul, 2018). All trajectories were intercepted at 20 ps intervals for visualization and analysis. FFmpeg was used to generate movies based on the intercepted trajectories.

We chemically synthesized the RAF1-WT target gene fragment by adding recombination sequences before and after the target gene based on the principle of seamless cloning. Through the use of agarose gel cutting, the synthesized PCR fragments were recovered. The pEGFP-C1 plasmid vector was double digested with XhoI/BamHI, and then the RAF1-WT fragment was reconstituted into the pEGFP-C1 restriction vector. Primers were designed to amplify the RAF1-WT template (Supplemental Methods). After PCR amplification of the products using the RAF1-WT plasmid as a template, the products were recombined. After the products were transformed, the positive clones were identified and screened, and the Sanger sequencing was performed to validate the sequence of positive clones.

The HEK 293T cells (Shanghai Mcellbank Biotechnology Co., Ltd, Shanghai) were grown adherent and maintained in high-glucose Dulbecco’s modified Eagle’s medium (DMEM, Gibco, 11965092) containing 10% premium fetal bovine serum (FBS, QmSuero, mu002SR) and 1% penicillin and streptomycin (MacGene, CC004). The constructed plasmids were transfected using Lipofectamine 3000 (Invitrogen, L3000015) following the protocol suggested by the manufacturer. At 48 h after transfection, cells were starved for 24 h in the medium free of FBS and then stimulated with or without EGF (100µg/mL, Solarbio, P00033) for 15 min.

Total RNA was extracted from transiently transfected 293T cells using rapid cell RNA extraction kit (Takara Bio, 9767) and cDNA was reverse transcribed using reverse transcriptase (Takara Bio, RR047A). Quantitative PCR (qPCR) was executed using a two-step quantitative reverse transcription-PCR kit (Takara Bio, RR820A). The mRNA level was normalized to GADPH and used 2^−ΔΔCT method to calculate the relative mRNA level (qPCR primers were listed in Supplementary Table S2).

The protein expressions of RAF1, MEK1/2, P-MEK1/2, ERK1/2, and P-ERK1/2 were detected by western blot. Total intracellular protein was extracted using RIPA lysis buffer (Beyotime, P0013B) supplemented with 1% PMSF (Beyotime, ST507), and protein concentration was determined using the BCA Protein Concentration Assay kit (Epizyme Biotech, ZJ103). Quantified proteins were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis followed by membrane transfer using PVDF membranes (Epizyme Biotech; WJ002). Tris-buffered saline-tween (TBST, Epizyme Biotech, PS103S) washed the PVDF membrane and blocked it with protein blocking solution for 1 hour. After washing the membranes with TBST, primary antibodies against RAF1 (1:1000, Abmart, T55225), MEK1/2 (1:5000, Abmart, T55168), P-MEK1/2 (1:1000, Abmart, TA8035), ERK1/2 (1:1000, Abmart, T40071), and P-ERK1/2 (1:1000, Abmart, TA1015) were added and overnight at 4°C. After washing PVDF membranes with TBST, horseradish peroxidase (HRP)-conjugated secondary antibodies (1:5000, Beyotime, A0216) were added and incubated for 1 hour at room temperature. The target bands were detected by an ECL chemiluminescence kit and observed in an automatic chemiluminescence image analysis system (Tanon 5200 Multi).

Identification of the transfected 293T cell, and detection of mutant RAF1 and wild-type RAF1 distribution in cell was performed by immunofluorescence staining. Slides were plated into six-well plates for 293T cell seeding and transfection. After 48 h, the slides were fixed with 4% paraformaldehyde (Solarbio, P1110) for 15 min and washed with PBS (Solarbio, P1020). The cells were permeabilized with Immunostaining Permeabilization Solution with Saponin (Beyotime, P0095) for 15 min and washed with PBS. After blocked with 10% immunofluorescence blocking solution for 1 hour and washed with PBS, the cells were incubated with primary antibodies against RAF1 (1:100, Abmart, T55225) at 4°C overnight. After washing with PBS, cells were incubated with CoraLite488-conjugated Goat Anti-Rabbit IgG (1:1000, Proteintech, SA00013-2) for 1 hour at room temperature. Cells were washed with PBS and stained cellular structure with phalloidin (Proteintech, PF00003), nuclei with DAPI (Beyotime, C1006). After washing with PBS, anti-fluorescence quencher was added to seal the slices. Finally, the cells were photographed using a confocal microscope (TCS SP8; Leica Microsystems, Biberach, Germany).