RASopathies: the Musculoskeletal Consequences and their Etiology and Pathogenesis

1. Introduction

The rat sarcoma viral oncogene/mitogen-activated protein kinase (RAS/MAPK) signaling pathway is a critical signaling pathway involved in cellular proliferation and differentiation, cell senescence, embryonal development, and oncogenesis (1, 2). The RASopathies are a group of at least 7 unique clinical syndromes (Table 1), but resulting from the unifying molecular mechanism of dysregulated signaling in the RAS/MAPK pathway. As such, they share certain overlapping clinical characteristics, including specific craniofacial and ocular dysmorphology, congenital cardiac malformations, neurodevelopmental deficits and a propensity for cancer. While RASopathies are rare disorders, as a group they affect ~1:1000 individuals (1, 2). Each RASopathy, however, is independently described and frequently attributable to specific germline mutations in RAS/MAPK pathway components (Table 1).

Table 1:

RASopathies with reported musculoskeletal features

Syndrome	Common Clinical Features (non-MS1)	Musculoskeletal (MS) Features and/or Pathology	Pathway Gene(s) and % contribution	Protein // Function	Inheritance: Germline vs. Mosaic
Neurofibromatosis type 1 (NF1)	Café-au-lait macules Intertriginous freckling Neurofibromas Lisch nodules Optic glioma Mild neurocognitive impairment Cancer predisposition	Short Stature Osseous dysplasia Macrocephaly Scoliosis Vertebral anomalies Low BMD/osteoporosis Pseudo arthrosis Pectus carinatum/excavatum Rhabdomyosarcoma	NF1: 100%	Neurofibromin // RasGAP	AD Germline + Mosaic
Noonan Syndrome (NS)	Distinct facial features Ocular hypertelorism, ptosis, downslanting palpebral fissures Congenital heart defects; HCM (↑ if RIT1) Mild intellectual deficits Delayed puberty, hypogonadism Bleeding disorders Cancer predisposition	Short stature (↑ if RAF1, SHOC2; ↓ if SOS1) Pectus carinatum/excavatum Scoliosis/kyphosis Vertebral anomalies Cubitus valgus Micrognathia Brachydactyly Low BMD Giant cell lesions of bone and soft tissue	PTPN11: 50% SOS1:13-15% RAF1: 5%RIT1: 5% KRAS: <5% NRAS: <1% SHOC2: <1% CBL: <1% BRAF MAP2K1 LZTR1	SHP2 // Phosphatase SOS1 // RasGEF CRAF // Kinase RIT1 // ELK1 transactivation KRAS // GTPase NRAS // GTPase SHOC2 // Scaffolding CBL // E3 ubiquitin ligase BRAF // Kinase MEK 1 // Kinase	AD Germline (de novo mutations) AD + AR
Noonan Syndrome with Multiple Lentigines (NSML)	NS facial features Ocular hypertelorism Freckling/lentigines EKG abnormalities Pulmonary stenosis Abnormal genitalia Deafness (sensorineural)	Short stature Pectus carinatum/excavatum Scoliosis Mandibular prognathism Craniosynostosis	PTPN11 RAF1 BRAF MAP2K1	SHP2 // Phosphatase CRAF // Kinase BRAF // Kinase MEK 1 // Kinase	AD
Capillary Malformation-Arteriovenous Malformation Syndrome (CM-AVM)	Multifocal capillary vs AV malformations Cardiovascular malformations	Leg length discrepancies, or limb overgrowth, related to vascular limb lesions	RASA1	P120-RasGAP // RasGAP	AD Germline + Mosaic (de novo mutations)
Costello Syndrome (CS)	Distinct facial features Dermatologic features Postnatal growth delay Cardiac abnormalities (HCM) Cutaneous papillomas Cancer predisposition	Short stature Joint laxity (small joints) Joint contractures (large joints) Hip dysplasia Tight heel cords Scoliosis/kyphosis Chest wall anomalies Low BMD/osteoporosis Rhabdomyosarcoma Hypo-mineralization of tooth enamel	HRAS: 90%	HRAS // GTPase	AD Germline + Mosaic (mostly de novo mutations) paternal bias/advanced paternal age
Cardio-facio-cutaneous Syndrome (CFCS)	NS facial features Ocular abnormalities Congenital heart defects; HCM; rhythm disturbances Failure to thrive; GI dysfunction Cutaneous abnormalities	Short stature Hypotonia/myopathy Scoliosis/kyphosis Pectus carinatum/excavatum Joint hyperextensibility/contractures Pes planus Clinodactyly/syndactyly Osteopenia	BRAF: 75% MAP2KI + MAP2K2: 25% together KRAS: <2%	BRAF // Kinase MEK 1 // Kinase MEK2 // Kinase KRAS // GTPase	AD (de novo mutations)
Legius Syndrome (LS)	Café-au-lait macules Intertriginous freckling Mild neurocognitive impairment	Macrocephaly Chest wall anomalies Scoliosis (rare) Unilateral polydactyly (rare)	SPRED1	SPRED1 // Negative regulator of RAS	AD Germline (de novo mutations)

¹MS, musculoskeletal; HCM, hypertrophic cardiomyopathy; AV, arteriovenous.

This review focuses on germline mutations and provides an overview of each RASopathy, specifically concentrating on the musculoskeletal abnormalities of those RASopathies in which musculoskeletal features are recognized, often as another overlapping component of these syndromes. Herein, we will emphasize the musculoskeletal phenotype and bone histology associated with each syndrome, along with a discussion of the cellular and molecular underpinnings of syndrome-specific musculoskeletal pathology, as delineated by in vivo and in vitro data. Finally, we will review the role of the RAS/MAPK signaling pathway in skeletal development, and discuss potential therapeutic targets for amelioration of musculoskeletal findings in the RASopathies.

2. Clinical Descriptions

2.1. Neurofibromatosis type 1 (NF1; OMIM #162200; Figure 1)

Figure 1: Neurofibromatosis type 1 (NF1).

A - Skeletal findings that have been reported to occur in NF1 in the appendicular and axial skeleton. B - Skeletal findings that have been reported to occur in NF1 in the cranium. C – Mutation (highlighted) in NF1 in the RAS/MAPK pathway.

2.1.1. Phenotypic overview:

The diagnostic criteria for NF1 were formalized by consensus in 1987 (3), although the disorder was first described ~100 years earlier by the German pathologist, Friedrich Daniel Von Recklinghausen (4). Clinical features of NF1 must include two or more of the following: café-au-lait macules (CALM), with ≥6 lesions, each being >5 mm (pre-pubertal) or >15 mm (post-pubertal) in size; skinfold freckling (typically axillary or inguinal); ≥2 neurofibromas (dermal, plexiform, or malignant peripheral nerve sheath tumors); optic pathway glioma; ≥2 Lisch nodules (iris hamartomas); a distinct bony dysplasia (sphenoid wing or tibial); and a 1^st degree relative with NF1, also diagnosed according to these criteria (5). NF1 is an autosomal dominant disorder, but unlike many of the other RASopathies, NF1 is a monogenic syndrome resulting from a germline mutation in the NF1 tumor suppressor gene for neurofibromin. As such, NF1 carries the highest malignancy risk among the RASopathies, contributing to a modestly reduced lifespan (6, 7). Because neurofibromin functions as a negative regulator of the RAS proto-oncogene, neurofibromin deficit leads to RAS hyper-activation. NF1 is estimated to affect 1:2500 - 1:3000 individuals worldwide (8, 9).

2.1.2. Musculoskeletal features:

Of all the RASopathies, musculoskeletal (MS) features of these disorders are perhaps best described in NF1. As listed in Table 1, MS features of NF 1 include: 1) pediatric and adult short stature; 2) a distinct bone dysplasia, occurring in the sphenoid wing of the cranium, the tibia/fibula, or in vertebrae; 3) macrocephaly; 4) osteopenia or osteoporosis; 5) pseudarthrosis; and 6) scoliosis. A propensity for the development of rhabdomyosarcomas is also characteristic of NF1. For a review of the orthopedic-specific management of NF1, see DeLucia, et al (10).

Short stature:

Short stature is a commonly reported feature of NF1, with a reported prevalence ranging between 13 – 30% (11-13). Children with NF1 are significantly shorter than age-matched population norms (12), and reportedly shorter than their unaffected siblings (14). However, a study of 170 individuals with NF1 followed longitudinally at a single center (and excluding those with potentially height-compromising skeletal anomalies), demonstrated that while this population had a mean height z-score of −0.5 ± 1.27 SD, and 68% of NF1 children had negative height Z-scores relative to the population mean, the prevalence of true short stature, defined as a height >2 SDs below mean, was observed in only 8% of children with NF1 (14). These data imply that while the height distribution curve for the NF1 population is clearly shifted to the left, the overall occurrence rate for true short stature, exclusive of other height-compromising skeletal pathology, may be somewhat less than previously expected (14).

The pathophysiology underlying short stature in NF1 remains unclear; however, suboptimal growth in NF1 has been attributed to: 1) a relatively frequent co-existence of growth hormone deficiency (GHD) (15); 2) organic GHD attributable to a propensity to develop destructive suprasellar lesions; 3) reduced peak height velocity and height acquisition occurring during the pubertal growth spurt (16); 4) height-compromising skeletal pathology (i.e. severe scoliosis, tibial dysplasia, bowing of long bones); and 5) a specific association between short stature and the presence of sphenoid wing dysplasia (17). A specific genetic influence of the NF1 mutation on height and/or growth has also been proposed (14).

Axial skeleton:

Abnormalities of both the thorax and vertebral column are characteristic of NF1. Both typical (non-dystrophic) scoliosis, and dystrophic scoliosis can occur in NF1, with prevalence estimates ranging from 10-21% of individuals (13, 18), to as high as 49% of individuals with NF1 (19). Non-dystrophic scoliosis is the more common of the two, usually presenting in early adolescence, with similarities in presentation and treatment approach to idiopathic scoliosis in the general population. In contrast, dystrophic scoliosis, described as “a short-segment, sharply angulated curve associated with underlying vertebral-body and rib abnormalities and sometimes with adjacent plexiform neurofibromas” (5) is apparent in earlier childhood (ages 6-10 years) (6), can be more rapidly progressive, at times associated with respiratory impairment (20), and typically requires surgical correction (5, 13, 21). Unfortunately, for a given individual, modulation between non-dystrophic and dystrophic scoliosis can also occur (10, 22); hence a true distinction between the two types may be inexact.

Consistent with many of the RASopathies, chest wall deformities, including pectus carinatum and excavatum, are also reported in NF1 (5), although with variable estimates of occurrence, ranging from 4.3% of individuals (18) to one-third of individuals with NF1 (5). Interestingly, while the NF1 mutation is known to exhibit extreme variability of clinical presentation, one study of monozygotic twin pairs with NF1 demonstrated a high concordance for pectus deformities, perhaps suggesting a less modifiable hereditary basis for this particular finding (23).

Appendicular skeleton:

The predominant abnormality of the appendicular skeleton in NF1 is congenital tibial dysplasia, heralded by anterolateral tibial bowing and/or fracture. Characteristic bowing becomes evident in later infancy or in the toddler and preschool years (6). Radiographic features of this anomaly include cortical thickening of the tibia, with narrowing of the medullary canal and the appearance of hyperplastic fibrous tissue. Tibial dysplasia occurs in ~5% of individuals with NF1 (18); less commonly, other long bones, including the femur, ulna, radius, and humerus, may exhibit this bone dysplasia (10).

Congenital pseudarthrosis of the tibia, while rare, is none-the-less a very specific diagnostic feature of NF1 (10, 24), in that two-thirds of infants identified with this lesion have NF1 (5). Non-union following long-bone fracture can also lead to acquired pseudarthrosis in persons with NF1 (25).

Cranium and jaw:

Macrocephaly is common in NF1, occurring in 24% of individuals (12). Although a clinical feature of NF1, the etiology of macrocephaly is not understood, and further evaluation or intervention is typically not needed (5). Other abnormalities of the skull and jaw include sphenoid wing dysplasia, a congenital, unilateral, frequently disfiguring defect in the greater wing of the sphenoid, often associated with periorbital plexiform neurofibroma. Sphenoid wing dysplasia is estimated to occur in 3-11% of individuals with NF1 (26). For a comprehensive review of orbital dysplasia associated with plexiform neurofibroma, see Friedrich et al. (27). Periapical cemental dysplasia, a benign condition of the jaw more common in African American women, is also common in women with NF1 (28).

Bone density and/or histology:

Lower bone mineral density (BMD), as assessed by dual energy x-ray absorptiometry (DXA), has been confirmed in pediatric patients with NF1. A study of 84 individuals with NF1 (ages 5-18 years; 54% male) compared with 293 healthy children without NF1 (ages 3-21 years) reported significant decrements in bone mineral content (BMC) and areal BMD (aBMD) of the lumbar spine, femoral neck, and whole body; and in bone mineral apparent density (BMAD; corrected for height) of the lumbar spine and femoral neck (29). Relatively greater decrements were noted in those individuals who also had localized osseous dysplasias (29). Similarly, among a study of 69 pediatric and young adult patients with NF1 (ages 5.2-24.8 years) and a high plexiform neurofibroma burden, 47% of these patients exhibited reduced BMAD by DXA (Z-score ≤ −2, compared with reference data), which was most severe at the lumbar spine (30). Findings of height-corrected, impaired bone density in NF1, meeting diagnostic criteria for osteopenia and/or osteoporosis, have been corroborated by several other pediatric studies (31-34).

A diagnosis of osteopenia is also very common among adults with NF1 (35, 36), and frequently progresses to osteoporosis by middle age (37). Consequently, osteoporosis is reported to occur in 20-40% of the adult NF1 population, and is associated with an increased risk for fracture (28, 35). As example, a Finnish population registry study involving 460 NF1 patients demonstrated that adults ≥41 years of age had a 5.2x risk ratio for fracture compared with age- and gender-matched controls (38). In this same study, children <17 years of age also had a 3.4x risk ratio for fracture compared to controls (38).

Contributing to this increased bone fragility in NF1, microstructural analyses of bone from NF1 patients have revealed: 1) increased osteocyte-related microporosity in cortical bone (39); 2) reduced trabecular bone, either Tb.BV by histomorphometry (40) or trabecular bone score (TBS) by DXA (41); and 3) increased osteoid volume (40). Many of these studies have also demonstrated that reduced bone density and/or quality in NF1 occurs independently of the more common demographic, social or physiological determinants for low BMD in the general population (37, 41, 42). Together, these studies imply that the neurofibromin mutation itself may have a detrimental effect on bone. Finally, hypovitaminosis D is also common among individuals with NF1 (34-36) and has been postulated to be a critical determinant of decreased BMD in the NF1 population (40).

2.2. Noonan Syndrome (NS; OMIM #163950; Figure 2)

Figure 2: Noonan Syndrome (NS).

A - Skeletal findings that have been reported to occur in NS in the appendicular and axial skeleton. B - Skeletal findings that have been reported to occur in NS in the cranium. C – Mutations (highlighted) in NS in the RAS/MAPK pathway.

2.2.1. Phenotypic overview:

As originally described in 1963 by pediatric cardiologists, Drs. Jacqueline A. Noonan and Dorothy A. Ehmke (43, 44), NS has been characterized by the phenotype of short stature; distinctive facial features (hypertelorism, down-slanting palpebral fissures, ptosis, shortened or webbed neck, low-set ears); congenital heart defects (pulmonary valve stenosis, atrial septal defects, hypertrophic cardiomyopathy); mild intellectual deficits; cryptorchidism or hypogonadism; and bleeding disorders (44, 45). NS is an autosomal dominant disorder, with an estimated incidence of 1:1000-2500 live births (2). RAS/MAPK pathway mutations identified in this genetically heterogeneous disorder are listed in Table 1 (1) and include gain-of-function mutations in genes for: Protein tyrosine phosphatase, non-receptor type 11 (PTPN11) (46); Son of Sevenless 1 (SOS1) (47, 48); Raf-1 proto-oncogene (RAF1) (49, 50); Ras like without CAAZ 1 (RIT1) (51); Kirsten rat sarcoma viral oncogene homolog (KRAS) (52); Neuroblastoma Ras viral oncogene homolog (NRAS) (53); Leucine-rich repeat scaffold protein (SHOC2) (54); and Casitas B-lineage lymphoma proto-oncogene (CBL) (55). Missense mutations in leucine zipper-like transcriptional regulator (LZTR1) have also been described (56). Finally, mutations in B-Raf proto-oncogene, serine/threonine kinase (BRAF) and Mitogen-activated protein kinase kinase 1 (MAP2K1) have been reported rarely (57, 58).

2.2.2. Musculoskeletal (MS) features:

As listed in Table 1, MS features are characteristic of NS and include: 1) suboptimal and/or delayed childhood growth, contributing to adult short stature; 2) deformities of the axial skeleton, including spine and rib cage (kyphosis, scoliosis, vertebral abnormalities, pectus deformities); 3) micrognathia and dental malocclusion; 4) cubitus valgus; 5) brachydactyly and syndactyly; and 6) decreased bone mineral density (although less commonly than in other RASopathies) (59). Giant cell lesions of bone are another associated feature of NS, as well as of other RASopathies (60-62).

Short stature:

Despite a normal birth weight and length, decreased growth velocity and/or short stature, present in ~50% of NS individuals, emerges as these children advance toward adult height, independent of the severity of underlying NS comorbidities (63). As alternately stated, ~50% of persons with NS, without intervention, will fall above the 3^rd-centile on standardized growth curves at final height, consistent with a reported mean final adult height of 162.5 cm for males and 152.7 cm for females (Noonan Syndrome Clinical Management Guidelines) (63, 64). By comparison, a more recent natural history study of the Noonan Syndrome Research Group, London, UK, reporting on the adult phenotype of 112 individuals with NS, recorded a greater final height of 169.8 vs. 153.3 cm (male vs. female), potentially indicative of treatment and/or population cohort differences across time (65). Finally, among a cohort of 35 individuals with NS in the Harvard Noonan Syndrome Genotype-Phenotype Correlation study, the mean final adult height for both men (164.6 cm) and women (152.7 cm) was between 3^rd-10^th centile (66).

The biological basis for growth failure in NS is arguable, but has been attributed to downregulation of growth hormone (GH)-receptor signaling (67), and/or relative insulin-like growth factor I (IGF-1) deficiency (67, 68). Consistent with dysregulation in GH signaling, retrospective analysis following 25 years of intervention, suggests a role for exogenous growth hormone administration to improve growth and height outcomes in NS (69), although the greatest gains are observed within the first two years of GH therapy (70, 71), with diminished efficacy thereafter (45). Mutations in RAF1 (72) and SHOC2 (73) have been associated with a relatively higher incidence of short stature, while mutations in SOS1 (72) are associated with a lower incidence of short stature among individuals with NS. In comparison, short stature is equally prevalent among individuals with PTPN11 mutations, as compared with the combined subgroup who are PTPN11 mutation-negative (44).

Axial skeleton:

Early descriptions of NS (based on clinical diagnostic criteria, prior to molecular confirmation) included reports of thoracic scoliosis in ~13% of individuals (74); later reports have also confirmed this prevalence rate (65). Reports by Dr. Noonan, herself, indicated that “scoliosis and kyphosis occur in about 15% of patients” (45). However, evaluation of a larger cohort of children with NS, diagnosed in South Korea, demonstrated a 30% incidence of spine deformities, including scoliosis or scoliosis associated with thoracic lordosis, with a mean age of detection of nine years (75), perhaps suggesting some population-specific differences. Additionally, scoliosis may become more evident as children age (66). Scoliosis appears to be equally prevalent among PTPN11-mutation positive and SOS1-mutation positive individuals (66).

Anterior chest wall anomalies, including superior pectus carinatum and inferior pectus excavatum along with a broad thorax are characteristic of NS, becoming apparent in early childhood (45, 59) and more obvious over time (76). The prevalence of chest wall anomalies has been estimated as affecting 28-95% of individuals with NS (59), and has been reported in NS attributed to PTPN11 (44), KRAS [broad chest (77)] and SOS1 mutations (73). Vertebral defects have also been reported (75).

Appendicular skeleton:

Upper limb anomalies include outward deviation of the elbow (cubitus valgus), along with brachydactyly and occasional reports of syndactyly (78). Cubitus valgus is the most common upper limb deformity, and has been reported in 45% (75) to 50% (78) of persons with NS.

Cranium and jaw:

In NS, the head has an appearance of macrocephaly, with a high, prominent forehead, yet bitemporal constriction, resulting in a turricephalic presentation (78). Interestingly, however, the Noonan Syndrome Research Group has reported a mean adult occipital-frontal head circumference of 56.4 cm in males (range 50-64 cm) and 54.9 cm in females (range 52-59 cm), which is near the 50^th-centile for the general population (65). In contrast, micrognathia, a highly arched palate and dental overcrowding are also relatively common maxillofacial findings in NS (65, 78), contributing to a more triangular overall facial shape with aging (78).

Bone density and/or histology:

Data regarding the frequency of osteopenia or osteoporosis in NS are limited. Small cohort studies have indicated that children with NS have lower bone mineral density (BMD), when compared with age, gender and height-matched control groups (79). Additionally, a diagnosis of osteopenia or osteoporosis has been reported in ~14% of adults with NS (66). Increased bone resorption, as evidence by an increase in urine pyridinium crosslinks has also been reported in the RASopathies, including NS specifically (80).

Bone tissue neoplasms:

Multiple giant cell lesions (MGCL) of bone and soft tissues occur in a subset of NS patients with PTPN11 mutations, often described separately as Noonan-like/multiple giant cell lesion syndrome (NL/MGCLS). However, these same PTPN11 mutations have been described in NS, Noonan Syndrome with Multiple Lentigines (NSML) and Cardio-facio-cutaneous Syndrome (CFCS), suggesting that other mechanisms may contribute to this comorbidity (57). Rarely, MGSL of the craniofacial region, specifically, have been reported in NS (81).

2.3. Noonan Syndrome with Multiple Lentigines (NSML; OMIM # 151100)

2.3.1. Phenotypic overview:

Noonan Syndrome with Multiple Lentigines [NSML; formerly named Leopard Syndrome (82)] is a rare disorder (~200 individuals reported) that is considered an allelic variant of NS. The genetic basis of disease is attributable to a heterozygous pathogenic mutation in one of four genes that overlap with NS (PTPN11, RAF1, BRAF, and MAP2K1) (57, 83). Although a more restricted spectrum of mutations has been associated with NSML, this genetic overlap with NS also contributes to a phenotypic overlap between NSML and NS. Similar to NS, NSML is characterized by craniofacial dysmorphism, short stature, chest wall abnormalities, intellectual deficits (mild), and genital anomalies (often cryptorchidism) (57, 84). However, NSML is further defined by the pathognomonic feature of multiple lentigines (ML), along with other specific features, including EKG abnormalities and sensorineural deafness. The characteristic lentigines are described as small (<0.5 cm), black-brown macules, concentrated in the face, neck and upper trunk, associated with melanocyte hyperplasia, and markedly increasing in number from the age of 4-5 years through puberty (84, 85). EKG abnormalities, progressive conduction anomalies, and hypertrophic cardiomyopathy (49) are common, with heart defects present in 70-85% of the NSML population (57, 84). Sensorineural deafness is also diagnosed in ~20% of individuals (57). Establishing the diagnosis of NSML requires ML plus 2 cardinal features (cardiac abnormalities, poor growth, pectus deformity, facial dysmorphia); or if no ML, 3 cardinal features and a 1^st degree relative with NSML (57, 84).

NSML is an autosomal dominant disorder, predominantly resulting from de novo mutations in the proband or parent. Approximately 90% of NSML cases are attributed to PTPN11 mutations (57) which predominate at residues critical to the switch between inactive vs. active conformations of the PTPN11 protein (SRC Homolog 2 domain-containing PTPase, or SHP-2). Hence, these mutations result in altered SHP-2 catalytic activity (1, 86). The true prevalence of NSML is unknown, perhaps due to phenotypic overlap with other RASopathies. However, a male predominance for NSML has been reported (87).

2.3.2. Musculoskeletal features:

As shown in Table 1, MS features of NSML share some similarity to those of NS, although the MS phenotype in NSML is milder.

Short stature:

Birthweight is normal or above average in NSML (88), while birth length is normal (89). Hence, in one study, macrosomia at birth was reported in 10 of 25 PTPN11-associated cases of NSML (89). Thereafter, growth failure develops in some individuals with NSML, although this is less frequent and less severe than in NS (89). Consequently, only 25% of individuals with NSML have an adult height at <3^rd percentile (59, 84). In NSML, short stature may be mutation-specific, being more common with PTPN11 mutations at the Tyr279 residue (84).

Axial skeleton:

Chest wall anomalies are the most common MS feature of NSML. A broad chest with either pectus carinatum or excavatum is described in 33% (88) to 75% (84) of individuals. Scoliosis has also been reported sporadically (84).

Appendicular skeleton:

Joint hyper-flexibility is a rarely reported component of NSML (82, 90).

Cranium and jaw:

Craniosynostosis can occur in NSML, both in individuals with PTPN11 mutations (91) and those with RAF1 mutations (90). Additionally, prognathism of the mandible can develop in individuals with NSML (81, 90).

2.4. Capillary Malformation-Arteriovenous Malformation Syndrome (CM-AVM, type 1; OMIM #608354)

2.4.1. Phenotypic overview:

The diagnostic hallmark of CV-AVM is the presence of multiple, multifocal, small (<1-3 cm), pinkish, capillary malformations (CM), along with an increased risk for arteriovenous malformations (AVMs) and arteriovenous fistulas (92, 93). Vascular malformations can involve many tissues, including the skin, subcutaneous tissues, muscles, bones (extremities and spine), and various internal organs (heart and brain) (1). CMs, specifically, are most frequently found on the trunk, extremities, head and neck (94). Cardiovascular malformations, including tetralogy of Fallot and septal or valve anomalies can also exist (1). CV-AVM, type 1 (CV-AVM1) is caused by autosomal dominant, loss-of-function mutations in the RAS p21 protein activator 1 (RASA1) gene (95). Mosaic RASA1 mutations have also been recently identified in individuals with CM-AVM (96).

2.4.2. Musculoskeletal features:

MS anomalies are not a characteristic component of CM-AVM. Instead, secondary skeletal abnormalities can develop due to bone overgrowth of an involved limb (94, 97). Intraosseous AVMs of the maxilla or mandible can also impact dental occlusion (81).

2.5. Costello Syndrome (CS; OMIM #218040; Figure 3)

Figure 3: Costello Syndrome (CS).

A - Skeletal findings that have been reported to occur in CS in the appendicular and axial skeleton. B - Skeletal findings that have been reported to occur in CS in the cranium. C – Mutation (highlighted) in CS in the RAS/MAPK pathway.

2.5.1. Phenotypic overview:

Compared with the other syndromes, Costello Syndrome (CS) is a much rarer developmental RASopathy; prevalence for CS has been estimated at <1:1,290,000 (98) to 1:300,000 (99). The predominant clinical characteristics of CS include (99-101): 1) dysmorphic craniofacial features; such as macrocephaly and coarse facial features consisting of thick eyebrows, epicanthal folds and down-slanting palpebral fissures, a depressed nasal bridge, low set ears with thickened helices and lobes, and a large mouth with full lips; 2) cardiac defects; including pulmonary valve stenosis, septal defects, hypertrophic cardiomyopathy and dysrhythmias; 3) ectodermal abnormalities; characterized by curly hair, soft, velvety, redundant, but thickened skin of the dorsum of hands and feet along with deep palmar/plantar creases, and papillomata; 4) musculoskeletal anomalies (detailed below); 5) abnormalities of growth and development; including postnatal failure to thrive and delayed puberty; 6) intellectual disability and developmental delays (estimated IQ, 25-50), associated with structural changes in the brain (enlarged ventricles and Chiari malformation); and 7) an increased risk for both benign (i.e. papillomata) and malignant neoplasms (i.e. rhabdomyosarcoma, transitional cell carcinoma, neuroblastoma), with malignancy onset mostly during childhood. CS is caused by an autosomal dominant germline mutation, typically a heterozygous de novo mutation, in the gene for the Harvey rat sarcoma viral oncogene homolog (HRAS) (99, 102). This RAS/MAPK pathway mutation results in a gain-of-function, or constitutive activation of H-RAS (99). Recognizing that CS shares certain features with other elastin-related disorders, the pathogenesis of CS has been attributed to disrupted elastogenesis, due to a functional deficiency in elastin-binding protein (101, 103).

2.5.2. Musculoskeletal features:

Due to the extreme rarity of this syndrome, available data on MS features of CS are generally limited to descriptive reports of specific anomalies among small CS cohorts. Never-the less, within the CS population, MS features are quite common, and the orthopedic manifestation of CS have been delineated in several orthopedic-focused studies (104-106).

Short stature:

While individuals with CS are large for gestational age (LGA) at birth, postnatal failure-to-thrive ensues, resulting in delays in growth and pubertal development, and ultimately in adult short stature. Adult height in CS is characteristically <5^th-centile (104), with mean adult heights of ~ 138-142 cm (107, 108). Adult height can be further diminished by a commonly stooped posture, or by scoliosis. The etiology of growth failure and adult short stature is unknown, although instances of growth hormone (GH) deficiency have been identified (107) and routine screening for GH deficiency is recommended (99).

Axial skeleton:

Similar to other RASopathies, spinal curvature and anterior chest wall deformities (pectus carinatum or excavatum) also occur in CS (59, 108). Kyphosis has been reported in 17% (104) to 58% (106) of individuals with CS, while scoliosis has been reported in 17% (104) to 63% (106) of patients. Unlike other RASopathies, however, curvature reversal of the spine (i.e., thoracic lordosis with lumbar kyphosis) is reportedly unique to CS (106).

Appendicular skeleton:

Ligamentous laxity and hyper-extensible digits are characteristic of the small joints in CS (e.g., hands), while joint contractures are characteristic of the large joints (e.g., ankles, hips, elbows, shoulders). Foot problems, in total, are the most prevalent abnormalities of the appendicular skeleton. Specific abnormalities of the foot, roughly in order of decreasing prevalence, include tight Achilles tendon, planovalgus feet, overriding toes, congenital vertical talus, metatarsus adductus, and hallux valgus (104, 106). Abnormalities of the upper extremities, beyond large joint contractures noted above, include a characteristic presentation of the hands, described as broad, short hands, with hyper-extensible fingers and ulnar deviation of the wrists (106). This CS hand phenotype has been reported in 75% (104) to 85% (106) of those examined. Hip dysplasia, either congenital or acquired, is also a common finding in CS (99, 106).

Cranium and jaw:

Craniofacial features of CS include a relative or absolute macrocephaly (100), with a prominent forehead but with bi-temporal narrowing. Individuals with CS also exhibit a highly arched palate, along with a dental phenotype characterized by delayed tooth development and eruption, malocclusion (81, 99), and hypo-mineralization of tooth enamel (109). For a review of craniofacial and dental development in CS, see Goodwin et al. (109).

Bone density and/or histology:

Osteopenia or osteoporosis are common in the CS population, specifically reported in 47% of patients evaluated by Detweiler et al. (n=34) (106), and in 8 of 8 patients evaluated by White et al. (108). However, the etiology of low BMD has often been attributed to several secondary causes, rather than to a primary bone defect. Possible secondary causes include: reduced overall mobility (104) due to prevalent foot anomalies (105), and ambulation disturbances or limitations; delayed puberty and hypogonadism; reduced GH secretion (107); or a disease-specific myopathy (106, 110). Even so, a study by Stevenson et al. demonstrated a significant elevation in urine pyridinium crosslinks in CS (p<0.0001; n=21 study participants) (80), indicating that increased bone resorption also contributes to the skeletal phenotype of this disorder (similar to data in NS and CFC).

2.6. Cardio-facio-cutaneous Syndrome (CFCS; OMIM #115150)

2.6.1. Phenotypic overview:

Cardio-facio-cutaneous Syndrome (CFCS), so named by Reynolds, et al (111), is characterized by a constellation of specific cardiac, craniofacial and cutaneous/ectodermal abnormalities, which occur along with early growth failure and a severity of neurologic findings or neurodevelopmental delays (112-114). Seizure disorders are also common in CFCS (115). Unlike NS and CS, however, CFCS is not considered a malignancy-prone syndrome. Considerable variability in phenotype can exist in CFCS, which may be mutation subtype-related. Consistent with this variability of presentation, specific heart defects, present in ~75% of individuals, can include any of the following: pulmonic stenosis, septal defects, hypertrophic cardiomyopathy, heart valve abnormalities, and heart rhythm disturbances (less commonly than in CS) (113). The distinctive facial dysmorphia is typified by relative or absolute macrocephaly with bi-temporal constriction, a high forehead, hyperteloric appearance, down-slanting palpebral fissures, a short nose with depressed nasal bridge, and low set, posteriorly rotated ears with thickened helices (81, 113). Dermatologic manifestations, a cardinal feature of CFCS, include: hyperkeratotic skin, keratosis pilaris, palmo-plantar keratoderma, or ichthyosis-like skin; sparse, coarse and curly hair; absent or sparse eyebrows; and dystrophic nails (113, 115). Other features characteristic of CFCS can include hypoplastic optic nerves and genitourinary abnormalities (although milder and less frequent than in NS) (115).

CFCS is an autosomal dominant disorder, equally present in males and females (113), and generally resulting from de novo heterozygous mutations in one of predominantly 3 genes, BRAF (75% of mutation-positive individuals)(116), and MAP2K1 or MAP2K2 (25% of mutation-positive individuals, combined)(117); mutations in KRAS have also been reported rarely (<2%), but in individuals with a phenotype that is intermediate to BRAF/MAP2K1 CFCS and NS (118). Most CFCS mutations result in activation of the RAS signaling pathway. CFCS is a rare disorder (perhaps a few hundred individuals), with one estimate of prevalence given as 1:810,000 live births (98).

2.6.2. Musculoskeletal features:

MS features are a relatively common component of CFCS, and descriptive information is available from several cohort analyses (105, 115, 118, 119). For a review of orthopedic conditions identified in CFCS, see Reinker et al. (105).

Short stature:

Individuals with CFCS characteristically exhibit a normal birth weight (113), but experience post-natal growth failure and significant failure to thrive, attributable to feeding difficulties and gastrointestinal abnormalities (115). Hence, short stature is typical of CFCS. In a clinical study of 186 children and young adults (age range; 6 months to 32 years) with mutation-confirmed CFCS, height at <3^rd-centile was present in ~60% of individuals at the time they were evaluated, irrespective of whether a BRAF or MAP2K1/MAP2K2 mutation was identified (118). Adult short stature is also common and severe, with final adult height measurements often at 2.5 to 5.0 SD below normal (119). Individual CFCS studies have identified short stature among 70-85% of adult participants (98, 115, 119). In a few cases, growth hormone deficiency has been documented, although much less commonly than in CS (119).

Axial skeleton:

A shield chest with pectus abnormalities, either carinatum or excavatum, are considered the most common anomaly of the axial skeleton, reported in up to 63% of individuals with CFCS (115). Scoliosis, reported in 25-33% (105, 115) of those studied, or kyphosis, in 19-23% (105, 115), are also frequent occurrences; kyphosis, in particular, is more commonly seen in CFCS than in NS (105). A relatively high incidence of cervical stenosis has also been noted in this syndrome (105).

Appendicular skeleton:

A long list of both upper and lower extremity anomalies have been reported in association with CFCS (115, 119, 120). The hands are characteristically short and broad, with shortened fingers, possibly with clinodactyly or campylodactyly. Joint hyper-extensibility is also noted (121). Other abnormalities of the arms include: ulnar deviation [in 38% (119)]; elbow contractures [13% (105)]; and cubitus valgus. The feet are also short and broad, with pes planus reported in 13% to 63% of individuals studied. Other lower extremity features of CFCS can include hip dysplasia (16%), knee contractures (22%), pes cavus, metatarsus varus, hallux valgus, and toe crowding (105).

Cranium and jaw:

Relative or absolute macrocephaly is a nearly universal finding in CFCS, reported in 86-97% of individuals (98, 122), typically with bi-temporal narrowing. Ventriculomegaly can co-exist in ~40% of these patients (114). Hypoplasia of the supraorbital ridge (in 52%) is also a common component of the craniofacial appearance (122). Within the jaw, a highly arched palate, seen in up to 80% of subjects, and more commonly noted with BRAF mutations, contributes to vertical dental malocclusion and the appearance of an open bite (122). Additionally, the presence of dysplastic teeth have been noted in several studies (115, 120).

Bone density and/or histology:

As noted above (See Bone density and/or histology; CS), a study by Stevenson et al. reported a significant increase in the concentration of urine pyridinium cross-links in patient with CFCS, similar to CS and NS, implying increased bone resorption in this syndrome as well (80). Currently, data on the prevalence of osteopenia, osteoporosis or fracture risk with CFCS are limited. However, in a recent report by Chen et al. describing 3 Chinese children with MAP2K1-mutation associated CFCS, osteoporosis was identified in one 10 year-old girl, in conjunction with short stature, multiple skeletal malformations and dysplasia of the hip joint (123).

2.7. Legius Syndrome (LS; OMIM #611431)

2.7.1. Phenotypic overview:

Legius Syndrome (LS), originally referred to as “NF1-like” syndrome, is characterized by a similarity to NF-1 in its cutaneous pigmentary abnormalities, but without neurofibromas or other diagnostic NF-1 criteria (124, 125). The diagnosis of LS is established by at least 2 of the following 3 criteria: ≥ 5 CALM, with or without axillary or inguinal freckling (present in 30-50% of individuals); a heterozygous pathogenic mutation in the gene for Sprouty-related, EVH1 domain containing protein 1 (SPRED1); and/or a parent with LS, diagnosed by the above 2 criteria (124, 126). Additional features of LS include a cognitive phenotype, milder than in NF-1 (127), but characterized by speech or language delays, learning disabilities, attention-deficit hyperactivity disorder, or autistic-type behaviors (128). Similar to NF-1, LS is an autosomal dominant disorder, resulting either from germline transmission or, less commonly, from de novo mutation. And, similar to neurofibromin, the SPRED1 protein product functions as a negative regulator of the RAS/MAPK pathway; hence the loss-of-function mutation leads to uninhibited RAS signaling. LS is, however, much less common than NF-1, with an estimated prevalence of 1:46,000-1:75,000 individuals (126).

2.7.2. Musculoskeletal features:

MS manifestations of LS are obtainable from the SPRED1-specific database of ~270 individuals with a mutation-confirmed diagnosis of LS (http://www.lovd.nl/SPRED1) (128). Overall, MS features appear to be a less pronounced component of LS, compared with NF-1 or other RASopathies. However, the relatively small number of identified LS individuals may preclude a complete understanding of the skeletal consequences of LS.

Short stature:

An association between LS and absolute short stature (height at >2 SD below population mean) is not clear-cut. Legius et al. reported a 12% frequency of short stature among this population (126), whereas Denayer et al. reported the finding in ~23% of individuals with LS (129). In contrast, other reports indicate that short stature in LS may be uncommon (130).

Axial skeleton:

Pectus excavatum or carinatum, typically mild, has been reported in 7.5-23% of individuals studied in LS cohorts (124, 129, 130). In one such report, scoliosis was also noted in 4 of 30 individuals studied (129).

Appendicular skeleton:

An increased prevalence of unilateral postaxial polydactyly has been reported in individuals with LS (129, 130). Additionally, 5^th finger clinodactyly has been infrequently noted (59, 131).

Cranium and jaw:

Relative or absolute macrocephaly is possibly the most common MS feature of LS, and has been reported in ~13% (129), 23% (130) or 42% (124) of individuals in specific LS cohorts (131).

3. Modeling of RASopathies in genetically modified mice

3.1. NF1

A number of mouse models have been generated to examine the impact of mutations within the Nf1 gene as they relate to skeletogenesis. Reports have elucidated how loss of function of NF1 impacts skeletogenesis and, in some aspects, these mouse models show similarities to the human condition. Initial attempts using the neurofibromin homozygous null (Nf1^−/−) mouse were unsuccessful as the mice were not viable. Later attempts were then made to study the haploinsufficient mice (Nf1^+/− mouse). Nf1^+/− mice displayed little to no skeletal phenotype, despite demonstrating some dysregulation of isolated osteoblasts in vitro, thus limiting the use of this mouse to model the human NF1 condition (132). In an effort to study the role of neurofibromin at different stages of skeletogenesis, various groups have employed the use of transgenic models in which Nf1 is eliminated in specific cell types like undifferentiated mesenchymal cells, osteochondroprogenitors, osteoprogenitors, osteoblasts and osteocytes. Nf1^flox/flox mice crossed with Prx1-cre mice results in the inactivation of Nf1 in undifferentiated mesenchymal cells in the developing limb (133-135). These mice demonstrated bowing of the tibia and growth retardation, as well as increased porosity, osteoidosis, decreased stiffness, and reduced bone mineral content, consistent with the human condition (133). Fracture healing was also impaired in Nf1 ^flox/flox/Prx1-cre mice, as was muscle development, resulting in muscle fibrosis and reduced numbers of muscle fibers (134, 135). Thus, multiple cells types can ultimately be impacted in the limb if Nf1 is eliminated from undifferentiated mesenchymal cells prenatally. When Nf1 is silenced in osteochondroprogenitors (Nf1^flox/flo/Col2a1–cre), impaired bone mineralization, bone strength and growth retardation are observed (136). In addition, Nf1^flox/−/Col2.3-cre mice demonstrate specific abnormalities in the lumbar vertebrae, similar to changes noted in the axial skeleton of humans with NF1 (137). When Nf1 is silenced in osteoprogenitors there have been some incongruences in outcomes depending on if Nf1^flox/flox mice are crossed with Col1a1-cre or Osx-cre mice. Nf1 ^flox/flox/Col1a1-cre mice display high bone mass and an increase in bone formation, along with increased osteoid and bone turnover (138). In contrast, when Nf1 ablation was induced in osteoprogenitors in the Nf1 ^flox/flox/Osx-cre mouse at post–natal day 14, observations included increased osteoid and cortical porosity, and lower bone mass, cortical thickness, and weaker bones (136). Neurofibromin has also recently been shown to be active in mature osteocytes. Abrogation of Nf1 in osteocytes (Nf1 ^flox/flox/DMP-1-cre) reveled a remarkable metabolic bone phenotype, including high serum fibroblast growth factor 23 (FGF23) levels, hypocalcemia, hypophosphatemia, and increased PTH (139). The bones of these mice demonstrated osteoidosis, reduced bone formation rate, increased porosity, and even spontaneous fractures. Overall, with some exceptions, these models do demonstrate certain skeletal characteristics commonly seen in NF1 in humans, including long-bone bowing, altered bone formation, and decreased long-bone length, and they suggest that osteoidosis/osteomalacia may be pivotal in the pathophysiology of NF1-related osteopathy. These studies also highlight that the timing of ablation (prenatally vs postnatally) as well as the cre-construct used may contribute to variability in skeletal outcomes and phenotypes.

3.2. NS

Several mouse models have provided a better understanding of how gain-of-function mutations in the Ptpn11 gene (encoding SHP2; and accounting for ~50% of NS in humans) can result in skeletal complications in NS. Mice heterozygous for the NS-associated D61G mutation in Ptpn11 demonstrated several skeletal features consistent with NS, including decreased body growth and cranio-facial anomalies, including a shortened skull length and triangular facial appearance (140). The growth delay noted in the D61G mutant mice has been linked to lower circulating concentrations of IGF-1; however, other work suggests there is a primary defect in the growth plate that is not amenable to IGF-1 therapy (141, 142). When another NS-associated Q79R mutation was expressed in neural crest cells, again the length of the skull was shortened, as was the jaw, features that commonly occur in the human condition (143). Kras mutations have also been explored in mouse models of NS. Expression of the NS-associated Kras V14I and T58I resulted in abnormalities of growth, along with craniofacial dysmorphia characterized by a reduced skull length, triangular facial appearance, blunter snout, and wider interorbital separation (144, 145). Many of the mouse models of NS are complimented by studies in zebra fish, which have demonstrated and confirmed that mutations in Ptpn11 and Kras can result in skeletal anomalies (146-148). To date, most of the studies using NS mouse models have focused on craniofacial abnormalities and growth delay; thus, bone microarchitecture, quality and strength are currently understudied.

3.3. CS

Only a few studies using mouse models of CS have examined skeletal consequences associated with mutations in the HRAS gene, the cause of Costello Syndrome (CS). Studies to date have reported on knock-in mouse models of the G12V HRAS mutation that is constitutively active; however, this specific mutation, while commonly found in cancers, is rarely the cause for CS. Therefore, there may be some limitations in drawing conclusions about specific outcomes. Nevertheless, two different approaches to expressing the G12V mutation resulted in similar dental defects, along with facial malformations impacting the nasal bridge, nasal septum and jaw bones, corresponding to clinical features of CS (149, 150). The tooth defects have been shown to evolve from dysfunctional ameloblasts and production of enamel that is hypo-mineralized (151). While these limited studies are informative, additional models using more relevant HRAS mutations may provide additional insights into skeletal development in CS.

3.4. CFCS

Germline mutations in the BRAF, MAP2k1, and MAP2k2 have been shown to be causative in CFCS and there are a few examples of how these mutations can be explored in mouse models. For instance, expression of BRAF L597V leads to increased kinase activity and facial abnormalities (152). Similarly, BRAF V600E allele expression in mice leads to facial dysmorphia and growth retardation (153). In another mouse model of CFCS, wherein the most frequent mutation in CFCS (BRAF Q257R) was expressed, the mutation proved to be lethal prenatally or shortly after birth, yet the mice demonstrated craniofacial anomalies (154). A follow-up study of the same mutation, but expressed on a different genetic background, revealed that facial dysmorphia was observed in mature mice (155). Furthermore, the growth retarded BRAF Q241R mice demonstrated reduced growth plate width as well as decreased serum levels of IGF-1 and its major serum binding protein, IGF binding protein 3, which may be relatable to the growth constriction noted in the BRAF Q257R mouse and is similar to what has been reported in the NS mouse model (156). Products of the Map2k1 and Map2k2 genes in mice, MEK1 and MEK2, respectively, have been shown to be necessary in osteoblasts for normal skeletal development, while over-activity of MEK1 in osteoprogenitors can lead to extra-cortical bone formation, osteoidosis, and increased porosity (157, 158). MEK1 Y130C is the most common MEK1 mutation reported in CFCS in humans. Its expression caused cranial abnormalities and the mutation was not lethal; however, more detail on skeletal outcomes were not reported (159). Overall, these studies suggest that most mutations that have been reported in the human condition of CFCS are amenable to being studied in mouse models and that future studies may reveal more detail on skeletal outcomes as well as mechanisms involved in those outcomes.

4. The RAS/MAPK signaling pathways in RASopathies

The Ras family (KRAS, HRAS, and NRAS) of small GTPases serves as a key node linking extracellular stimuli, i.e., growth factors, cytokines, extracellular matrix, and mechanical loading, to the MAPK intracellular signaling pathway, culminating in cellular responses such as replication, growth, differentiation, senescence, and death. Mutations within the RASopathy spectrum result in activation of the RAS/MAPK pathway via inactivation of inhibitors of the RAS/MAPK pathway, disruption of autoinhibitory intramolecular interactions within pathway activators, decreased degradation of RAS proteins, or promotion of transphosphorylation and activation of kinases within the pathway.

RAS/MAPK pathway activation is initiated by ligand binding to receptor tyrosine kinases (RTK), receptor dimerization, and autophosphorylation on tyrosine residues. Phosphotyrosine residues recruit growth factor receptor bound 2 (GRB2) and it’s binding partner SOS to the inner cell membrane proximal region. SOS is a RAS-guanine nucleotide exchange factor (RAS-GEF), promoting exchange of bound GDP for GTP within RAS, forming the active RAS-GTP complex. SOS mutations associated with NS disrupt intramolecular autoinhibitory interactions within SOS, promoting increased GDP – GTP exchange in RAS. RAS mutants in NS, CFCS and CS affect RAS activity predominantly by reducing guanine nucleotide binding and/or intrinsic GAP activity with a recent study reporting that some RAS mutations associated with NS may bind to SOS in such a way as to favor allosteric SOS autoactivation and thus increased activation of RAS (160).

Additional RAS regulatory proteins SHP2, p120GAP, NF1, and LZTR1 have been reported to be mutated in the RASopathies. SHP2 is mutated in NS and NSML, with mutations clustering in regions that disrupt intramolecular autoinhibitory interaction between the SH2 and phosphotyrosine phosphatase (PTP) domains, destabilizing the inactive form, and allowing SHP2 PTP activity to increase. Active SHP2 dephosphorylates autoinhibitory phosphotyrosines within RAS-GTP (161, 162), decreasing binding of RAS inhibitor p120GAP (protein product of Rasa1 gene) and increasing binding of RAF (161), a downstream mediator of the MAPK pathway. Non-functional mutations of p120GAP and another RAS-GAP, NF1, are associated with CM-AVM and NF1, respectively, due to reduced extrinsic RAS-GAP activity and increased RAS-MAPK pathway activity. LZTR1, leucine zipper-like transcriptional regulator 1, is a component of the E3 ubiquitin ligase complex, which has recently been shown to facilitate ubiquitination and degradation of RAS proteins. Dominant mutated forms of LZTR1 appear to be associated with increased levels of RAS proteins (163), thereby increasing RAS-MAPK signaling, possibly due to reduced ubiquitin-proteasomal degradation. The mechanism is unclear regarding how recessive mutant forms of LZTR1 may increase RAS-MAPK signaling, although interaction between LZTR1 and other proteins implicated in RASopathies (RAF1, SHOC2, PP1) have been reported to promote inhibitory RAF1 Ser259 phosphorylation (164), which is reduced by some mutations within the complex.

Activated RAS binding to the RAS binding domain (RBD) of RAF1 recruits RAF1 to the cell membrane and triggers a conformational change, dephosphorylation of Ser259 of RAF1 by protein phosphatase 1 (PP1) and phosphorylation of multiple sites within RAF1, relieving autoinhibition of the intrinsic kinase domain. Missense mutations of PPP1CB, encoding the catalytic subunit of PP1, have been reported in patients having phenotypic characteristics similar to the rare NS variant, Noonan-like syndrome with loose anagen hair (NS-LAH) (165). At this time, the functional implications of these mutations have not been determined.

Mutation of RAF1 Ser259 or nearby amino acids represents a large group of activating RAF1 mutations in NS and NSML (166, 167), in which autoinhibition of RAF1 occurs possibly due to reduced binding to 14-3-3 protein and resulting decreased stabilization of an inactive RAF1 conformation (168). A recent report has demonstrated that 14-3-3 protein encoding gene YWHAZ may be a new RASopathy-associated allele, with the mutant protein able to bind RAF1 with higher affinity than the wild-type protein, possibly helping to dimerize RAF1 and making RAF1 activation more efficient (169). SHOC2, which is mutated in NS, contains multiple leucine rich repeats and is a scaffolding protein linking RAS-GTP to RAF and also serves as a regulatory subunit of PP1C, which dephosphorylates the regulatory serine residue in RAF1. A p.S2G mutation in SHOC2 is believed to promote myristoylation of SHOC2, causing persistent localization of SHOC2 to the cell membrane where it can more readily and constitutively promote dephosphorylation and activation of RAF1 (54).

Clustering of RAF at the cell membrane promotes Raf transphosphorylation [reviewed in (170)]. Interestingly, in some cases of NS, CFCS and NSML, kinase impaired versions of BRAF and RAF1 have been identified. Contrary to the supposition that impairment of kinase activity would reduce signaling through the RAS-MAPK pathway, kinase impaired mutations appear to promote BRAF/RAF1 isoform homo- and heterodimerization, increasing kinase transactivation and thus increasing RAS-MAPK pathway activation (171). Interestingly, the vast majority of somatic tumorigenic BRAF mutations occur in the kinase domain. It remains to be determined whether similar mutations in NS, NSML, and CFCS are more prone to develop cancer compared to mutations in other BRAF domains (170).

Activated RAF1/BRAF recruits MEK1/MEK2 and phosphorylates the proteins on two serine residues (Ser 218 and Ser 222). Missense mutations in MEK1 and MEK2 associated with CFCS have been studied and shown to activate kinase activity (117, 172). Activated MEK1/MEK2 phosphorylate and activate ERK1/ERK2, which phosphorylate a wide variety of effector proteins such as transcription factors and protein kinases, which ultimately regulate cell proliferation, growth, and differentiation.

Activated RAS-GTP is specifically phosphorylated by SRC on tyrosine residues, inactivating RAS-GTP. This phosphorylation inhibits BRAF/RAF1 binding and promotes binding of a GAP protein (p120-GAP, or possibly NF1) (161). RAS Tyr64 phosphorylation reduces association of GEF activity (173). SHP2 can dephosphorylate RAS tyrosine residues, returning the RAS-GTP to an active state. Multiple skeletal cell types including osteoblasts, osteoclasts, and chondrocytes are impacted by RAS/MAPK signaling. (174-176).

Mutation of the SPRED1 EVH1 domain eliminates SPRED1 inhibition of Ras-mediated activation of RAF1 (177). Mutations are usually truncations and loss of function. SPRED1 may also normally function by binding to NF1, recruiting it to the cell membrane in proximity to RAS, where NF1 can exert its negative regulation of RAS activity (178). Hirata et al. showed that Legius Syndrome missense mutants of SPRED1 within the EVH1 domain have reduced binding to NF1, thus recruiting less NF1 to the membrane and resulting in reduced GAP activity toward RAS (179).

5. RAS/MAPK signaling molecules involved in RASopathies and their effects on bone cells

Many components of the RAS-MAPK pathway play an important role in proliferation, differentiation, and survival of osteoblasts, osteocytes, osteoclasts, and chondrocytes. Mineralized bone is produced by the action of osteoblasts, which derive from mesenchymal progenitors that differentiate into osteoblasts and ultimately bone-encased osteocytes. The skeleton is continuously remodeled by the coordinated activity of osteoclasts, which remove mineralized bone matrix, and osteoblasts, which replace the mineralized matrix. An imbalance in skeletal remodeling caused by a reduction in osteoblast numbers or activity, an increase in osteoblast death, or increased activity of bone resorbing osteoclasts can lead to reductions in mineralized bone and osteoporosis. Linear growth occurs due to the coordinated proliferation, hypertrophy, calcification, and death of growth plate chondrocytes. Several of the RASopathy-related proteins have been studied in the context of their activities in osteoblasts, osteoclasts and/or chondrocytes, as discussed below.

5.1. RAS

RAS activity likely regulates osteoblasts differently depending on the stage of differentiation. For example, Papaioannou et al. (180) showed that expression of an oncogenic KRAS protein in COL2-positive osteoprogenitors increased bone marrow stromal cell proliferation, the number of COL2 positive progenitor cells and their descendants, and trabecular bone mass, all of which was dependent on a functional MAPK pathway. In contrast, expression of oncogenic KRAS in more mature osterix (OSX)- or COL1-positive osteoblasts did not increase the number of descendant cells or trabecular mass. Growth factor stimulation of RAS can also promote or inhibit osteoblast differentiation depending on the timing and duration of growth factor stimulation. Peak expression of epiregulin and heparin-binding epidermal growth factor (HB-EGF) correlates with the proliferative phase in MC3T3-E1 preosteoblasts, with expression waning during mineralization (181). When exogenous HB-EGF is present continuously, proliferation is favored, but differentiation is severely attenuated, an effect that is reversed by dominant negative RAS (181) and is possibly mediated due to antagonism of BMP signaling (182). In contrast, RAS activity is required for expression of runt-related transcription factor 2 (RUNX2) in a strontium treated C3H10T1/2 mesenchymal cell line (183). The RAS-related protein MRAS, in which mutations have recently been implicated in NS, is sufficient to induce osteoblastic differentiation of C3H10T1/2 mesenchymal cells, while reduced expression of MRAS inhibited osteoblastic differentiation (184). Moreover, constitutively active MRAS was capable of inducing trans-differentiation of C2C12 myocytes into osteoblasts, an activity that was blunted by inhibitors of p38 and JNK kinases, but not MEK1/MEK2 inhibitors.

Bone development requires migration of mesenchymal stem cells in response to gradients in chemotactic factors released from the surrounding matrix as well as mechanical signals received from the matrix through integrin-collagen interactions. Jiang et al. have shown that connective tissue growth factor-induced migration of mesenchymal stem cells in vivo and MC3T3-E1 cells in vitro resulted in RAS activation downstream of integrin alpha V and was inhibited by RAS inhibitor salirasib (185). Kanno et al. have shown that RAS activity was required for stretch-induced ERK activation and upregulation of RUNX2 in the MC3T3-E1 osteoblast cell line (186).

Osteoporosis with increased bone resorption has been documented in some of the RASopathies. Macrophage colony stimulating factor (M-CSF) and receptor activator of nuclear factor kappa-B ligand (RANKL) act in concert to promote proliferation, survival and differentiation of multinucleate bone resorbing osteoclasts from monocytic cell precursors (187). Activation of the RAS pathway in response to M-CSF and RANKL is important for osteoclast differentiation from mononuclear cells (188). M-CSF binding to the colony stimulating factor 1 receptor (c-FMS) triggers RAS activation and subsequent activation of MEK/ERK and PI3K signaling pathways, which is necessary for monocyte proliferation and survival (189). Interestingly, RANKL expression in osteoblasts is suppressed by mechanical stimulation in a mechanism that requires HRAS activity, which would be predicted to reduce RANKL-dependent osteoclast differentiation (190).

5.2. BRAF

BRAF activation in response to WNT3a mediated signaling activates the MAPK pathway and results in chondrocyte de-differentiation (191). Using induced pluripotent stem cell-derived mesenchymal cells produced from a CFCS patient harboring a mutant BRAF protein, Choi et al. (192) demonstrated that in vitro osteoblast differentiation was impaired and correlated with activation of MAPK and increased phosphorylation of SMAD2, while SMAD1 phosphorylation was decreased. Defective osteoblast differentiation was rescued by inhibiting MAPK or SMAD2 activation, or by inducing SMAD1 phosphorylation, suggesting a possible treatment paradigm to rescue impaired bone development in CFCS patients with BRAF mutations. Provot et al. (193) reported that ARAF and BRAF are dispensable for endochondral bone formation by knocking out the two isoforms in the chondrocyte lineage. CRAF is the predominant isoform expressed in hypertrophic growth plate chondrocytes. Papaioannou et al. knocked out ARAF, BRAF, and CRAF in chondrocytes in mice and showed that chondrocytes underwent reduced apoptosis, causing expansion of the hypertrophic zone and decreased vascular invasion (194). In a separate study, Liu et al. (195) knocked out CRAF in chondrocytes and showed that VEGF production was reduced, possibly explaining reduced growth plate vascular invasion and expansion of the hypertrophic chondrocyte zone. Patients with activating mutations in RAF family members would be expected to have increased chondrocyte apoptosis and attenuated expansion of the hypertrophic zone, potentially resulting in impaired endochondral ossification and growth. Hutchinson et al. (196) described the role of ERK/MAPK and p38 MAPK in chondrocyte proliferation and differentiation in which proliferation is favored when ERK/MAPK activity dominates over that of p38, while greater p38 activity promotes differentiation and suppresses proliferation in part via direct inhibition of RAF. RAF activity in chondrocytes may promote differentiation and support endochondral bone formation via upregulation of the cyclin dependent kinase inhibitor p21/WAF1, although hyperactivation of this pathway may result in premature slowing of chondrocyte proliferation, leading to shortened growth plates and impaired endochondral bone formation. Osteoblast ARAF and CRAF were identified as critical mediators of MAPK activation in response to mechanical stretch (197), working to promote stretch-induced differentiation (ARAF) or survival (CRAF). BRAF was identified as a mediator of cAMP induced ERK activation and cell proliferation in cultured osteoblast cell lines (198, 199).

5.3. /SHP2

Targeted deletion of Ptpn11 encoding SHP2 in the myelomonocytic lineage showed that myelomonocytic deletion of Ptpn11 caused decreased M-CSF induced ERK1/2 activation and mild osteopetrosis in a mouse model (200). Ptpn11 has also been deleted in osteocalcin-expressing osteoblasts and chondrocytes, which resulted in profound reductions in cortical and trabecular bone as well as decreased bone strength, possibly due to reduced RUNX2 and Osterix transcriptional activity (201). Wang et al. (201) also demonstrated that SHP2-deficient mice had increased osteoclast activity, possibly due to increased RANKL expression in osteoblasts and chondrocytes. Ramifications of SHP2 deficiency in cartilage include development of enchondromas and exostoses, as well as abnormal expansion of prehypertrophic chondrocytes in the growth plate (201-203). Tajan et al. (142) described hyperactivation of the RAS/MAPK pathway and reductions in growth plate length in SHP2 mutant mice, primarily due to a shorter hypertrophic zone. Interestingly, treating SHP2 mutant mice with statins improved growth plate architecture in vivo and chondrocyte differentiation in vitro.

5.4. NF1

Neurofibromin is expressed throughout development in mouse periosteal osteoblasts and osteoclasts, cortical bone osteocytes, and all stages of chondrocyte maturation, with the notable exception of proliferating chondrocytes, where expression is low to undetectable (204). Heterozygous loss of Nf1 function in mouse embryos resulted in increased ERK1/ERK2 phosphorylation in hypertrophic chondrocytes (204) and hyperactivation of RAS in osteoblast progenitors (132). Although the Nf1 haploinsufficient mouse has a very subtle osteopenic phenotype (132), committed osteoprogenitors derived from Nf1 haploinsufficient mice were shown to have increased proliferation rates accompanied by decreased expression of osteoblast differentiation marker genes (132, 205). Expression of the NF1 GAP-related domain, or inhibition of MEK, was sufficient to rescue in vitro osteoblast differentiation of committed osteoprogenitors derived from Nf1 haploinsufficient mice (205, 206), while Sullivan et al. (207) reported that inhibition of JNK activity, in combination with rhBMP-2, promoted osteoblast differentiation of NF1 deficient osteoprogenitors.

Fibroblast growth factor receptor 1(FGFR1), FGFR3, epidermal growth factor receptor (EGFR) and insulin-like growth factor receptor (IGFR1) initiate signaling events through the RAS-MAPK pathway that are critical for growth plate chondrocyte development [reviewed in (175)]. Neurofibromin expression overlaps with that of FGFR1 and FGFR3 in mouse growth plate chondrocytes (208), where deletion of Nf1 in Col2a1-positive chondrocytes and osteoprogenitors in mice results in constitutive activation of the MAPK/ERK pathway, reduced chondrocyte proliferation, shortened growth plates, defective endochondral bone formation, post-natal growth retardation, and other skeletal abnormalities seen in some humans with NF1 (209, 210). Nf1 deletion can restore normal MAPK/ERK signaling and aberrant growth plate organization in FGFR1 deficient hypertrophic chondrocytes, where it suppresses osteoclast development by way of reducing RANKL expression, and supports normal endochondral bone formation (208). Expression of constitutively active forms of MEK1 in growth plate chondrocytes, similar to deletion of Nf1, resulted in reduced chondrocyte hypertrophy and delayed endochondral bone formation (211).

Although most of the focus on skeletal issues in NF1 have been on bone forming cells, other data suggests that development of osteoclast lineage cells may also be affected by Nf1 haploinsufficiency (212). Rhodes et al. (213) showed that myeloid progenitor-restricted, but not mature osteoclast-restricted, Nf1 haploinsufficiency resulted in enhanced osteoclast development and resorptive activity, suggesting that regulation of RAS/MAPK activity is critical for early stages of osteoclastogenesis. Mononuclear cells isolated from mouse models of Nf1 haploinsufficiency or human NF1 patients exhibit increased osteoclast activity (214) (215, 216), although one study of isolated mononuclear cells from patients with or without NF1 showed that despite increased basal MAPK/ERK activation in cells from NF1 patients, there was no difference in osteoclast development in vitro (217).

Cell autonomous and non-autonomous mechanisms may be operative in the increased osteoclast activity observed in Nf1^+/− cells. Hypersensitivity of osteoclast precursors to M-CSF and RANKL, critical osteoclast developmental regulators, has been documented in Nf1 deficient cells (215). The bone resorbing capacity of osteoclasts is highly dependent on the actin cytoskeleton, which is regulated in part via the RHO/RAC family of small GTPases (218, 219), downstream mediators of the RAS pathway. Deletion of RAC1 restores to normal osteoclast formation and activity in Nf1^+/− mice. Other studies have demonstrated that production of osteoclast regulatory proteins by cells other than osteoclasts may help to explain the osteoclast hyperactivation in NF1. Rhodes et al. (220) reported that serum concentration of TGF-β1 was increased in both a mouse model of osteoblast-restricted Nf1 haploinsufficiency and human patients with NF1, while Nf1 deficient osteoclasts were more responsive to TGF-β1 signaling, forming greater numbers of mature, bone resorbing osteoclasts. Osteopontin (OPN), another osteoblast product that can potentiate osteoclast development and activity, was shown to be produced at higher levels in Nf1^+/− osteoblasts (132, 221), with Nf1^+/− osteoblast conditioned media stimulating migration and bone resorption in vitro, an effect that could be attenuated by OPN neutralizing antibodies (221). Furthermore, Nf1^+/− osteoclasts were more responsive to the potentiating effects of OPN (221). Overproduction of RANKL by Nf1^−/− osteoprogenitors suggests that RANKL overproduction could also contribute to the enhanced osteoclastogenesis observed in NF1 (138).

5.5. MEK 1/2

In vitro studies of mouse calvarial osteoblasts expressing constitutively-active or dominant-inhibitory forms of MEK1 showed increased or decreased osteoblastic differentiation, respectively (222). Consistent with this observation, inhibition of MEK1/2 activity reduced osteoblast gene expression in MC3T3 mouse osteoblast cells (223). The osteoblast differentiation promoting activity of MEK/ERK signaling acts at least in part via phosphorylation of the critical osteoblast transcription factor, RUNX2 (222, 224, 225). Other studies have shown opposite effects of MEK/ERK signaling, i.e., chronic inhibition of MEK/ERK signaling was reported to stimulate osteoblast differentiation and mineral formation in vitro (226), while stimulation of MEK/ERK signaling by FGF-2 and EGF was shown to inhibit MC3T3 response to BMP-2, an effect that was reversed by dominant inhibitory RAS or a MEK1 inhibitor (182).

M-CSF and RANKL interact with c-FMS and RANK, respectively, to stimulate proliferation (M-CSF) and differentiation (RANKL) of osteoclast precursors (227, 228). Each pathway activates the MAPK/ERK cascade with different kinetics; c-FMS activation causes a strong, sustained activation of MAPK/ERK signaling that is linked to osteoclast progenitor proliferation, while RANKL-RANK induced MAPK/ERK signaling is weaker and biphasic that is linked to differentiation (reviewed in (176). Importantly, altering the dynamics of MAPK/ERK signaling as occurs in RASopathies would be predicted to lead to altered osteoclast differentiation. Consistent with the observation that overactive RAS-MAPK signaling promotes osteoclastogenesis, Wang et al. (200) showed that myelomonocytic deletion of SHP2 caused decreased M-CSF induced ERK1/2 activation and mild osteopetrosis in a mouse model.

5.6. CBL

CBL (or c-CBL; named after Casitas B-lineage Lymphoma) plays many roles in cellular functions, including ubiquitylation-dependent events and adaptor functions, and it is involved in multiple intracellular signaling events (229). The function of c-CBL has been studied in the context of skeletal biology, especially as it relates to osteoclast physiology. However, the specific c-CBL interactions most commonly explored in bone cells are those involving SRC and PI3K. When c-CBL is eliminated in osteoclasts, which would affect all its various functions, osteoclasts have been noted to have decreased motility and defective bone resorption capabilities (230). In relation to osteogenesis, c-CBL has been shown to be involved in osteoblast proliferation and differentiation (231). In fracture healing models c-CBL-PI3K interactions are important in the osteogenic response of activated periosteal cells in the healing process (232). Specific knock-in mutations known to occur in CBL-mediated RASopathies are yet to be explored in the context of specific bone cell types.

6. Potential therapeutic interventions

A number of investigations in mice have been carried out to understand the underlying mechanisms and signaling pathways that manifest in skeletal consequences of RASopathies. These studies have inevitably led to intervention studies to determine if targeted approaches could be used to intervene in preventing skeletal pathology in mouse models of RASopathies. Increased RAS-MAPK signaling in NF1 was shown to cause accumulation of inorganic pyrophosphate (PPi), an inhibitor of hydroxyapatite formation as a result of increased expression of genes known to regulate pyrophosphate synthesis (136). This pathway was subsequently targeted in mice lacking neurofibromin in osteochondroprogenitors or osteoblasts using asfotase alfa enzyme (human tissue nonspecific alkaline phosphatase isoenzyme) therapy, resulting in improvements in stature, bone mineralization and strength in these mice, thus providing evidence that bone mineralization defects in NF1 might be modifiable (136). In a single case report, asfotase alfa was used in conjunction with zolendronic acid after spinal surgery in an individual with NF1. Good healing was reported; however, it was not possible to know what specific role asfotase alfa has in the healing process (233). Fracture healing, which is impaired in mouse models in which neurofibromin is ablated in osteoprogenitors, is improved when co-therapy with a MEK inhibitor and BMP2 is used (234). Other studies exploring specifically the role that hyperactivity of TGF-β may play in skeletal deficits in NF1 have shown that a TGF-β receptor 1 kinase inhibitor improves bone mass deficits and fracture healing in the in Nf1^flox/− /COL2.3-cre mouse (220). Prenatal inhibition of the RAS-MAPK pathway has also been successful in preventing some of the skeletal deficits observed in mice in which neurofibromin is ablated in osteoprogenitors (209), as well is in mouse models of Noonan Syndrome (143, 144) and CFC (154). Together, these studies suggest that specific skeletal deficits observed in RASopathies can be targeted from a pharmacologic approach; however, specificity and off-target effects of these potential therapies, as well as the optimal window for their application, remain outstanding considerations and limitations to advancing their use in humans.

7. Summary

RASopathies are an ever expanding group of complex phenotypes representing a wide range of mutations in critical signaling molecules functioning within the RAS/MAPK pathway. There can be significant overlap in phenotype amongst different RASopathies, yet each presents also with unique features. Because the RAS/MAPK pathway is important in normal skeletogenesis, the skeleton is commonly affected in many RASopathies as described herein. Despite understanding more about the specific genetic mutations involved in each RASopathy, many questions still remain as to why skeletal phenotypes vary amongst RASopathies and even within the same RASopathy; which bone cell types are most affected by specific mutations and how they affect their behavior; why specific bones within the skeleton are affected more than others, and what could be effective and safe treatments to manage the skeletal outcomes of RAS/MAPK mutations.

HIGHLIGHTS.

• RASopathies are caused by germline mutations in the Ras/MAPK pathway

• RASopathies exhibit craniofacial, cardiac, and neurodevelopmental features

• RASopathies demonstrate overlapping yet distinct musculoskeletal phenotypes

• RAS/MAPK signaling is integral for osteoblast, osteoclast, and chondrocyte biology

• Mouse models of RASopathies display musculoskeletal findings similar to humans