Noonan syndrome (NS [MIM 163950]) is a clinically variable disorder defined by short stature, facial dysmorphism, multiple skeletal defects, and congenital heart disease.^1–3 The distinctive facial features consist of a broad forehead, hypertelorism, down-slanting palpebral fissures, high-arched palate, and low-set, posteriorly rotated ears. Cardiac involvement is present in up to 90% of affected individuals, with pulmonic stenosis, hypertrophic cardiomyopathy, atrioventricular septal defects, and aortic coarctation representing the most common lesions.⁴ Additional relatively recurrent features of NS are webbed and/or short neck, mental retardation, cryptorchidism, and hematologic anomalies. Although precise epidemiological data are not available, the prevalence of NS is estimated to be between 1 in 1,000 and 1 in 2,500 live births.⁵

NS is an autosomal dominant trait and is genetically heterogeneous.⁶ Using a positional candidacy approach, we established PTPN11 as the NS disease gene residing at NS1 (12q24),⁷ and studies by our groups and others have demonstrated that mutations in PTPN11 account for nearly 50% of affected individuals.⁵ The PTPN11 gene encodes the cytoplasmic protein tyrosine phosphatase SHP-2, a widely expressed signal transducer with crucial roles in the cellular response to growth factors, hormones, and cytokines.⁸ Accumulating genetic, modeling, and biochemical data indicate that NS-causing PTPN11 mutations promote SHP-2 gain of function.^7,9–12 In contrast, a limited number of germline PTPN11 missense mutations that seem to engender dominant negative effects cause the phenotypically related LEOPARD syndrome (MIM 151100).^12–15 Finally, a distinct class of somatic PTPN11 mutations promoting stronger activation of SHP-2 has been identified as contributing to leukemogenesis.^{11,12,16–18}

SHP-2 is an essential component of the machinery required for the activation of the RAS-mediated intracellular signaling pathways, and converging data indicate that germline and somatic PTPN11 mutations cause deregulation of RAS function.^{15,16,19–22} RAS proteins act as GDP/GTP–regulated molecular switches to control intracellular signal flow.^23,24 They exhibit high affinity for both GDP and GTP and low GTPase activity. GDP/GTP cycling is controlled by GTPase-activating proteins (GAPs), which accelerate the intrinsic GTPase activity, and by guanylyl-exchanging factors (GEFs), which promote release of GDP. RAS proteins cycle from a GDP-bound inactive state to a GTP-bound active state, the latter allowing signal flow by protein interaction with multiple downstream transducers. RAS proteins share a structure that includes a conserved domain (residues 1–165), known as the “G domain,” which is required for signaling function, and a less conserved C-terminal tail, called the “hypervariable region,” that guides posttranslational processing and plasma membrane anchoring (fig. 1A). Within the conserved region, five motifs (G-1 to G-5) direct the GTP/GDP binding and exchange and the GTP hydrolysis.²⁵ Furthermore, two tracts, usually denoted as “Switch I” (residues 32–38) and “Switch II” (residues 59–67), undergo major conformational changes on GTP/GDP exchange and mediate binding to effectors, GAPs, and GEFs.^24,26,27

Open in a separate window

Figure 1.

Heterozygous KRAS missense mutations causing a severe Noonan syndrome phenotype with features overlapping those of CFC and CS. A, Schematic representation of the structural and functional domains defined within RAS proteins.²⁵ The conserved domain (G domain) is indicated, together with the motifs required for signaling function. The hypervariable region is shown (gray), as is the C-terminal motifs that direct posttranslational processing and plasma membrane anchoring (dark gray). The arrow indicates the location of affected residues. B, KRAS gene organization and transcript processing that produces the alternative KRAS isoforms A and B (^Gene accession numbers NC_000012, NM_004985, and NM_033360). The numbered black and gray boxes indicate the invariant coding exons and the exons undergoing alternative splicing, respectively. KRASB mRNA results from exon 5 skipping. In KRASA mRNA, exon 6 encodes the 3′ UTR. The arrows indicate locations of mutations. C, DHPLC elution profiles (left) and electropherograms (right) of KRAS exon 6 PCR products showing the de novo heterozygous 455T→G (above) and 458A→T (below) changes. D, Dysmorphic facial features of individual 2 (with Asp153Val).

Given the link between the functions of SHP-2 and RAS, we hypothesized that other genes encoding for proteins functioning as transducers in RAS signaling might be implicated in NS pathogenesis. Here, we report the identification of heterozygous KRAS gene mutations in a small percentage of subjects exhibiting a severe NS phenotype.

Eighty-seven subjects with NS and eight subjects with cardiofaciocutaneous (CFC) syndrome (MIM 115150) were included in the study. All the individuals were of European origin and lacked a PTPN11 coding-sequence mutation¹² (M.T., B.D.G., and B.D., unpublished data). Informed consent for genetic analyses was obtained from all subjects considered in the study. For the vast majority of these individuals, clinical features satisfied diagnostic criteria for NS²⁸ and CFC syndrome,²⁹ but a few subjects who did not have features sufficient to make a definitive diagnosis were also considered. Genomic DNA was isolated from peripheral blood leukocytes, and the entire KRAS sequence coding for both the KRASA and the KRASB isoforms (exons 2–6) (fig. 1B) was screened for mutations. Primer pairs designed to amplify exons, exon/intron boundaries, and short intron flanking stretches are listed in table 1. The entire coding sequences of the NRAS, HRAS, and DUSP6 genes, as well as the NF1 genomic portion coding for the putative cysteine/serine–rich (exons 11–17) and RASGAP (exons 21–27) domains of neurofibromin, were also screened for mutations. Primer sequences and PCR conditions are available on request. Mutation analysis of the amplimers was performed with denaturing high-performance liquid chromatography (DHPLC) by use of the Wave 2100 System (Transgenomic) at column temperatures recommended by Navigator software, version 1.5.4.23 (Transgenomic). Amplimers with abnormal denaturing profiles were purified (Microcon PCR [Millipore]) and were sequenced bidirectionally using the ABI BigDye Terminator Sequencing Kit v.1.1 (Applied Biosystems) and ABI Prism 310 Genetic Analyzer (Applied Biosystems). Paternity was confirmed by STR genotyping (AmpF/STR Identifiler PCR amplification kit [Applied Biosystems]). Structural analyses and molecular graphics were performed using the program MOLMOL.³⁰

Mutation analysis allowed the identification of heterozygous KRAS mutations in two individuals (fig. 1C). Both mutations were missense and affected exon 6, which encodes the C-terminal portion of KRAS isoform B.

A T→G transversion at position 455, predicting the substitution of Val¹⁵² by a glycine residue (Val152Gly), was identified in a 1-year-old girl who had macrocephaly with high and broad forehead, curly and sparse hair, hypertelorism, strabismus, epicanthic folds, down-slanting palpebral fissures, hypoplasic nasal bridge with bulbous tip of the nose, high palate and macroglossia, low-set and posteriorly rotated ears, short neck with redundant skin, wide-set nipples, and umbilical hernia. She was born at 32 wk of gestation by cesarean section after a pregnancy complicated by a cystic hygroma detected at 12 wk and polyhydramnios at 30 wk. At birth, her weight was 2.550 kg (+3 SD), length was 43 cm (−0.5 SD), and head circumference was 32.5 cm (+2 SD), and she showed edema of lower limbs. Clinical examination at age 1 year showed severe failure to thrive because of poor intake, with weight, length, and head circumference at −5 SD, −5 SD, and +1 SD, respectively. No skin anomaly was noted, and no cardiac defect was documented by echocardiographic examination. The phenotype was compatible with a severe NS condition with features overlapping Costello syndrome (CS [MIM 218040]) (polyhydramnios, neonatal macrosomia and macrocephaly, loose skin, and severe failure to thrive) and, to a lesser extent, CFC syndrome (macrocephaly and sparse hair).

An A→T change at position 458, predicting the substitution of the adjacent aspartic acid residue by valine (Asp153Val), was observed in a 14-year-old girl with NS and some features of CFC syndrome (fig. 1D). She had short stature and growth retardation with delayed bone age, cardiac defects (moderate ventricular hypertrophy, mild pulmonic stenosis, and an atrial septal defect), dysmorphic features (hypertelorism, down-slanting palpebral fissures, strabismus, low-set and thick ears, relative macrocephaly with high forehead, and a depressed nasal bridge), short and mildly webbed neck, wide-set nipples, and developmental delay. There was hyperpigmentation of the skin and a large café-au-lait spot on the face. The creases of the hands and feet were deep. There was no keratosis or ichthyosis. The nails and teeth were normal. The hair was dry, thick, and without fracture and was wavy but not curly. Abnormalities of the extremities included cubitus valgus and joint hyperextensibility but not clinodactyly. Her gestation was complicated by polyhydramnios. She had feeding difficulties until age 4 years and subsequently experienced cyclic vomiting. She had a single seizure, at age 6 years. Application of the CFC index, a comprehensive scoring system developed to aid in diagnosing this disorder,²⁹ yielded a score of 9.483, which is just below the 5th percentile among patients with CFC.

Genotyping of parental DNAs confirmed paternity in both families and demonstrated the de novo origin of the mutations. Analysis of genomic DNA obtained from buccal cells of the affected individuals documented the occurrence of mutations, indicating that both lesions were germline events.

Val¹⁵² and Asp¹⁵³ reside far from the portion of RAS typically altered in cancer,³¹ which suggests a different pathogenetic mechanism for NS. To explore this, we used crystallographic information available on RAS family proteins. Although the crystallographic structure of KRAS per se has not been resolved, the available structures for other members of the RAS family are largely superimposable. The three-dimensional structures of HRAS (Protein Data Bank codes 4Q21 [for GDP-HRAS] and 5P21 [for GTP-HRAS]) were considered in our analysis because many structures of this protein have been resolved, including those of cancer-related mutants and complexes with GEF and GAP proteins. HRAS is 94% homologous to KRAS isoform B within the G domain.²⁵ In HRAS, a glutamic acid residue replaces Asp¹⁵³ in KRAS. We assumed, however, that the structure of this region was retained, because the substitution was conservative and Glu¹⁵³ and Asp¹⁵³ are almost equally represented among RAS proteins. Moreover, the structure of RRAS (Protein Data Bank code 2FN4), which has an aspartic acid residue at this position, shows almost no structural alteration (fig. 2A). The mutated residues Val¹⁵² and Asp¹⁵³ are located at the N-terminus of the α-helix (α5) spatially close to the G-5 region (residues 145–147)²⁵ (fig. 2B). The conformation of this region, which constitutes part of the guanine-binding pocket, is marginally affected by GDP/GTP switching (fig. 2C). The guanine binding is maintained by an H-bonding network involving residues Asp¹¹⁹ and Ala¹⁴⁶, which interact with the purine ring, and residues Gly¹³, Val¹⁴, Asn¹¹⁶, Lys¹¹⁷, Thr¹⁴⁴, and Ser¹⁴⁵, which strengthen those interactions. Val¹⁵² and Asp¹⁵³ contribute to stabilizing the guanine-binding pocket. The side chain of Val¹⁵² participates in the stabilization of a very tight hydrophobic core with residues Leu¹⁹, Leu²³, and Ala¹⁴⁶ and the aliphatic chains of Gln²² and Arg¹⁴⁹ (fig. 2B). The interactions with these amino acids strongly guide the orientation of the α5 helix and, as a consequence, of the loop containing the G-5 tract involved in the interaction with the purine ring. The introduction of a glycine residue at position 152 is predicted to affect the stability of that hydrophobic core. In addition, this change would endow the peptide chain with higher flexibility due to the presence of a Gly¹⁵¹Gly¹⁵² motif.³⁴ Glu¹⁵³ forms a salt bridge with Arg¹⁴⁹ (fig. 2B), an interaction that is retained in the comparable Asp residue in RRAS protein, with a shorter distance between the charged groups (fig. 2A). Loss of this salt bridge, caused by the Asp153Val mutation, is predicted to destabilize the conformation of the loop connecting the α5 and β6 tracts (residues 144–151). Overall, these observations strongly indicate that the Val152Gly and Asp153Val substitutions destabilize the conformation of domains that contribute extensively to the interaction of KRAS with the GTP/GDP guanine ring and, consequently, are predicted to affect binding to both GTP and GDP. Since these residues are also involved in the binding of GTP-RAS with the human GEF protein SOS,³⁵ we cannot exclude the possibility that additional mechanisms, including disruption of possible regulatory interactions with GEF proteins, might also contribute to the pathological effects of these mutations. Since the 455T→G and 458A→T changes resided close to the splice-acceptor site, we considered the possibility of disruptive effects on an exonic splicing enhancer (ESE) site, using an informatics approach. ^ESEfinder analysis³⁶ identified a single site for the splicing factor SC35 (positions 13–20), which was not altered by these mutations. We cannot exclude the possibility that the mutations altered a site for an as-yet-unidentified ESE.

Open in a separate window

Figure 2.

Structural analyses. A, Superimposition of the crystallographic structures of HRAS (yellow) and RRAS (light blue), in the region encompassing the β6 strand and α5 helix. The side chain of HRAS residues Arg¹⁴⁹ (β6-α5 loop) and Glu¹⁵³ (α5) and the corresponding RRAS residues Arg¹⁷⁶ and Asp¹⁸⁰ are shown as a stick model. B, Crystallographic structure of GTP-HRAS³² showing the interactions of residues 152 and 153. The side chain of Val¹⁵² (red surface) participates in a hydrophobic core involving residues Leu¹⁹, Gln²², and Leu²³ (gray surfaces); Ala¹⁴⁶ (pink surface); and Arg¹⁴⁹ (semitransparent yellow surface). The side chains of residues 149 and 153, forming a salt bridge, are shown as a stick model. The yellow dashed line indicates the hydrogen bond formed by the N-H of the Ala¹⁴⁶ backbone and the O6 atom (red stick) of the GTP molecule (blue sticks). C, Superimposition of the crystallographic structures of GTP-HRAS³² (light blue) and GDP-HRAS³³ (yellow) complexes. A ribbon representation is used for the protein backbone, whereas GDP and GTP are shown as a stick model. The positions of the mutated residues (N-terminus of the α5 helix) and the G-5 motif are shown in red and pink, respectively. Regions assuming a different conformation in the two complexes (Switch I [residues 32–38] and Switch II [residues 59–67]) are shown in orange (in GDP-HRAS) and blue (in GTP-HRAS).

Following the discovery of PTPN11 mutations as a major underlying cause of NS and LEOPARD syndrome,^7,13,14 other developmental disorders clinically related to NS have been demonstrated to be caused by defects in genes that encode proteins functionally related to SHP-2, including NF1 mutations and/or deletions in neurofibromatosis-Noonan syndrome (NFNS [MIM 601321]),^37,38 HRAS mutations in CS,³⁹ and, most recently, BRAF, MEK1, and MEK2 mutations in CFC syndrome.⁴⁰ Here, we documented that KRAS gene mutations affecting KRAS isoform B can cause a severe form of NS, with a phenotype including features of CS and CFC syndrome. Studies published while our work was in review confirmed independently the causative role of KRAS mutations in NS and CFC syndrome pathogenesis.^41,42 Although the initial finding concerning PTPN11 in NS implicated signaling downstream of RAS, it was unclear whether one or more pathways were affected. Taken together, the newer genetic findings specifically implicate perturbed signaling through the mitogen-activated protein kinase (MAPK) cascade. NS, LEOPARD syndrome, CFC syndrome, CS, and NFNS can now be grouped as developmental disorders resulting from mutations in functionally related genes. Although a significant portion of NS cases remain unexplained genetically, it appears highly likely that several additional genes related to RAS/MAPK signal transduction will ultimately be implicated.

RAS mutations associated with cancer generally upregulate RAS function by impairing the switch between the active and inactive conformation, favoring a shift in the equilibrium toward the former.^31,43,44 The same is true of HRAS mutations causing CS³⁹ and a few KRAS mutations recently identified in NS⁴² and CFC syndrome.⁴¹ The NS-associated KRAS mutations described in this report, one of which was independently reported elsewhere,^41,42 are predicted to represent a novel gain-of-function mechanism. This appears to stem from an increased rate of GDP-GTP exchange related to a reduced affinity for GDP and GTP. This assertion is supported by biochemical data^45–48 from related RAS mutants characterized by substitutions involving residues Asn¹¹⁶, Asp¹¹⁹, and Thr¹⁴⁴. In particular, biochemical and functional characterization of two HRAS mutants (Asp119Asn and Thr144Ile) demonstrated that an increase in rates of dissociation of GDP/GTP from the protein does not prevent the mutant RAS proteins from binding GTP in vivo and inducing transformation.⁴⁹ The mechanism leading to this gain-of-function effect has been characterized. Briefly, the inactive complex between wild-type RAS and GDP is normally stable, requiring interaction with GEF proteins for its dissociation. Mutations destabilizing the GDP-bound state favor spontaneous dissociation of this complex. In these mutants, unbound RAS tends to complex with GTP, because GDP and GTP have similar dissociation constants⁴⁷ but GTP has a significantly higher concentration in the cytoplasm (GTP:GDP ratio of 25:1).⁵⁰ On the basis of our structural analysis and these related biochemical data, we suggest that the Val152Ala and Asp153Val substitutions perturb the conformation of the guanine ring-binding pocket, increasing the GDP/GTP dissociation rate. This would shift the equilibrium toward the GTP-bound, active form and bypass the requirement for a GEF. These mutations would define a novel class of activating lesions that are inherited germline and that perturb development. To our knowledge, they have not been documented to occur as somatic lesions contributing to human hematologic malignancies or solid tumors. Functional studies to verify this hypothesis are in progress.

Both KRAS mutations identified in this study affected exon 6, which contributes to KRASB but not to KRASA.⁵¹ KRASA and KRASAB differ at their C-termini, regions subjected to posttranslational modifications.⁵² Similar to HRAS and NRAS, KRASA is palmitoylated at cysteine residues upstream of the conserved CAAX motif, which are replaced with a polylysine stretch in KRAS isoform B. This differential processing has profound functional effects, leading to alternative trafficking pathways to the plasma membrane and distinct membrane localization.⁵¹ Recent evidence demonstrates that the two KRAS isoforms play distinct roles in development. While KRASB is ubiquitously expressed in embryonic and adult tissues, KRASA expression is restricted temporally and spatially and is not expressed in the adult heart.⁵³ Consistent with these data, loss of both KRAS isoforms is embryonic lethal,^54,55 whereas absence of only KRASA does not perturb development.⁵³ Although KRAS mutations affecting domains shared by the two isoforms can cause NS and CFC syndrome,^41,42 the identification of exon 6 mutations documents that isolated KRASB gain of function is sufficient for disease pathogenesis—further evidence that KRAS isoform B plays the major role in development. Additional studies are required to delineate the phenotypic spectrum resulting from germline KRAS mutations as well as their molecular diversity and functional consequences for intracellular signaling and development.