Exome Sequencing Identifies GNB4 Mutations as a Cause of Dominant Intermediate Charcot-Marie-Tooth Disease

Abstract

Charcot-Marie-Tooth disease (CMT) is a heterogeneous group of inherited neuropathies. Mutations in approximately 45 genes have been identified as being associated with CMT. Nevertheless, the genetic etiologies of at least 30% of CMTs have yet to be elucidated. Using a genome-wide linkage study, we previously mapped a dominant intermediate CMT to chromosomal region 3q28–q29. Subsequent exome sequencing of two affected first cousins revealed heterozygous mutation c.158G>A (p.Gly53Asp) in GNB4, encoding guanine-nucleotide-binding protein subunit beta-4 (Gβ₄), to cosegregate with the CMT phenotype in the family. Further analysis of GNB4 in an additional 88 unrelated CMT individuals uncovered another de novo mutation, c.265A>G (p.Lys89Glu), in this gene in one individual. Immunohistochemistry studies revealed that Gβ₄ was abundant in the axons and Schwann cells of peripheral nerves and that expression of Gβ₄ was significantly reduced in the sural nerve of the two individuals carrying the c.158G>A (p.Gly53Asp) mutation. In vitro studies demonstrated that both the p.Gly53Asp and p.Lys89Glu altered proteins impaired bradykinin-induced G-protein-coupled-receptor (GPCR) signaling, which was facilitated by the wild-type Gβ₄. This study identifies GNB4 mutations as a cause of CMT and highlights the importance of Gβ₄-related GPCR signaling in peripheral-nerve function in humans.

Main Text

Charcot-Marie-Tooth disease (CMT) is a group of inherited neuropathies characterized by progressive distal muscle atrophy, distal sensory loss, depressed tendon reflexes, and foot deformities; they can be categorized into either the demyelinating form or the axonal form by their electrophysiological and pathological features.^1,2 Clinically, median motor nerve conduction velocity (NCV) is the most frequent parameter (with a cutoff value of 38 m/s) used for distinguishing between demyelinating and axonal CMT.³ Mutations in approximately 45 genes have been identified as being associated with CMT, but these account for fewer than 70% of all CMT cases.^1,2 Within the CMTs, dominant intermediate CMT (DI-CMT) is a rare group that shows autosomal-dominant inheritance and variable NCVs, which can look like either demyelinating or axonal types of CMT within a pedigree.^3–6 Previously, we characterized a three-generation Chinese family afflicted with a DI-CMT and mapped the locus of the candidate gene to chromosomal region 3q28–q29.⁷ In this study, 13 family members, including six affected individuals, five unaffected individuals, and two married-in spouses, were enrolled (Figure 1A). Disease status was determined by both the clinical manifestations and electrophysiological evidence.³ In brief, the age of onset ranged from 10 to 45 years, and individuals ranged in clinical severity from being asymptomatic to being wheelchair-bound. The median motor NCVs ranged from 16.5 to 45.7 m/s. Men tended to be more severely affected (Table 1). Sural nerve biopsies of two affected individuals detected loss of myelinated fibers and also demyelinating changes as evidenced by multiple onion-bulb formations. Written informed consent was obtained from all subjects, and the study protocol was approved by the institutional review board at Taipei Veterans General Hospital, Taipei, Taiwan. Genomic DNA was isolated from peripheral-blood leukocytes according to standard protocol.

Figure 1.

Pedigrees and Electropherograms of the GNB4 Mutations

(A and B) Pedigree structure of families 1 (A) and 2 (B). The probands are denoted by an arrow. The squares and circles denote males and females, respectively, and the diagonal black lines represent deceased individuals. Filled symbols represent affected members with neuropathy, and open symbols indicate unaffected individuals. The numbers in diamonds indicate the number of unaffected siblings or children. The gender and birth order have been partially hidden for the sake of confidentiality. “M” represents the GNB4 mutant allele, “W” stands for the wild-type allele, and asterisks denote members who underwent whole-exome sequencing.

(C) The electropherograms of the heterozygous GNB4 mutations, c.158G>A (p.Gly53Asp) and c.265A>G (p.Lys89Glu).

(D) RT-PCR analysis of the RNA isolated from the leukocytes of a normal control and the individual carrying GNB4 c.265A>G with the use of primers specific to GNB4 exons 5 and 6. Both samples revealed a single band at the same location (218 bp), indicating that GNB4 mRNA from the affected individual was not differentially spliced.

(E) The cDNA electropherograms of the GNB4 exon 5 and 6 junction of a normal control and the affected individual carrying GNB4 c.265A>G reveal similar splicing patterns.

Table 1.

Clinical Features of the Individuals with GNB4 Mutations

	Family 1 (c.158G>A [p.Gly53Asp])						Family 2 (c.265A>G [p.Lys89Glu])
	III-2	III-3	IV-1	IV-2	IV-3	IV-4	II-1
Age of onset (years)	13	45	10	13	17	22	5
Age at exam (years)	49	48	23	18	17	22	9
Gender	male	female	male	male	female	female	female
Upper-limb weakness (MRC)	distal: 4	normal	normal	normal	normal	normal	normal
Lower-limb weakness (MRC)	distal: 2	distal: 4- to 4	distal: 4- to 4	distal: 4	distal: 4+	distal: 4+	distal: 4
Upper-limb muscle atrophy	distal: moderate	normal	distal: mild	normal	normal	normal	normal
Lower-limb muscle atrophy	distal: severe	distal: mild	distal: severe	distal: mild	distal: mild	distal: mild	distal: moderate
Knee and ankle DTRs	absent	absent	absent	absent	absent	reduced or absent	absent
Gait	bound in wheelchair	steppage	steppage	steppage	normal	normal	steppage
Pinprick sensation	reduced	mildly reduced	mildly reduced	mildly reduced	normal	normal	mildly reduced
Vibration	reduced	mildly reduced	mildly reduced	mildly reduced	normal	normal	mildly reduced
Median nerve MNCV (m/s)	25.2	35.4	16.5	28.6	44.2	45.7	20
Median nerve cMAP (mV)	7.7	8.3	3.9	3.2	9.3	5.9	2.6
Sural nerve SNAP	NR	NR	NR	NR	NR	7	NR
Sural nerve biopsy	demyelination	demyelination	ND	ND	ND	ND	ND

The following abbreviations are used: MRC, Medical Research Council scale; DTR, deep tendon reflex; MNCV, motor nerve conduction velocity; cMAP, compound motor action potential; SNAP, sensory nerve action potential; NR, not recordable; and ND, not done.

About 165,000 exonic regions from the genomic DNA of the two affected first cousins of family 1 (IV-1 and IV-4 in Figure 1A) were captured and enriched with the Agilent SureSelect Human All Exon v.2 kit. The DNA-sequencing libraries were prepared with an Illumina Paired-End DNA Sample Preparation kit. A total of 100 bp paired-end reads were obtained by sequencing on an Illumina Genome Analyzer GAIIx. All sequence reads were mapped to the human reference genome (GRCh37 patch 2) with the Burrows-Wheeler Aligner.⁸ Removal of PCR duplicates and variant calling were performed with SAMtools/BCFTools.⁹ Only the variants with Phred-like scores of at least 30.0 were collected for the investigation of candidate variants of high confidence. On average, 6 Gb of sequence was generated from each individual, resulting in a coverage depth of 83× per base per targeted region. After analysis (Table S1, available online), 64 heterozygous coding variants were found in both individuals; these variants were not present in dbSNP, the 1000 Genomes Project,¹⁰ or the in-house exome data from ten non-CMT ethnic controls. Targeted sequencing of eleven other members of family 1 revealed that only the c.158G>A (p.Gly53Asp) variant in GNB4 (MIM 610863; RefSeq accession number NM_021629.3), encoding guanine-nucleotide-binding protein (G protein) subunit beta-4 (Gβ₄), perfectly segregated with the CMT phenotype (Figures 1A and 1C). This variant was also located within the chromosomal region 3q28-q29, which was the previously mapped locus for this type of CMT.

To support that mutations in GNB4 cause CMT and to investigate the frequency of GNB4 mutations in our CMT population, we further sequenced all nine coding exons of GNB4 in another 88 unrelated individuals with molecularly unassigned CMT (Table S2); these individuals included 66 with the demyelinating variant and 22 with the axonal variant. Among them, one additional de novo mutation, c.265A>G (p.Lys89Glu) (Figures 1B and 1C) in the heterozygous form, was identified. Neither the GNB4 c.158G>A mutation nor the c.265A>G mutation was present in the Exome Variant Server of the National Heart, Lung, and Blood Institute (NHLBI) Exome Sequencing Project or in 1,920 ethnically matched control chromosomes. Gly53 and Lys89 are highly conserved residues in Gβ4 (Figure S1). SNPs&GO,¹¹ Panther,¹² and Polyphen2¹³ predicted both mutations to be deleterious to Gβ₄ functioning¹⁴ (Table S3). Because the c.265A>G mutation occurs at the nucleotide adjacent to the exon-intron boundary (Figure 1C), we analyzed the mRNA from leukocytes of the individual carrying this mutation by RT-PCR and nucleotide sequencing. No splicing alteration was found (Figures 1D and 1E). We also utilized a minigene assay to demonstrate that the GNB4 c.265A>G mutation did not change the mRNA splicing pattern (Figure S2). Because the CMT individuals in this study were selected from 251 pedigrees affected by CMT after the cases with clear genetic diagnosis were excluded,¹⁵ the frequency of GNB4 mutations in our CMT population was approximately 0.8% (2/251).

The GNB4 c.265A>G (p.Lys89Glu) mutation was found in a 9-year-old girl who had presented with slowly progressive weakness of her distal lower limbs since the age of 5 years. She had started to walk at a normal age. Physical examination at age 9 years revealed high-arched feet, hammer toes, atrophy and weakness of the intrinsic muscles of the feet (score 4−/5 on the Medical Research Council Scale), impaired dorsiflexion of the feet (score 4/5), generalized areflexia, and mildly diminished sensation for all modalities in regions distal to the ankles despite a lack of sensory complaints. Nerve conduction studies demonstrated the presence of demyelinating sensorimotor polyneuropathy with axonal loss. The median and tibial motor NCVs were 20 and 15 m/s, respectively. Neither parent had the CMT phenotype or the GNB4 c.265A>G mutation (Figure 1B).

G proteins are heterotrimeric and are composed of α, β, and γ subunits; they are important to cellular signal transduction. They transmit extracellular signals into cells by acting in concert with different G-protein-coupled receptors (GPCRs); these GPCRs can be activated by many different signal factors, such as neurotransmitters, hormones, and cytokines.¹⁶ In humans, there are 16 Gα subunits, 5 Gβ subunits, 12 Gγ subunits, and a number of splice variants, which thus contribute to a large diversity of G protein heterotrimers.¹⁷ In the inactive state, the guanosine-diphosphate (GDP)-bound Gα is associated with the Gβγ dimer. When a GPCR is activated by a ligand, it catalyzes the G protein to exchange the GDP for guanosine triphosphate (GTP) on Gα, which results in the dissociation of the Gα from the Gβγ.¹⁶ The Gα-GTP and released Gβγ dimer each then transduce signals to a wide range of effectors, including adenylyl cyclases, phospholipases, ion channels, phosphodiesterases, and many more proteins,^17,18 thereby regulating various intracellular signaling pathways and cellular functions. As a result of this wide range of G protein signaling activities, GNB4, which encodes Gβ₄, is likely to be important to a number of cellular signal-transduction systems.

To understand the location of Gβ₄ in peripheral nerves, we analyzed the sural nerves of one neurologically normal control and one individual (III:3 in Figure 1A) with the c.158G>A (p.Gly53Asp) GNB4 mutation by double immunofluorescence staining. Gβ₄ was found to be colocalized with neurofilament heavy chain and S100 (Figure 2 and Table S4), indicating that Gβ₄ is expressed in both axons and Schwann cells. Intriguingly, some myelinated fibers of the normal control had a target-like appearance because of the presence of dense Gβ₄ staining in the axons and the outer rim of the myelin, but not in the compacted myelin sheath (Figure 2). Onion-bulb formations in the sural nerve of individual III:3 showed a “rosette” pattern with Gβ₄ staining in the axons and the cytoplasm of surrounding Schwann cells (Figure 2).

Figure 2.

Gβ₄ Expression in Both Axons and Schwann Cells of Peripheral Nerves

Double immunostaining of Gβ₄ with neurofilament heavy chain (NF-H) (A) or S100 (B) with sural nerve samples from a neurologically normal control and family 1 individual III:3, who harbors the GNB4 c.158G>A (p.Gly53Asp) mutation. The colocalization of Gβ₄ with NF-H and S100 indicates that Gβ₄ is expressed in both axons and Schwann cells. Some myelinated fibers of the normal control have a target-like appearance as a result of dense Gβ₄ staining in the axons and the outer rim of the myelin, but not in the compacted myelin sheath (arrows). Onion-bulb formations in the sural nerve of individual III:3 have a “rosette” pattern with Gβ₄ staining in the axons and the cytoplasm of surrounding Schwann cells (arrowheads). The scale bars represent 10 um.

We next compared Gβ₄ expression in the sural nerve samples from two individuals (III:2 and III:3) with the GNB4 c.158G>A (p.Gly53Asp) mutation, one individual with CMT1A, and one neurologically normal control by immunohistochemistry (IHC). In addition to staining Gβ₄, we also doubly stained the cell nuclei with TDP-43 antibody as a reference (Table S4). Immunostaining revealed that Gβ₄ was abundantly expressed in the axons and Schwann cells of the sural nerve of the normal control (Figure 3A). Some myelinated fibers had a target-like appearance (Figure 3A). Onion-bulb formations in the sural nerve of the CMT1A control were also densely stained with Gβ₄ antibody (Figure 3B). In contrast, Gβ4 staining was significantly weaker in the sural nerves of the two individuals with the GNB4 mutation than in those of the normal control and the individual with CMT1A (Figures 3C and 3D).

Figure 3.

Gβ₄ Immunohistochemistry and Toluidine-Blue Staining of Sural Nerves

Double immunostaining of Gβ4 (red) and TDP-43 (dark blue) in sural nerve samples from a neurologically normal control (A), a disease control with CMT type 1A (CMT1A) (B), and family 1 individuals III:2 (C) and III:3 (D), who harbor the GNB4 c.158G>A (p.Gly53Asp) mutation. Some myelinated fibers in the normal control feature a target-like appearance (A, arrows). The Gβ₄ staining of onion-bulb formations appears weaker in the two subjects with the GNB4 mutation (C and D, arrowheads) than in the individual with CMT1A (B, arrowheads). The sural nerves of the two individuals with the GNB4 mutation have obviously weaker Gβ₄ staining than do those in the normal control and the individual with CMT1A. The cell nuclei were stained with TDP-43 antibody as a reference. The toluidine-blue staining of sural nerve samples of individuals III:2 and III:3 from family 1 reveals a loss of myelinated fibers and a presence of multiple onion-bulb formations (C and D), associated with a demyelinating change. The scale bars represent 10 um.

To test the hypothesis that GNB4 is important for peripheral nerve functioning, we investigated Gβ₄ expression in the sciatic nerves of the rats 3 days after sciatic nerve transection, nerve conditioning with a 20 s transient crush injury, or a sham operation. The nerve tissue 10 mm in length and distal to the injured site was analyzed for Gβ₄ expression by immunoblotting and IHC. Compared with Gβ₄ expression in the sham-operated group, the expression of Gβ₄ appeared to increase in the nerve conditioning group and decrease in the nerve transection group (Figure S3). These findings suggest that GNB4 plays a role in peripheral nerve regeneration and support its importance in peripheral nerve functioning.

To investigate the effect of the GNB4 mutations on protein stability and G-protein-coupled signaling by Gβ₄, we constructed Gβ₄ expression plasmids to allow further in vitro analysis. The coding region of GNB4 from a human GNB4 cDNA clone (Invitrogen; GenBank accession number AF300648.1) was subcloned into pcDNA3.1⁺ (Invitrogen). Mutations c.158G>A (p.Gly53Asp) and c.265A>G (p.Lys89Glu) were introduced by site-directed mutagenesis according to the Quick-Change method (Stratagene) (Table S5). pECFP-C1 was purchased from Clontech. A human PLCβ₂ expression clone, pCMV5-PLCβ₂, was a kind gift from E.M. Ross (Southwestern Medical Center, University of Texas, TX, USA). All constructs were verified by DNA sequencing.

In order to carry out a protein-stability assay, we incubated human embryonic kidney (HEK) 293T cells maintained in Dulbecco’s modified Eagle’s medium (DMEM; Hyclone) containing 10% fetal bovine serum (FBS; Invitrogen) at 37°C and in 5% CO₂ and transfected them with plasmids expressing wild-type, p.Gly53Asp, or p.Lys89Glu forms of Gβ₄ by using calcium-phosphate precipitation.¹⁹ At 48 hr after transfection, the cells were treated with 100 μg/ml cyclohexamide for 0, 2, 4, 8, or 24 hr. Total-protein cell extracts were analyzed by immunoblotting (40 μg of protein was loaded from each cell lysate). The immunoblot signals were used for comparing the relative degradation profiles between the Gβ4 variants and actin. Representative graphs and quantification obtained from six independent experiments are shown in Figure S4. These results show that the degradation rates of the two altered forms of Gβ₄ were not significantly different from that of their wild-type counterpart, suggesting that these alterations did not affect protein stability.

We next investigated the functional impact of the Gβ₄ alterations on GPCR signaling. In this functional study, we used bradykinin to activate the B2-receptor-coupled Gqα signaling pathway in a PLCβ₂-overexpressing COS-7 cell model.^20,21 Activation of PLCβ₂ by Gqα signaling leads to hydrolysis of phosphatidylinositol 4,5-bisphosphate and the formation of the second messengers inositol trisphosphate (IP₃) and diacylglycerol, which is followed by IP₃-induced intracellular Ca²⁺ release. Thus, IP₃ production and cytosolic calcium changes were used to reflect GPCR signaling activity. In brief, COS-7 cells were maintained in DMEM containing 10% FBS at 37°C and in 5% CO₂; LipofectAmine-2000 (Invitrogen) was used for transfecting these cells at a 1:1:1 ratio with plasmids expressing ECFP, expressing PLCβ₂, and expressing wild-type, p.Gly53Asp, or p.Lys89Glu forms of Gβ₄ or empty vectors. Forty-eight hours after transfection, the cells were harvested for subsequent second-messenger analysis. To measure IP₃ production, we subcultured the cells into a 96-well plate at a density of 2 × 10⁴ cells per well with 20 μl PBS. The cells were then stimulated with 100 nM bradykinin for 1 min and lysed with 0.2 N perchloric acid. Intracellular IP₃ production was then measured with a HitHunter IP₃ Fluorescence Polarization Assay Kit (DiscoveRx) as outlined in the manufacturer’s manual. To measure cytosolic calcium changes, we incubated cells on coverslips with 5 μM of the calcium indicator dye fura-2 AM (Molecular Probes) in loading buffer (LB) for 45 min at 37°C and then washed them with LB. Cells coexpressing ECFP with a similar expression level (mean pixel intensity = 526.6 ± 32.8 AU; n = 148) were selected for the subsequent measurements. These cells were stimulated with a pulse of 100 nM bradykinin for 10 s while the fluorescence of fura-2 was simultaneously monitored under an inverted microscope (IX-71, Olympus Optical) coupled with a monochromator (Polychrome IV, TILL Photonics, Gräfelfing, Germany) with a 40× oil-immersion objective (UApo/340 40×/1.35 numeric aperture oil objective, Olympus Optical).²² Use of the 340/380 nm excitation ratio with fura-2 allowed measurements of the intracellular calcium concentration ([Ca²⁺]_i).^23,24

The study revealed that the function of both the p.Gly53Asp and p.Lys89Glu altered forms of Gβ₄ was impaired, that is, they showed reduced bradykinin-induced PLCβ₂ activation, namely an inhibition of IP₃ production and a lower increase in cytosolic calcium. Compared to control cells, PLCβ₂-overexpressing cells showed a bradykinin-induced increase in IP₃ production by 25% (25.7% ± 6.2%, n = 3). When cells coexpressed wild-type Gβ₄ and PLCβ₂, their IP₃ generation was found to be about 3.5-fold higher than that of the PLCβ₂-overexpressing cells (87.9 ± 11.1% versus 25.7% ± 6.2%). However, compared to the control, the p.Gly53Asp and p.Lys89Glu altered forms of Gβ₄ reduced bradykinin-induced IP₃ production by 42.1% ± 2.8% and 45.7% ± 6.0%, respectively (Figure 4A). Similar results were observed when the bradykinin-induced [Ca²⁺]_i increases were monitored. Bradykinin induced a similar increase in [Ca²⁺]_i when the controls, the PLCβ₂-overexpressing cells, and the cells coexpressing PLCβ₂ and wild-type Gβ₄ were compared. However, when the cells coexpressing either PLCβ₂ and p.Gly53Asp Gβ₄ or PLCβ₂ and p.Lys89Glu Gβ₄ were examined, there were no significant changes in [Ca²⁺]_i upon bradykinin stimulation (Figures 4B and 4C). These findings indicate that the mutations in GNB4 cause a defect in the GPCR signaling cascade and impair the activation of PLCβ₂.

Figure 4.

Inhibition of the GPCR Signaling Results from GNB4 Mutations

(A) The p.Gly53Asp and p.Lys89Glu altered forms of Gβ₄ had an inhibition of inositol trisphosphate (IP₃) production upon bradykinin stimulation. COS-7 cells were transfected with pcDNA3.1 (control, first bar on the left), pcDNA3.1-PLCβ₂ alone, or pcDNA3.1-PLCβ₂ and pcDNA3.1-Gβ₄ or altered forms of Gβ₄. The IP₃ generated upon stimulation with 100 nM bradykinin for 1 min was then measured. The IP₃ production level was normalized against the control as 100%. The data shown are the mean ± SEM from three independent transfections.

(B) Absence of changes in intracellular calcium ([Ca²⁺]_i) upon bradykinin stimulation in the cells expressing altered forms of Gβ₄. COS-7 cells were transfected with pcDNA3.1 (control), pcDNA3.1-PLCβ₂ alone, or pcDNA3.1-PLCβ₂ and pcDNA3.1-Gβ₄ or altered forms of Gβ₄ in conjunction with pECFP-C1. Changes in the F₃₄₀/F₃₈₀ ratio of fura-2 upon 100 nM bradykinin stimulation were monitored by a fluorescence microscope. Bradykinin stimulation started at 30 s and lasted for 10 s. The [Ca²⁺]_i changes are shown as the F₃₄₀/F₃₈₀ ratio of fura-2 fluorescence.

(C) Maximal [Ca²⁺]_i changes upon bradykinin stimulation. The maximal [Ca²⁺]_i changes were obtained by subtraction of the averaged fluorescence ratio in the 10 s before stimulation from the maximal fluorescence ratio. The data shown are the mean ± SEM from 26–34 cells for each group (148 cells in total). The cells were obtained from at least three independent transfections.

Utilizing exome sequencing, bioinformatic analysis, and mutational analysis, we have identified two mutations that affect Gβ₄-encoding GNB4 and are present in subjects with CMT. Both GNB4 mutations alter the evolutionarily conserved amino acid residues, are absent from the NHLBI Exome Variant Server, and do not occur in 1,920 ethnic control chromosomes. Immunohistochemical studies revealed that Gβ₄ is abundantly expressed in peripheral nerves, including both axons and Schwann cells. In vitro studies demonstrated that both GNB4 mutations impair bradykinin-induced GPCR signaling. Our findings provide strong genetic and functional evidence supporting that mutations in GNB4 are the cause of this type of CMT.

According to the tissue expression databases of UniGene²⁵ and GeneNote (now retired),²⁶ Gβ₄ is widely expressed in many human tissues, including the spinal cord and peripheral nerves. Gβ₄ has also been found to be expressed in peripheral nerve myelin by proteome analyses.²⁷ We have been able to consistently demonstrate that Gβ₄ is readily detected in the axons and Schwann cells of sural nerves, which indicates that the Gβ₄ signal pathway is essential for peripheral nerve functioning. Decreased Gβ₄ immunostaining in the sural nerves of the two CMT individuals with GNB4 mutations further highlights the important role of Gβ₄ dysfunction in the pathogenesis of this hereditary neuropathy.

Our functional studies have shown that both the p.Gly53Asp and p.Lys89Glu Gβ₄ alterations impair GPCR signaling pathways via a dominant-negative effect. When compared to cells expressing PLCβ₂ alone, cells coexpressing PLCβ₂ and p.Gly53Asp or p.Lys89Glu altered Gβ₄ showed a significantly reduced IP₃ production and an absence of cytosolic calcium change when stimulated with bradykinin. This phenomenon suggests that both altered forms of Gβ₄ have a negative effect on the endogenous wild-type Gβ₄ molecules already present in the cells and that this affects GPCR signaling. Gly53 and Lys89 are highly conserved across the Gβ protein family (including Gβ₁–Gβ₅) across the various kingdoms (including mammals, fungi, nematode, and amoeba). This suggests that they both most likely play critical roles in GPCR signaling.²⁸

The functional defect caused by the two mutations appears to be related to their roles in the specific binding of Gα to the Gβγ dimer. On the basis of the three-dimensional (3D) structure of the native heterotrimer, Giα₁β₁γ₂, the first two WD40 repeats of Gβ₁ interact with the N-terminal helix of Giα₁, and it has been suggested that this interaction confers a certain degree of specificity to Gα-Gβγ complex formation.²⁹ Gβ₄ shares 96% amino acid sequence homology with Gβ₁.³⁰ Gly53 and Lys89 are located within the first and second WD40 repeats, respectively, of the β propeller, and according to high-resolution structure analysis, Gly53 contacts one residue and Lys89 contacts multiple residues within the N-terminal helix of Giα₁ (Figure S5).²⁹ Therefore, it is likely that the interaction between Gβ₄ and Gα, but not the interaction between Gβ₄ and Gγ, is affected in both the p.Gly53Asp and p.Lys89Glu Gβ₄ altered proteins. Although the dimerization of Gβ and Gγ is obligatory in nature,³¹ the interaction between Gγ and the altered forms of Gβ₄ might render a decrease in the functional wild-type Gβ₄-Gγ dimer and thus exert a dominant-negative effect that abolishes relevant GPCR signaling.

All of the individuals carrying the GNB4 c.158G>A (p.Gly53Asp) or c.265A>G (p.Lys89Glu) mutation feature a typical CMT phenotype that involves distal motor and sensory function; however, the subject carrying the c.265A>G (p.Lys89Glu) mutation had an earlier age of onset, indicating that she was more severely affected. The c.265A>G (p.Lys89Glu) mutation seemed to affect Gβ₄ functioning more than the c.158G>A (p.Gly53Asp) mutation in our small subject group. This is supported by the 3D structure of the heterotrimeric Giα₁β₁γ₂. Whereas Gly53 of Gβ₁ only interacts with Leu23 of Giα via van der Waals interactions, Lys89 interacts with multiple residues, namely Giα via hydrogen bond with Ser16, via a salt bridge with Asp20, and via van der Waals interactions with Ile19 and Leu23 (Figure S5).²⁹ Thus, Lys89 might be more important than Gly53 in terms of the association between subunits Gα and Gβ, and this might therefore cause the c.265A>G (p.Lys89Glu) mutation to produce a more severe disease phenotype.

The diagnosis of DI-CMT in family 1 was based on the NCVs and the sural nerve biopsies of subjects III:2 and III:3, who had slow NCVs consistent mainly with demyelinating neuropathy. Because NCV cannot faithfully reflect the underlying pathological changes, the full pathological spectrum of Gβ₄-related neuropathy is yet to be delineated. Given that Gβ₄ is highly expressed in both axons and Schwann cells, mutations in GNB4 might result in myelinopathy and/or axonopathy; this might partially explain the phenotypic heterogeneity of the family 1 individuals, in whom variable NCVs that blurred the division between demyelinating and axonal CMT were documented.

In the family carrying the GNB4 c.158G>A (p.Gly53Asp) mutation, men tended to be more severely affected than women. If this proves to be true in a larger cohort, it will imply that the Gβ₄ signal-transduction pathway might be influenced by gender-preferential genes in the peripheral nerve tissues. One recent study uncovered 159 genes that show male or female preference when expressed in the brain.³² Our study suggests the interesting possibility of identifying genes with sex-preferential expression in peripheral nerves.

The fact that no significant difference in protein stability exists between wild-type and altered forms of Gβ₄ suggests that the pathomechanism of the GNB4 mutations is not due to a haploinsufficiency of the Gβ₄ molecules. This finding is consistent with the possible dominant-negative mechanism. Because the stability of Gβ₄ is not affected by the GNB4 mutations, the reduced Gβ₄ expression in the sural nerves of the individuals with GNB4 mutation suggests that the altered Gβ4 might be toxic. If the Gβ₄-related GPCR signaling function is vital to some cells, the dominant-negative effect of the altered Gβ₄ could be lethal or toxic to them. Alternatively, the altered Gβ₄ might have gained a toxic function that is unrelated to the dominant-negative effect. Such a toxic function could result in a situation where cells expressing the most altered form of Gβ₄ are likely to die first. In such a circumstance, the remaining cells in the sural nerves of these individuals would be those with relatively lower Gβ₄ expression. This speculation is supported by the lower cell and nerve-fiber densities in the sural nerve samples of the two individuals with a GNB4 mutation than in those in the control samples (Figure 3).

We have also screened for mutations in the genes encoding the Gγ subunits that are known to be expressed in the peripheral nerves; these genes included GNG3 (MIM 608941), GNG4 (MIM 604388), GNG7 (MIM 604430), GNG8, and GNG10 (MIM 604389) across 87 molecularly unassigned and unrelated CMT individuals. So far, we have not identified any relevant mutation in them (data not shown). The possibility that other proteins that interact with Gβ₄ might also be pivotal to the etiology of some of these hereditary neuropathies cannot be ruled out.

In conclusion, our findings demonstrate that mutations in GNB4 cause CMT, and this emphasizes the importance of Gβ₄-related GPCR signaling to peripheral nerve functioning in humans.

Web Resources

The URLs for data presented herein are as follows:

• 1000 Genomes Project, http://www.1000genomes.org

• dbSNP, http://www.ncbi.nlm.nih.gov/projects/SNP/

• GATK, http://www.broadinstitute.org/gatk/

• GRCh37 patch 2, ftp://ftp.ensembl.org/pub/release-61/

• NHLBI Exome Variant Server Exome Sequencing Project (ESP), http://evs.gs.washington.edu/EVS/

• Online Mendelian Inheritance in Man (OMIM), http://www.omim.org/

• Panther, http://www.pantherdb.org/

• PolyPhen-2, http://genetics.bwh.harvard.edu/pph2/

• RefSeq, http://www.ncbi.nlm.nih.gov/RefSeq

• SAMtools, http://samtools.sourceforge.net/

• SNPs&GO, http://snps-and-go.biocomp.unibo.it/snps-and-go/

• UniGene, http://www.ncbi.nlm.nih.gov/unigene/