Gain-of-function mutations in TRPV4 cause autosomal dominant brachyolmia

Abstract

The brachyolmias constitute a clinically and genetically heterogeneous group of skeletal dysplasias characterized by a short trunk, scoliosis and mild short stature¹. Here, we identify a locus for an autosomal dominant form of brachyolmia on chromosome 12q24.1–12q24.2. Among the genes in the genetic interval, we selected TRPV4, which encodes a calcium permeable cation channel of the transient receptor potential (TRP) vanilloid family, as a candidate gene because of its cartilage-selective gene expression pattern. In two families with the phenotype, we identified point mutations in TRPV4 that encoded R616Q and V620I substitutions, respectively. Patch clamp studies of transfected HEK cells showed that both mutations resulted in a dramatic gain of function characterized by increased constitutive activity and elevated channel activation by either mechano-stimulation or agonist stimulation by arachidonic acid or the TRPV4-specific agonist 4α-phorbol 12,13-didecanoate (4αPDD). This study thus defines a previously unknown mechanism, activation of a calcium-permeable TRP ion channel, in skeletal dysplasia pathogenesis.

Endochondral ossification requires the orchestrated actions of regulatory and structural molecules that together are responsible for linear bone growth. Early in development, regulatory factors, including transcription factors, receptors and their ligands, pattern the early axial and appendicular skeletal anlage (reviewed in ref. 2). Subsequently, the cartilage model ossifies and the growth plates of endochondral bones are established. Proliferation and expansion of growth plate chondrocytes followed by apoptosis and mineralization require additional regulatory factors, including the PTHrP/PTHR and FGF/FGFR3 regulatory pathways as well as structural components of the cartilage extracellular matrix. Frequently, quantitative or qualitative defects in these molecules lead to skeletal dysplasia phenotypes³. Indeed, many of the diverse molecules that are essential for normal skeletal development have been identified by the characterization of genes associated with skeletal dysplasias.

The brachyolmias are a heterogeneous group of skeletal dysplasias of unknown etiology that primarily affect the spine. At least three types of brachyolmia have been described¹. Type 1 brachyolmia includes the Hobaek and Toledo forms and is inherited in an autosomal recessive fashion^4,5. Both forms of type 1 brachyolmia are characterized by scoliosis, platyspondyly with rectangular and elongated vertebral bodies, overfaced pedicles and irregular, narrow intervertebral spaces. The Toledo form is distinguished by corneal opacities and precocious calcification of the costal cartilage. Type 2 brachyolmia (Maroteaux type) is also an autosomal recessive disorder, primarily distinguished from type 1 by rounded vertebral bodies and less overfaced pedicles^1,6. Some cases are associated with precocious calcification of the falx cerebri. Type 3 brachyolmia is an autosomal dominant form with severe kyphoscoliosis and flattened, irregular cervical vertebrae⁷. Paradoxically, although the limbs are mildly shortened in all three types of brachyolmia, they show minimal epiphyseal and metaphyseal abnormalities on radiographs.

Figure 1 shows the pedigree of a large family (International Skeletal Dysplasia Registry (ISDR) reference number R99–102) with autosomal dominant brachyolmia. The clinical phenotype was characterized by moderately short-trunk short stature, mildly short limbs, mild brachydactyly and no extraskeletal clinical findings. The most characteristic radiographic features were scoliosis with platyspondyly and overfaced pedicles, which were most prominent in the lumbar vertebrae (Fig. 2a,b). There were mild irregularities at the metaphyses of the proximal femora (Fig. 2c) and the hands showed delayed epiphyseal and carpal ossification (Fig. 2d).

Figure 1. Pedigree and haplotypes for family R99–102.

Filled symbols, affected individuals; open symbols, unaffected individuals; gray boxes, probable affected individuals; ^*, DNA sample collected. Microsatellite markers are listed to the left of generation III. The haplotype segregating with the disease is boxed, and the shaded boxed loci were excluded by recombination mapping.

Figure 2. Radiographs of the proband at age 8 years, 3 months.

(a) Anterior-posterior view of the spine showing scoliosis and overfaced pedicles. (b) Lateral view of the spine showing anterior rounding of the vertebrae accompanied by flattening of the vertebral bodies. (c) Anterior-posterior view of the proximal femur and acetabulum showing mild irregularity at the proximal femoral metaphyses. (d) Anterior-posterior view of the hands showing delayed ossification.

Using a two-stage genome scan (see Methods), we identified a locus for the phenotype at chromosome 12q24.1–24.2 with a maximum lod score of 3.04 at a recombination fraction of zero for the marker at locus D12S79. Two-point lod scores for five marker loci from this region are listed in Supplementary Table 1. We constructed haplotypes according to the NCBI physical map (Fig. 1 and identified recombinations that limited the genetic interval transmitted with the disorder.

Because the brachyolmia phenotype is restricted to the skeleton, we hypothesized that the gene involved would be selectively expressed in cartilage. We assessed microarray data from cartilage and noncartilage tissues⁸ to define the expression profiles of genes within the 11.1-Mb interval defined by recombination mapping. Expression analysis for all probe sets within the interval showed that TRPV4 had about tenfold higher cartilage-selectivity than the average of all other genes (Fig. 3). On the basis of these data, we analyzed theTRPV4 coding region and exon–intron boundaries for mutations in the index family and in a second family (R99–457) with a similar autosomal dominant brachyolmia phenotype.

Figure 3. Gene expression analysis throughout the linked region of chromosome 12.

Shown are 230 Affymetrix probe sets representing genes within 12q24.1–24.2 in chromosomal order across the bottom. Data for TRPV4 are marked by the arrow. Cartilage-selective expression (defined in ref. 8) is plotted using the right y axis (black line), and the level of cartilage expression compared to noncartilage tissue (purple bars) or dedifferentiatied chondrocytes (yellow bars) is plotted on the left y axis. Median cartilage selectivity (computed from the linear regression of all points) is represented by the dashed black line. The boundary of the linked region is denoted by the marker loci shown on the x axis.

We found heterozygosity for single-base changes within exon 12 of TRPV4 in both families (Supplementary Fig. 1 online). In family R99–102, a 1847G>A transition was found exclusively among affected family members and predicted a single amino acid substitution, p.R616Q, in the fifth transmembrane region of TRPV4 (ref. ⁹). This change was not present among 107 alleles of ancestry-matched unaffected individuals. Similarly, all five affected individuals studied in family R99–457 carried a 858G>A transition that predicted a V620I substitution. Supplementary Figure 2 online shows a diagram of TRPV4 indicating the locations of the substitutions. A comparison of amino acid sequences among TRPV4 orthologs showed that both Arg616 and Val620 are conserved among the human, rat, mouse, chicken, stickleback and zebrafish proteins (Supplementary Fig. 2 and ref. 9), suggesting that the sequence changes are likely to be pathogenic. In addition, for Arg616, sequence alignment of TRPV4 with the paralogous TRPV1 and TRPV2 and the more highly selective calcium channel proteins TRPV5 and TRPV6 showed complete conservation (Supplementary Fig. 2). Subsequently, sequence analysis of TRPV4 excluded structural gene mutations in two additional autosomal dominant and two sporadic brachyolmia cases, suggesting that autosomal dominant brachyolmia may be genetically heterogeneous.

The TRPV4 channel (reviewed in refs. ^10,11,12,13) gates in response to a large variety of stimuli, including cell swelling, warm temperatures, 4αPDD and endogenous lipid arachidonic acid. Activation by cell swelling and arachidonic acid requires cytochrome P450 (CYP) epoxygenase activity to convert arachidonic acid to epoxyeicosatrienoic acids, which then act as TRPV4 agonists^{11, 12, 13, 14}. The fraction of constitutively active channels can be increased by elevation of intracellular Ca²⁺ or by heat^10,14.

To determine the effect of the brachyolmia-associated mutations on TRPV4 activity, we expressed human TRPV4 in HEK cells. In comparison to wild-type TRPV4 (Fig. 4a,b), the R616Q channel yielded a much larger constitutive current before agonist application (Fig. 4c,d). The shape of the IV curve and the reversal potentials were not changed (Fig. 4b,d). Activation with the TRPV4-specific agonist 4αPDD also resulted in significantly increased currents. The data show significantly increased constitutively activated current in the mutant (Fig. 4e). Similar data were obtained for the mutation encoding V620I, with a significantly increased constitutive current at +100 mV, albeit somewhat less increased than for the R616Q mutant, but a markedly increased maximal 4αPDD-activated current (Supplementary Fig. 3). Cell surface biotinylation assays showed no significant differences in the level of TRPV4 at the plasma membrane between the wild type and mutants (Supplementary Fig. 4), excluding the possibility that the increased channel activity was due to altered channel trafficking. In addition, mechano-activation by hypotonic cell swelling (reduction of extracellular osmolarity from 320 to 200 mOsm) and channel activation by application of 10 μM arachidonic acid^15,16 markedly increased the activity of both mutant channels (Fig. 5). Thus both the R616Q and V620I substitutions confer a ‘gain-of-function’ phenotype by significantly increasing the fraction of constitutively open channels and by potentiated agonist activation.

Figure 4. Expression of human TRPV4 and the R616Q mutant in HEK293 cells.

(a) Time course obtained from voltage ramps measured at −100 mV (black boxes) and +100 mV (red circles). 1 μM 4αPDD was applied at the time indicated. HEK cells were transfected with human wild-type TRPV4. The data reflect typical activation of TRPV4. Before application of the specific agonist 4αPDD, currents measured at −100 mV and +100 mV reflect constitutively open channels. After application of 4αPDD, more channels were activated, causing an increase in current at both potentials. (b) Current voltage relationships obtined from the time course shown in a and measured at the points indicated by the filled circles. Note that the IV curves reverse at slightly positive potentials and show a modest inward and outward rectification as typical for TRPV4. No such currents were observed in nontransfected or mock transfected cells (that is, those transfected with the GFP plasmid without the TRPV4 construct (data not shown)). (c) Same experiment as in a; however, cells were transfected with the R616Q mutant of TRPV4. Note the increased current at both potentials before activation with 4αPDD. (d) IV curves from the experiment shown in c. (e) Summarized data from the constitutive current and the 4αPDD activated current measured at −100 mV and +100 mV from wild-type TRPV4 and the mutant channel (mean ± s.e.m., ^*P < 0.05, n = 14).

Figure 5. Expression of TRPV4 in HEK 293 cells.

Same experiments as in Figure 4 and Supplementary Figure 3, except that TRPV4 was stimulated by hypotonic cell swelling (200 mOsm instead of 320 mOsm) and application of 10 μM arachidonic acid (AA). (a) Current voltage relationships (IV) for wild type, R616Q and V620I upon activation by cell swelling. Note the unchanged shape of the IV curves obtained from voltage ramps of the controls before the stimulus (black traces represent the constitutive activity; red traces reflect activity during stimulation). (b) Current activation by 10 mM arachidonic acid measured from voltage ramps at −100 mV and +100 mV. (c) Pooled data from the same series of experiments as shown in a for wild-type TRPV4 (n = 6), R616Q (n = 7) and V620I (n = 7). Averaged values before stimulation (white bars, averaged and s.e.m.) and maximal values during stimulation by hypotonic cell swelling (gray bars). Note that all data on mutants are significantly higher as compared with that of wild type (P < 0.05). Current values are shown for +100 mV and −100 mV. (d) Pooled data from same series of experiments as shown in b for wild-type TRPV4 (n = 6), R616Q (n = 8) and V620I (n = 8). Data are averaged values and s.e.m. Note that values for mutants are significantly higher (P < 0.05) compared with that for wild type, for both constitutively active channels and after stimulation with 200 mOsm or arachidonic acid.

These results demonstrate that the mutations lead to the brachyolmia phenotype by increasing constitutive and stimulus-dependent TRPV4 activity¹³. The increase in constitutive open channel activity is similar to that resulting from in vitro substitutions including Y591A in the TRPV4 TM4 domain¹⁷, F707A in the sixth transmembrane helix and E797A or E797K in the TRPV4 C terminus¹⁸. The consequences of the Y591A substitution and a second substitution (R594Q), which is similar to the brachyolmia mutations, were suggested to reveal involvement in the general gating mechanism of TRPV4 (ref. ¹⁷). The data presented here are thus compatible with the hypothesis that Arg616 and Val620 are structurally important residues that modulate gating function, causing an increased Ca²⁺influx both under resting and stimulated conditions.

Recently, it has been shown that overexpression of mouse TRPV4 in zebrafish caused marked shortening and curvature of the axial skeleton¹⁹. These developmental abnormalities might be caused by uncontrolled accumulation and activation of TRPV4 at the cell surface, which in turn compromises normal calcium homeostasis. A newly identified TRPV4 binding protein, OS-9, diminishes this effect by reducing the TRPV4 traffic to the plasma membrane. Thus, overexpression of normal TRPV4 may have similar consequences as the activating mutations found in individuals with brachyolmia. Notably, knockout of Trpv4 in the mouse resulted in mice of normal size and growth parameters²⁰.

The data presented here contribute to increasing evidence that TRPV4 is a key regulatory molecule in the growth plate²¹. Although the specific mechanism by which TRPV4 activation causes brachyolmia is unknown, there are several possibilities. First, pharmacological activation of TRPV4 induces Sox9 expression in ATDC5 cells²¹. Sox9 has an essential role in chondrocyte differentiation in the growth plate, and both increased and decreased Sox9 can cause defects in endochondral ossification. Haploinsufficiency for Sox9 in humans causes campomelic dysplasia²². In contrast, a mutation in Prkg2 that results in the persistence of Sox9 in growth plate chondrocytes causes growth retardation in the Komeda miniature rat Ishikawa²³. Thus, Sox9 is tightly titrated, and increased Sox9 due to TRPV4 activation could affect growth plate chondrocyte differentiation.

Second, TRPV4 has a functional role in keratinocyte cell volume regulation²⁴. The tenfold volumetric increase in hypertrophic chondrocytes contributes up to 80% of longitudinal endochondral bone growth²⁵. Perhaps activation of TRPV4 in chondrocytes decreases the extent of chondrocyte hypertrophy during endochondral ossification.

Finally, TRPP1 and TRPP2 (also known as PKD1 and PKD2) have been localized in cilia-like structures of osteoblasts and osteocytes, and PKD1-null mice develop articular cartilage and growth plate defects^26,27. TRPV4 associates with TRPP2 (ref. ²⁸) and, together with TRPV2, has been implicated in the regulation of cilia activity in the bronchioles¹¹. This suggests that TRPV4 activation could possibly lead to altered ciliary function.

This study describes activating mutations in TRPV4 that result in an autosomal dominant form of brachyolmia. The mutations increase constitutive TRPV4 activity, and identify an essential role for cation channel activity in the growth plate. Abnormal skeletogenesis resulting from alterations in calcium homeostasis is a previously unknown mechanism in the etiology of skeletal dysplasias, and the specific skeletal abnormalities observed in brachyolmia reveal a particularly important function for this process in the growth and stability of the spine. The nature of the mutations also suggests that modulation of TRPV4 activity using calcium channel inhibitors could represent an avenue for treatment.

METHODS

Subjects

Under an IRB protocol approved by Cedars-Sinai, we collected blood and isolated DNA from cases referred through the ISDR. Informed consent was obtained for all family members. We determined the clinical status of each family member by clinical and radiographic criteria.

Genome scan

We carried out an initial microsatellite genome scan using DNA from 7 family members and 384 microsatellite markers from the ABI Prism Linkage Mapping Set Version 2 (Applied Biosystems). We identified seven chromosomal regions with lod scores of >1.0, and we analyzed these with additional microsatellite markers identified through the Genome Database using DNA from 15 family members. Amplified PCR products were resolved by gel electrophoresis on an ABI Prism 377 DNA sequencer and output genotypes were processed using ABI Prism Genescan version 3.1.2 software for allele size determination. We then imported genotypes into the linkage analysis program Mendel 4.0 and carried out two-point linkage analysis to calculate lod scores²⁹. Brachyolmia was modeled as an autosomal dominant, fully penetrant disease with an allele frequency of 0.0001. The allele frequencies for each marker were set at 1/N, where N was the number of alleles observed in the pedigree.

Mutation detection

PCR fragments covering exons 2 to 16 of the TRPV4 coding sequence were amplified from 100 ng of genomic DNA using AmpliTAQ polymerase and standard PCR protocols in the Gene Amp PCR System 9700 (Applied Biosystems). The primers used (at an annealing temperature of 60 °C) are listed in Supplementary Table 2. Exon 1 was amplified using the Qiagen HotStart Hifidelity kit and buffer Q according to the manufacturer‘s instructions, using a touchdown PCR reaction with the annealing temperature starting at 65 °C and decreasing 1 °C every 3 cycles until reaching 62 °C, at which the reaction was cycled a further 25 times. Sequences of amplified PCR products were determined using the ABI Prism Big-Dye Terminator Cycle Sequencing Kit, version 3.1 (Applied Biosystems), and analyzed on an ABI Prism 377 DNA Sequencer (Applied Biosystems). Sequence data were processed using ABI software and analyzed using Sequencher (Genecodes). We compared the DNA sequences against the cDNA reference sequence for TRPV4, with nucleotide numbering starting from the A of the ATG initiation codon of the reference sequence. Mutations and polymorphisms were confirmed in two separate PCR products using bidirectional sequencing.

Allele frequency analysis

The sequence change detected in TRPV4 in family R99–102 was within an Hpy188I restriction endonuclease cleavage site. The change altered the recognition sequence from [TCC▼GA]TTC to [TCCAA]TTC, ablating the Hpy188I site in the putative mutant allele. The target sequence was amplified with primers for exon 12 using DNA from more than 50 normal Ashkenazi Jewish controls and cleaved with Hpy188I (New England Biolabs). PCR products derived from an unaffected subject only showed Hpy188I cleavage products of 180 bp and 117 bp. PCR products derived from an affected individual showed one uncleaved PCR product of 297 bp in addition to the two cleavage products derived from the normal allele.

Expression of TRPV4 and the R616Q and V620I variants in HEK293 cells

We grew human embryonic kidney cells HEK293 in DMEM containing 10% (v/v) human serum, 2 mM L-glutamine, 2 U/ml penicillin and 2 mg/ml streptomycin at 37 °C in a humidity-controlled incubator with 10% CO₂. For electrophysiology, the HEK293 cells were transiently transfected with the human TRPV4 vector and cloned as a BamHI fragment into the BcII-site of the pCAGGS/IRES-GFP vector, which allows detection of transfected cells based on their green fluorescence when illuminated at 475 nm. For transfection, we used L-alanyl-L-glutamine (Merck Eurolab) and GeneJuice Transfection Reagent (Novagen). Green fluorescence–negative cells from the same batch were used as controls. We introduced mutations in TRPV4 using the QuickChange Site-Directed Mutagenesis Kit (Stratagene). The nucleotide sequences of the mutants were verified by sequence analysis of the corresponding cDNAs.

Cell surface biotinylation and immunodetection

HEK293 cells were transfected with equal amounts of pCAGGS or IRES-GFP vectors carrying wild-type or mutant TRPV4 proteins (similar transfection efficiency) and, after overnight growth, they were washed twice with PBS and incubated with 1 mM EZ-link Sulfo-NHS-SS-Biotin (Pierce) for 30 min at 4 °C. To quench any nonreacting biotin reagents, we washed cells once with ice-cold 50 mM Tris (pH 8.0) and then twice with PBS. After whole-cell protein extractions in PBS supplemented with 1.5% Triton X-100, 1 mM PMSF and protease inhibitor cocktail (10 μg/ml leupeptin and antipain, 2 μg/ml chymostatin and pepstatin), biotinylated proteins were precipitated with immobilized streptavidin, using the manufacturer‘s protocol (Pierce), and analyzed by SDS-PAGE and immunodetection with antibodies to TRPV4 (ref. ¹⁵), pan-cadherin (Abcam) and β-actin (Sigma).

Electrophysiology

We measured whole-cell membrane currents with an EPC-9 (HEKA Elektronik, sampling rate, 1 ms; 8-Pole Bessel filter 3 kHz) using ruptured patches. Patch electrodes had a DC resistance between 2 and 4 MΩ when filled with intracellular solution. An Ag-AgCl wire was used as a reference electrode. We monitored capacitance and access resistance continuously. Between 50% and 70% of the series resistance was electronically compensated to minimize voltage errors. We applied a ramp protocol, consisting of a voltage step from the holding potential of 0 mV to −100 mV followed by a 400 ms linear ramp to +100 mV. This protocol was repeated every 2 s. Cell membrane capacitance values were used to calculate current densities. For electrophysiological measurements, the standard extracellular solution contained 150 mM NaCl, 6 mM KCl, 1 mM MgCl₂, 1.5 mM CaCl₂, 10 mM glucose and 10 mM HEPES, buffered at pH 7.4 with NaOH. The pipette solution was composed of 20 mM CsCl, 100 mM Asp, 1 mM MgCl₂, 10 mM HEPES, 4 mM Na₂ATP, 10 mM BAPTA and 2.93 mM CaCl₂. The free Ca²⁺ concentration of this solution is 50 nM. The osmolality of this solution, measured using a vapor-pressure osmometer (Wescor 5500, Schlag), was 320 ± 5 mOsm. The non–PKC-activating phorbol ester, 4α-phorbol 12,13-didecanoate (4αPDD, Alexis Biochemicals), was applied at a 1 μM concentration from a 10 mM stock solution in ethanol. Arachidonic acid was purchased from Sigma (applied from a 10 mM DMSO stock to a final concentration of 10 μM). For hypotonic cell swelling, we superfused cells with a solution containing 95 mM NaCl, 6 mM CsCl, 1 mM MgCl₂, 1.5 mM CaCl₂, 10 mM glucose and 10 mM HEPES, buffered to pH 7.4 with NaOH. Isoosmolarity was achieved by adding mannitol (125 mM) to this solution to reach 320 mOsm. Hypotonic cell swelling was induced by removing mannitol to lower the extracellular osmolarity to 200 mOsm (see ref. 30 for details). All measurements were carried out at room temperature, 22–25 °C.

Statistics

We tested significance between individual groups using the unpaired Student‘s t-test (P < 0.05). Data are expressed as the mean ± s.e.m.