Effect of the achondroplasia mutation on FGFR3 dimerization and FGFR3 structural response to fgf1 and fgf2: A quantitative FRET study in osmotically derived plasma membrane vesicles

Abstract

The G380R mutation in the transmembrane domain of FGFR3 is a germline mutation responsible for most cases of Achondroplasia, a common form of human dwarfism. Here we use quantitative Föster Resonance Energy Transfer (FRET) and osmotically derived plasma membrane vesicles to study the effect of the achondroplasia mutation on the early stages of FGFR3 signaling in response to the ligands fgf1 and fgf2. Using a methodology that allows us to capture structural changes on the cytoplasmic side of the membrane in response to ligand binding to the extracellular domain of FGFR3, we observe no measurable effects of the G380R mutation on FGFR3 ligand-bound dimer configurations. Instead, the most notable effect of the achondroplasia mutation is increased propensity for FGFR3 dimerization in the absence of ligand. This work reveals new information about the molecular events that underlie the achondroplasia phenotype, and highlights differences in FGFR3 activation due to different single amino-acid pathogenic mutations.

1. Introduction

The growth of long bones occurs via endochondral ossification, a process which replaces cartilage with bone in the growth plates of long bones. It is mediated by the chondrocytes in cartilage, which proliferate, undergo hypertrophy, and die. In the final stage, the extracellular matrix secreted by the chondrocytes is invaded by bone. This process is impaired in achondroplasia (ACH), the most common form of human dwarfism [1,2]. The achondroplasia phenotype, which afflicts 1 in ~10,000 live births, is characterized by short stature, megalocephaly, prominent forehead, and diminished muscle tone.

98% of all achondroplasia cases are due to a G380R mutation in the transmembrane domain of fibroblast growth factor receptor (FGFR3) [3]. FGFR3 is known as a negative regulator of long bone development [4–6]. It is highly expressed in growth plate chondrocytes, and plays a critical role in chondrocyte differentiation [2,7]. It signals in response to fgf ligands, in the presence of cell surface heparan sulfate proteoglycans [8,9].

FGFR3 belongs to the large receptor tyrosine kinase (RTK) superfamily. It consists of an N-terminal extracellular domain (comprised of three immunoglobulin-like domains D1–D3), a single pass transmembrane domain, and an intracellular kinase domain. FGFR3 transduces biochemical signals via lateral dimerization in the plasma membrane [10]. FGFR3 dimerization brings the two kinase domains in close proximity, promoting their cross-phosphorylation and activation. The G380R mutation in FGFR3 has been shown to increase FGFR3 phosphorylation in the absence of ligand, as compared to the wild-type [11–14]. The mutation-induced FGFR3 activation suppresses proliferation and alters the hypertrophic differentiation of growth plate chondrocytes, leading to the ACH phenotype [5,15–20].

According to the canonical model of RTK activation, FGFR3 can be expected to be monomeric in the absence of ligand, and to undergo dimerization and activation in response to ligand binding [21]. We have shown, however, that FGFR3 forms dimers even in the absence of ligand, in a concentration dependent manner [22]. It has been hypothesized that the achondroplasia mutation increases dimerization in the absence of ligand [23], but studies in lipid vesicles with the isolated FGFR3 TM domain have not supported this hypothesis [24]. Later, a very modest increase in dimerization was observed due to the achondroplasia mutation in chemically derived plasma membrane vesicles [25]. However, this model system may introduce artifacts due to the presence of DTT and formaldehyde [26,27], and thus one goal of the current study was to re-measure dimerization in osmotically derived plasma membrane vesicles.

The FGFR3 ligands fgf1 and fgf2 are expressed in the proliferating and hypertrophic zones and the resting and proliferating zones of the human growth plate cartilage, respectively [28]. Thus, fgf1 and fgf2 are relevant for long bone development, and are likely involved in the induction of the ACH phenotype. Moreover, their differential localization in the growth plate suggests that they elicit distinct biological responses. We have already shown that wild-type FGFR3 dimers have different structures when bound to fgf1 and fgf2, and are stabilized by distinctly different TM helix–helix contacts in the membrane [22]. The observed difference in structure provides an explanation as to how FGFR3 transmits information about the identity of the ligand across the membrane. The structural response of the G380R mutant to fgf1 and fgf2, however, has not been compared. The second goal of this study therefore was to determine if the G380R FGFR3 dimers adopt different structures in response to fgf1 and fgf2, similarly to the wild-type.

The fgf2-bound FGFR3 dimer structure exhibits the smallest separation between the transmembrane (TM) domains and the highest possible phosphorylation [22]. Investigations of a different pathogenic mutation in the TM domain of FGFR3, A391E, have shown that this mutation emulates the action of fgf2, trapping the FGFR3 dimer in its most active state even in the absence of fgf2 [22]. The A391E mutation, which also occurs in the TM domain of FGFR3, has been linked to Crouzon syndrome with acanthosis nigricans, characterized by premature ossification of the skull [29]. The G380R and A391E mutations occur very close to each other in the FGFR3 sequence, but the phenotype associated with these two mutations is very different. We therefore asked if there are mechanistic differences in the activation of the two mutants that underlie the different phenotypes. Thus, the third goal of this study was to compare the responses of the G380R mutant and the A391E mutant to the ligands fgf1 and fgf2. All experiments described here utilized a FRET-based methodology with capabilities to probe RTK interactions in quantitative terms, as well as report on their response to ligands [27,30].

2. Materials and methods

2.1. Plasmid constructs

The detailed cloning procedure for the plasmid constructs used for this work (see Supplementary Figure 1) has been published [25,31]. The plasmids encode truncated FGFR3 receptors, with the intracellular domains substituted with fluorescent proteins such that FRET can be used for detection of dimerization. In particular, the wild-type plasmid encodes the 22-amino-acid signal peptide of FGFR3 (MGAPACALALCVAVAIVAGASS), the extracellular (EC) and transmembrane (TM) domains of FGFR3, a (GGS)₅ flexible linker, and a fluorescent protein, either YFP or mCherry (a FRET pair). The G380R mutation was engineered in the wild-type plasmid using the QuikChange mutagenesis kit (Agilent Technologies, Inc., USA).

2.2. Cell culture and transfection

Chinese Hamster Ovary (CHO) cells were received from Dr. M. Betenbaugh (Johns Hopkins University). The cells were cultured at 37 °C with 5% CO₂ for 24 h. 2 × 10⁴ cells were then seeded in each well of a six-well plate. Transfection was carried out using Fugene HD transfection reagent (Promega, USA) according to the manufacturer's protocol. The cells in each well were cotransfected with a total of 3 μg DNA. No expression of FGFR3 in CHO was detected via immunostaining or Western blots unless the cells were transfected, demonstrating that CHO cells do not express measurable amounts of endogenous FGFR3.

2.3. Production of mammalian plasma membrane vesicles

CHO cells were rinsed twice with 30% phosphate-buffered saline (pH 7.4) and incubated with 1 mL of chloride salt vesiculation buffer overnight at 37 °C. The vesiculation buffer consisted of 200 mM NaCl, 5 mM KCl, 0.5 mM MgSO₄, 0.75 mM CaCl₂, 100 mM bicine and protease inhibitor cocktail (Complete mini EDTA-free tabs, Roche Applied Science), adjusted to pH of 8.5 [32]. A large number of vesicles were produced after 12 h, and the vesicles were transferred into 4-well Nunc Lab-Tek II chambered coverslips for imaging.

In ligand treatment experiments, vesicles were treated with 5 μg/mL of either fgf1 or fgf2 and were imaged after 1 h incubation at room temperature.

2.4. Fluorescence image acquisition

Vesicles were imaged using a Nikon (Melville, NY) Eclipse confocal laser scanning microscope and a 60× water-immersion objective. All the images were collected and stored at 512 × 512 resolution. Three different scans were performed for each vesicle: 1) excitation at 488 nm, with a 500–530-nm emission filter (donor scan); 2) excitation at 488 nm, with a 565–615-nm emission filter (FRET scan); and 3) excitation at 543 nm, with a 650-nm longpass filter (acceptor scan). Gains of 8.0 and pixel dwell time of 1.68 μs were used for the three scans. To minimize the bleaching of fluorescent proteins, ND8 filters were used during excitation with the 488-nm laser. The imaged vesicles exhibited uniform fluorescence intensities (see Fig. 1), which allowed us to determine the concentrations of the fluorescent proteins in the membrane using purified YFP and mCherry solutions of known concentration as described [33]. The fluorescent protein solutions were prepared as described [34]. They were imaged in the microscope using the same settings used for vesicle imaging, in order to allow direct comparison of solution and vesicle intensities.

Fig. 1.

A vesicle, co-expressing FGFR3 G380R EC + TM-YFP and FGFR3 G380R EC + TM-mCherry, imaged and analyzed in the FRET, acceptor, and donor channels. Images were acquired with a Nikon laser scanning confocal microscope. The images are analyzed with a Matlab code that has been discussed in detail in a previous publication [3]. The intensity across the membrane (open blue symbols) is fit to a Gaussian (solid line) after background correction [30]. The residual from the fit is also shown. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Each vesicle was analyzed using a Matlab® program to determine the fluorescence intensity across the membrane, which was fitted to a Gaussian function; the background intensity was approximated as an error function [30] (see Fig. 2). The donor, acceptor, and FRET intensities for each vesicle were used to determine (i) the donor concentration, (ii) the acceptor concentration, and (iii) the FRET efficiency in each vesicle as described in detail in previous work [27,30]. A brief description of the method is given in Supplementary Information.

Fig. 2.

(A) FRET data for FGFR3 G380R EC + TM construct (magenta circles) are compared to previously published data for the wild-type (blue diamonds) [22]. The FRET efficiency is shown as a function of the acceptor concentration. Each data point represents a single vesicle. (B) Donor versus acceptor concentrations in each vesicle. (C) Dimeric fraction as a function of total receptor concentration, calculated as described in Supplemental Information. The solid lines are the best fits to the data. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

3. Results

3.1. Effect of the ACH mutation on FGFR3 unliganded dimerization

We first assessed the effect of the G380R mutation on FGFR3 dimerization in the absence of ligand. Dimerization experiments were performed using an established quantitative FRET methodology [27, 30], with FGFR3 constructs in which the intracellular domains were substituted with fluorescent proteins (either YFP or mCherry) to allow for FRET-based detection of dimerization. Thus, the constructs used (termed FGFR3_G380R EC + TM-YFP and FGFR3_G380R EC + TM-mCherry) contained the EC domain, the TM domain, and the fluorescent proteins attached to the TM domains via a flexible (GGS)₅ linker (see Supplementary Figure 1). The (GGS)₅ linker has been shown to be unstructured [35], to behave like a random coil, and to not inhibit dimerization [36].

CHO cells were cultured and co-transfected with a total of 3 μg DNA encoding FGFR3_G380R EC + TM-YFP and FGFR3_G380R EC + TM-mCherry. 24 h after transfection, after the receptors were trafficked to the plasma membrane, the cells were incubated with an osmotic stress vesiculation buffer [32]. This resulted in the formation of many plasma membrane derived vesicles [26,27]. FGFR3 dimerization was measured in about 600 such vesicles. Each vesicle was imaged in three channels: a donor channel, a FRET channel, and an acceptor channel (a representative vesicle is shown in Fig. 1). The vesicle image was analyzed using a Matlab® program to obtain the donor concentration, the acceptor concentration, and the FRET efficiency in the vesicle [30].

Fig. 2A shows the measured FRET efficiencies for the G380R EC + TM construct, with each data point representing a single vesicle. For comparison, we also show previously published data for the wild-type in osmotically derived plasma membrane derived vesicles [22]. FRET is higher for the mutant than the wild-type (Fig. 2A), suggesting that the mutant exhibits higher dimerization. Fig. 2B shows the donor concentration versus the acceptor concentration in each vesicle. As discussed in Supplementary Information, these concentrations are required in order to quantify the dimerization propensities.

From these data, we calculate the dimeric fraction versus the total concentration using equation (8) in Supplementary Information. A dimerization model, given by equation (10) in Supplementary Information, is fitted to all single-vesicle data points, yielding the dimerization free energy ΔG and the structural parameter Intrinsic FRET, Ẽ (Table 1). In Fig. 2C, we show the dimerization curve calculated for the optimal parameters, along with the binned experimental dimeric fractions.

Table 1.

Dimer stabilities ΔG and Intrinsic FRET values (Ẽ) for the G380R EC + TM mutant, compared to published results for the wild-type and the A391E mutant [27]. Uncertainties are 67% confidence intervals (standard errors). The average distance between the fluorescent proteins, d, is calculated under the assumption of free fluorescent protein rotation, justified by the fact that the fluorescent proteins are attached via long flexible linkers to the TM domains.

	Kdiss	ΔG (kcal/mol)	Ẽ	d(Å)
G380R FGFR3	160 (134 to 185)	–5.2 (–5.1 to –5.3)	0.59 (0.58 to 0.6)	50.0 (49.6 to 50.4)
G380R FGFR3 + fgf1	100% dimer	100% dimer	0.56 (0.55 to 0.57)	51.0 (50.7 to 51.4)
G380R FGFR3 + fgf2	100% dimer	100% dimer	0.74 (0.73 to 0.75)	44.6 (44.2 to 45.0)
WT FGFR3	3235 (2670 to 3660)	–3.4 (–3.3 to –3.5)	0.52 (0.49 to 0.55)	52.4 (51.4 to 53.5)
WT FGFR3 + fgf1	100% dimer	100% dimer	0.55 (0.54 to 0.56)	51.4 (51.0 to 51.8)
WT FGFR3 + fgf2	100% dimer	100% dimer	0.72 (0.71 to 0.73)	45.4 (45.0 to 45.8)
A391E FGFR3	290 (251 to 360)	–4.8 (–4.7 to –4.9)	0.72 (0.70 to 0.73)	45.5 (45.0 to 46.1)
A391E FGFR3 + fgf1	100% dimer	100% dimer	0.75 (0.74 to 0.76)	44.2 (43.8 to 44.6)
A391E FGFR3 + fgf2	100% dimer	100% dimer	0.78 (0.77 to 0.79)	43.0 (42.6 to 43.4)

The dimerization free energy for the G380R mutant is determined as –5.2 ± 0.1 kcal/mol. On the other hand, the dimerization of the wild-type is –3.4 ± 0.1 kcal/mol. Thus, the G380R mutation stabilizes the unliganded FGFR3 dimer by –1.8 ± 0.2 kcal/mol. This value is slightly higher than the one measured previously for the A391E mutation, which stabilizes the dimer by –1.4 ± 0.2 kcal/mol [36,37].

The Intrinsic FRET value for the G380R mutant is 0.58 ≤ 0.59 ≤ 0.60 and is thus similar to a previously measured value for the wild-type, 0.49 ≤ 0.52 ≤ 0.55 (the uncertainties are 67% confidence intervals calculated from the Matlab fit). Intrinsic FRET is a parameter that does not depend on dimerization propensity, but only on the relative position and orientation of the fluorescent proteins in the dimer. Since the fluorescent proteins are attached via long flexible linkers, in our experiments Intrinsic FRET is expected to depend primarily on the separation between the fluorescent proteins. The distances between the fluorescent proteins can be estimated using eq. 9 in Supplementary Information, assuming free rotation of the fluorescent proteins. We thus calculate the distances between the fluorophores as 50.0 ± 0.4 Å and 52.4 ± 1.1 Å in the G380R and wild-type dimers, respectively (see Table 1). While these values are similar, the Intrinsic FRET for the A391E mutant is distinctly different in the absence of ligand, 0.70 ≤ 0.72 ≤ 0.73, corresponding to fluorophore separation of 45.5 ± 0.6 Å [22].

3.2. Effect of the G380R mutation on FGFR3 response to fgf1 and fgf2

Next, we performed FRET experiments with the G380R mutant in the presence of very high concentrations of the ligands fgf1 and fgf2 (5 μg/mL, ~280 nM), similar to previous experiments with the wild-type [22]. This ligand concentration exceeds the ligand-receptor dissociation constants and the total FGFR concentration by at least two orders of magnitude, such that all receptors are expected to be in the ligand-bound dimeric state [12,38,39]. Indeed, at these concentrations the phosphorylation as a function of ligand concentration reaches a plateau and does not increase further as more ligand is added [12,39]. The data for G380R EC + TM FGFR3 are shown in Fig. 3 in black in the presence of fgf1 and in green in the presence of fgf2.

Fig. 3.

FRET efficiencies, donor concentrations and acceptor concentrations, measured for the G380R EC + TM FGFR3, in the presence of 5 μg/ml fgf1 (black) or fgf2 (olive). For comparison, we also show the G380R EC + TM FGFR3 data in the absence of ligand (open magenta circles). Each data point represents a single vesicle. In each case, 300 to 500 individual plasma membrane derived vesicles were imaged 1 hour after adding the ligand. FRET does not depend on the receptor concentration, as expected in the case of saturating ligand. The Intrinsic FRET for each vesicle was determined according to equation (11) in Supplementary Information. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

We see that in the presence of the two ligands, the FRET efficiency is higher than in the unliganded case, and does not depend on the concentration. This fact indicates that the receptors are 100% dimeric in the presence of the ligands, as expected. Furthermore, we could not fit a monomer-dimer equilibrium model to these data. For 100% dimers, the Intrinsic FRET in each vesicle can be calculated directly by dividing the measured FRET by the acceptor fraction (see equation (12) in Supplementary Information). The results are summarized in Fig. 4A, which shows the histograms of the single-vesicle Intrinsic FRET values calculated using equation (12).

Fig. 4.

Histograms of measured Intrinsic FRET in single vesicles for the FGFR3 G380 EC + TM construct (top panel), compared to published data for the wild-type (middle panel) and the A391E mutant (bottom panel) [22]. Also shown is a graphical representation of the changes in the average distance between the fluorescent proteins, based on changes in Intrinsic FRET (not to scale).

The Intrinsic FRET histograms in the cases of fgf1 and fgf2 are distinctly different (p < 0.01). Thus, there exist two distinct ligand-bound states for FGFR3 carrying the G380R mutation. When fgf1 is bound, the Intrinsic FRET is 0.56 ± 0.01, corresponding to an interfluorophore distance of 51.0 ± 0.4 Å. When fgf2 is bound, the Intrinsic FRET is 0.74 ± 0.01, corresponding to a distance of 44.6 ± 0.4 Å between the fluorescent proteins. These results are similar to the Intrinsic FRET results for the wild-type in the presence of fgf1 and fgf2, respectively (see Table 1). On the other hand, the Intrinsic FRET for the A391E mutant has been previously measured as 0.75 ± 0.01 in the presence of fgf1, very similar to the case of fgf2. In Fig. 4, we show a schematic representation of the observed changes in the separation between the fluorescent proteins.

4. Discussion

4.1. Findings

We have previously shown that FGFR3 has a strong propensity for dimerization in the absence of ligand [22]. Here we show that the dimerization propensity of a FGFR3 EC + TM construct, in which the intracellular domain is substituted with a fluorescent protein, is increased due to the G380 achondroplasia mutation. The effect is significant, –1.8 kcal/mol, supporting the idea that an increase in the dimerization propensity in the absence of ligand is a contributor to the Achondroplasia phenotype. The Intrinsic FRET, on the other hand, is not affected by the G380R mutation.

The Intrinsic FRET is a structural parameter that depends on the distance and the orientation of the fluorescent proteins in the dimer [30]. The Intrinsic FRET is expected to depend primarily on the average separation between the fluorescent proteins in the dimer, because the fluorescent proteins are attached via flexible linkers and the orientation effects are averaged [35]. Furthermore, the separation between the fluorescent proteins should depend on the distance between the points of attachment of the flexible linkers, i.e. on the distance between the C-termini of the TM helices. Thus, a change in the separation of the fluorescent proteins should report on a change in the separation of the TM domain C-termini.

A significant difference in the separation of the TM domain C-termini should be measurable as a difference in Intrinsic FRET. However, subtle structural changes in the TM domain dimer could occur without causing a measurable change in Intrinsic FRET. Thus, the finding that the Intrinsic FRET is similar for the wild-type and the G380R mutant in the absence of ligand does not necessary imply that their high resolution dimer structures are identical. On the other hand, the structures of the wild-type and the G380R mutant dimers are significantly different from the A391E dimer, which exhibits much higher Intrinsic FRET, 0.72.

When fgf1 is bound to the EC + TM dimer, the Intrinsic FRET for both the G380R mutant and the wild-type is 0.55, while it is 0.75 for the A391E mutant [22]. In the presence of fgf2, however, the Intrinsic FRET was 0.73 for all three variants. Previously, this high Intrinsic FRET has been linked to a high propensity for FGFR3 phosphorylation and activation [22]. Thus, the A391E mutation traps the dimer in this state, even when fgf1 is bound or when the dimer is unliganded. The G380R mutation, however, does not cause similar large-scale structural perturbations, and behaves similarly to the wild-type in the FRET experiments in the presence of ligand.

4.2. Comparison with previous studies of the achondroplasia mutation

The effect of the achondroplasia mutation on FGFR3 dimerization, measured here, is higher than the value measured previously in vesicles produced via chemical vesiculation, ≃ 0.5 kcal/mol [25]. These previous measurements were carried out in vesicles produced using a well-established chemical vesiculation method that utilizes DTT and formaldehyde [40–42]. While the exact reason for the difference is unknown, it may be due to the very presence of DTT and formaldehyde in the vesiculation buffer in the earlier experiments.

Previous studies of the achondroplasia mutation have employed traditional biochemical methods. For instance, chemical cross-linking in conjunction with Western blotting has been used to assess the dimerization propensities. In Western blot experiments, we did not observe a significant increase in cross-linking due to the G380R mutation [12]. Thus, the cross-linking efficiencies do not scale with the dimerization propensities measured in the FRET experiments. The disagreement may reflect structural changes in the EC domain due to the G380R mutation. Indeed, cross-linking efficiency depends both on dimerization and structure, as it requires the proximity of suitable amine groups to be cross-linked. Thus, even a modest perturbation in dimer structures may alter cross-linking efficiencies.

4.3. Plasma membrane derived vesicles as a model of the plasma membrane

Recent work has revealed that the plasma membrane of live cells has a complex topology, with a membrane area that is 2 to 3 times larger than the area required to sustain the cell's shape [43,44]. This creates challenges in determining the exact two-dimensional protein concentration in the membrane [45]. The plasma membrane derived vesicles have a simple spherical topology. When they are imaged through their equators, their membrane is perpendicular to the field of view. Thus, measured three dimensional concentrations of fluorophores can be easily converted to two-dimensional concentrations in the membrane [30]. Such concentrations are required in order to obtain binding curves and determine dimerization constants. Thus, the use of plasma membrane derived vesicles allows the characterization of membrane protein interactions in quantitative terms.

A method, which utilizes chemical vesiculation relying on formaldehyde and DTT [40–42], has been used to characterize the lipid and protein organization in membranes [46–48]. Previously, we have used these vesicles to study the interactions of membrane proteins [30,31,49]. As there are reports of chemical modifications of proteins due to the presence of formaldehyde [50], we set to develop an alternative method that uses osmotic vesiculation rather than chemical vesiculation and works with the CHO cell line that is widely used in biomedical research [32]. This osmotic method gives vesicles of the same size, and of very similar lipid composition, as the chemical method [26,32]. Yet, there are distinct differences between the vesicles produced with the two methods [26]. First, we have shown that some membrane proteins incorporate more efficiently in vesicles produced with the osmotic stress method. Second, we have shown that the osmotic stress vesicles contain no cytoplasm, while the chemically produced vesicles retain their cytoplasm.

Some membrane proteins have similar dimerization propensities in the two types of vesicles [32,51]. Here we find, however, that the G380R mutant behaves differently in the two types of vesicles. Since Arg is known to react easily in the presence of formaldehyde [50], it can be hypothesized that the TM domain of mutant G380R FGFR3 is cross-linked to other cellular components, particularly proteins with free amines. The homodimerization of G380R FGFR3 is therefore weaker in the presence of formaldehyde, likely due to cross-linking to other proteins in the plasma membrane.

4.4. Biological implications

The main difference between the wild-type and the mutant, observed in our study, is in the propensity for unliganded dimerization, which is increased due to the G380R mutation. This finding is consistent with previous findings that (i) the achondroplasia mutation increases FGFR3 phosphorylation in the absence of ligand [11,12,23], and (ii) the effect of the mutation on phosphorylation gradually disappears as the ligand is titrated [12,52].

Recently, we proposed a new model of RTK activation, according to which all RTKs are capable of forming dimers in the absence of ligand, but the stability of the unliganded dimers can vary significantly between the different receptors [22]. These unliganded dimers are likely important signaling intermediates, as they have been proposed to “prime” the receptors for efficient response to ligand [53]. The biological significance of RTK unliganded dimers is underscored by the fact that elevated dimeric populations due to enhanced expression are linked to cancer [54–58]. Here we show that the main consequence of the achondroplasia mutation is also enhanced unliganded dimerization, providing further proof for the physiological significance of unliganded FGFR dimers.

We have shown that FGFR3 undergoes a structural change upon ligand binding [22]. We have further shown that the TM domain senses the identity of the ligand and adopts a ligand-specific dimer configuration [22]. The fgf1-specific and fgf2-specific TM dimer configurations are different, and this difference is manifested in differences in the Intrinsic FRET measured when the fluorescent reporters are attached directly to the TM domains via flexible linkers [22].

The ligands fgf1 and fgf2 are physically separated in the growth plate: fgf1 is expressed in areas with hypertrophic chondrocytes, while fgf2 is localized in areas with resting/proliferative cartilage [28]. Thus, these two ligands seem to elicit distinctly different FGFR3-mediated biological responses in the growth plate. The differences in the biological responses are likely due to differences in ligand binding affinities reported earlier [9], as well as due to the structural differences in the fgf2-bound and fgf1-bound FGFR3 dimers, observed here. The current work shows that the ACH mutant, just like the wild-type, adopts two different configurations when bound to fgf1 or fgf2. Thus, both the wild-type and the mutant transmit information about the identity of the ligand across the plasma membrane in a very similar way.

The G380R and the A391E mutations both occur in FGFR3 TM domain and are only 11 amino acids apart in the FGFR3 sequence. The G380R mutation impedes the development of the long bones, which occurs through endochondral ossification [1,5]. The A391E mutation, however, affects intramembranous ossification and leads to premature bone fusion in the coronal area of cranium [29,59]. Thus, the phenotypes associated with these two mutations are very different. The results of our FRET study demonstrate that the structural effects due to the G380R mutation are distinctly different from the effect of the A391E mutation. There are also measurable differences in the effect of the mutation on unliganded dimerization, with dimerization propensities increased by –1.8 ± 0.2 and –1.4 ± 0.2 kcal/mol due to the G380R and A391E mutations, respectively. Thus, the mechanism through which these two mutations affect the early stages of FGFR3 signaling is very different, a novel finding which provides an explanation for the observed phenotypic differences due to the two mutations.