BRIL/IFITM5 is a membrane protein present almost exclusively in osteoblasts, which is believed to adopt a type III (N-out/C-out) topology. Mutations in IFITM5 cause OI type V, but the characteristics of the mutant protein and the mechanism involved are still unknown. The purpose of the current study was to re-assess the topology, localization, and biochemical properties of BRIL and compare it to the OI type V mutant in MC3T3 osteoblasts. Immunofluorescence labeling was performed with antibodies directed against BRIL N- or C-terminus. In intact cells, BRIL labeling was conspicuously detected at the plasma membrane only with the anti-C antibody. Detection of BRIL N-terminus was only possible after cell permeabilization, revealing both plasma membrane and Golgi labeling. Trypsinization of live cells expressing BRIL only cleaved off the C-terminus, confirming that it is a type II protein and that its N-terminus is intracellular. A truncated form of BRIL lacking the last 18 residues did not appear to affect localization, whereas mutation of a single leucine to arginine within the transmembrane segment abolished plasma membrane targeting. BRIL is first targeted to the endoplasmic reticulum as the entry point to the secretory pathway and rapidly traffics to the Golgi via a COPII-dependent pathway. BRIL was found to be palmitoylated and two conserved cysteine residues (C52 and C53) were critical for targeting to the plasma membrane. The OI type V mutant BRIL, having a five residue extension (MALEP) at its N-terminus, presented with exactly the same topological and biochemical characteristics as wild type BRIL. In contrast, the S42 > L mutant BRIL was trapped intracellularly in the Golgi. BRIL proteins and transcripts were equally detected in bone from a patient with OI type V, suggesting that the cause of the disease is a gain of function mediated by a faulty intracellular activity of the mutant BRIL. © 2014 American Society for Bone and Mineral Research.
Bone-restricted IFITM-like (BRIL), also called IFITM5, is part of an evolutionarily conserved family of so-called small interferon inducible transmembrane (IFITM) proteins, for which there are at least 3 closely related members (IFITM1, 2, 3).[1-3] The mouse has 2 other members (IFITM6 and 7), all of which fall under the larger “dispanin” family of proteins that are predicted to possess 2 transmembrane passages. Our group discovered BRIL using a high throughput screen for cDNAs encoding secreted and membrane proteins in osteoblastic cells. The relationship of BRIL to the other members, however, is based on structural rather than functional considerations. For instance, IFITM5, IFITM1, IFITM2, and IFITM3 are all clustered on chromosome 11 (in humans); they possess a similar gene architecture (2 coding exons), and they are proposed to encode transmembrane proteins that have a similar predicted topology (type III; N-out/C-out) and 2 transmembrane domains.
BRIL, however, is distinct from the other IFITMs in several aspects. Unlike IFITM1-3, BRIL is not responsive to interferons at the transcriptional level. Rather, we have shown that it is regulated by the hedgehog signaling pathway (GLI2) in conjunction with Sp1/Sp3/Sp7, and further controlled by CpG methylation of the promoter region. BRIL is expressed almost exclusively in osteoblasts,[1, 8, 9] in contrast to the ubiquitous nature of IFITMs. Lastly, BRIL localizes predominantly to the plasma membrane, whereas IFITMs are targeted mostly to the endosomal compartment.[10-12]
Functionally, BRIL was ascribed a role as a positive modulator of mineralization in vitro. IFITM1, 2 and 3 have a prominent role in the inhibition in cell entry and infection by various viruses, a function that is dependent on palmitoylation of conserved cysteine residues.[14, 15] Evaluation of the role of BRIL by genetic ablation in mice has not clearly confirmed an in vivo role in mineralization.[16, 17] However, the contribution of BRIL as a major determinant of skeletal integrity is clearly demonstrated by the discovery that a mutation in IFITM5 causes OI type V.[18, 19] OI type V is inherited in an autosomal dominant fashion, and is characterized by distinct clinical features that are not usually observed in any other OI type, such as hyperplastic callus formation, interosseous membrane ossification, and a meshlike lamellation pattern. All cases reported to date have a single recurrent mutation (c.-14C > T) in the 5′UTR of IFITM5.[18, 19, 21-25] The base change creates a novel in-frame ATG upstream of the natural coding start of BRIL, resulting in the addition of 5 residues (MALEP) at its N-terminus. Only one other distinct point mutation in the coding region of BRIL, converting serine 40 into a leucine (S40 > L), has recently been reported to cause severe OI.[26-28] The mechanism by which these BRIL mutants contribute to disease is still unclear but likely involves a gain of detrimental function. In one study, the S40 > L BRIL mutation was associated with reduced PEDF expression and function, as reflected by the increased unmineralized osteoid thickness at the histological level.
We originally reported that, in HEK293 cells, FLAG-tagged BRIL is a type III transmembrane protein, and thus has both extremities extruding into the extracellular milieu. This topology was also previously demonstrated for IFITM1 and 3.[15, 29] However, more recent investigations of the structural properties of IFITM1 and IFITM3 have challenged this long assumed model.[11, 14, 30-33] These new data support a type II conformation (N-in/C-out) and that the predicted first transmembrane passage could be “intramembranous.” This new model prompted us to re-evaluate the topology and biochemistry of BRIL. We demonstrate that the bulk of BRIL resides in a type II configuration at the plasma membrane and Golgi of osteoblasts. Palmitoylation is also necessary for the stability and proper targeting of BRIL to the cell surface. We also present comparative data on the localization of the two OI BRIL mutants (MALEP and S40 > L).
Materials and Methods
Cell cultures and treatment
UMR106 and MC3T3-E1 (subclone #4, hereafter designated MC3T3) were obtained from ATCC and used up to passage 20. UMR106 were grown in DMEM and MC3T3 in αMEM, all supplemented with 10% FBS (Life Technologies). For differentiation of MC3T3, cells were seeded at 100,000 cells per well in 6-well plates (Sarstedt) and grown for 72h until confluency. From this point on, which was considered day 0, cells were fed αMEM + 10% FBS supplemented with 50μg/ml ascorbic acid (Sigma) and 3mM beta-glycerophosphate (Sigma). Medium was changed every 2 or 3 days. To assess palmitoylation, cells were pre-incubated for 18h with media containing 10% charcoal stripped (CS)-FBS. Metabolic labelling was performed for 3h in CS-FBS media supplemented with 0.25mCi/ml [9,10-3H(N)]-palmitic acid (PerkinElmer).
The antibodies against BRIL were raised in rabbits by immunization with the following mouse peptides: D2TSYPREDPRAPSS16C (as described previously); CGS114KLAKDSAAFFSTKFD129; MALEPMDTGGC. These antibodies will be referred to as anti-N, anti-C, and anti-MALEP, respectively. The anti-MALEP peptide included the first 3 residues of the wild type BRIL (bolded) to improve immunogenicity. Exogenous glycine residues (underlined) were added to act as neutral spacer. C-terminal cysteine was conjugated to activated keyhole limpet hemocyanin with maleimide. The peptide/carrier complex was mixed with complete Freund's adjuvant and injected into rabbits according to standard protocols (EZBiolab, Carmel, IN, USA). Antibodies were affinity-purified using the same peptides coupled to SulfoLink Immobilization Resin (Pierce). Anti-FLAG M2 was from Sigma, Alexa Fluor-488 phalloidin, Alexa Fluor (448 and 594) coupled goat-anti-rabbit or donkey-anti-mouse secondary, anti-GAPDH (clone 6C5) antibodies were from Life Technologies, and the anti-58K Golgi protein (clone 58K-9) was from Abcam.
Plasmids and transfection
Plasmids were constructed in the CMV-promoter driven backbone of pcDNA. Except when the natural 5′UTR of Ifitm5 was evaluated (Figs. 7 and 8), translation was in the context of an optimal Kozak consensus sequence (GCCACC) preceding the natural ATG of BRIL. A plasmid encoding GFP was used as a negative control for transfections. Point mutations and deletions were introduced in BRIL by whole plasmid amplification using Phusion DNA polymerase (New England Biolabs, Ipswich, MA, USA) with phosphorylated primers covering the targeted codons. The list of oligonucleotides used for all mutagenesis and cloning is provided in supplemental Table 1. PCR conditions included an initial denaturing at 98°C for 2 minutes and 26 cycles at 98°C for 10 seconds, 58°C for 25 seconds, 72°C for 80 seconds. The linear plasmid products were purified from agarose gels on MinElute columns (QIAGEN) and religated. Plasmids were prepared using Midiprep Qiafilter Kit (QIAGEN). The plasmid expressing osteocrin (OSTN) having an N-terminal FLAG epitope located just after its cleavable signal peptide was described previously.[34, 35] The FLAG-OSTN-BRIL construct was made by subcloning a PCR product corresponding to the mouse BRIL entire coding sequence (from ATG to STOP), into the blunted HindIII site of the FLAG-OSTN plasmid. The mouse Sar1 cDNA was amplified by RT-PCR on RNA from MC3T3 cells and cloned into pCDNA. The H79G mutation was introduced in SAR1 as described above with primers H79G-F and H79G-R. The identity of all constructs was confirmed by Sanger sequencing on an Applied Biosystem 3730xl DNA Analyzer through the McGill University and Genome Quebec Innovation Centre. For transient transfection experiments, cells were seeded at 190,000 cells per well in 6-well plates. The next day, medium was changed and cells were transfected with plasmid DNA (1μg/well) using XtremeGENE 9 (Roche) at a 1:6 ratio. Cells were collected typically 24 h thereafter.
Immunofluorescence (IF) and western blotting
Cells were processed for IF and western blotting essentially as described.[35-37] Briefly for IF, cells grown directly in 6-well culture plates were washed with phosphate buffered saline (PBS), fixed for 10 min with paraformaldehyde (3% w/v in PBS), and left intact or permeabilized for 5 min with either Triton X-100 (0.1% v/v PBS) or digitonin (0.005% w/v in 125 mM sucrose). After washing with PBS, cells were blocked for 1 h with 2% skim milk with 0.1% BSA in PBS. Primary and secondary antibodies were diluted in blocking solution and incubated sequentially for 1 h at room temperature. At the end of each incubation, cells were washed three times for 5 min with PBS, and mounted with ProlongGold Antifade reagent with DAPI (Life Technologies). Cells were imaged by epifluorescence microscopy on a Leica DMRB equipped with an Olympus DP70 digital camera. All images within each experiment were captured at the same exposure. The primary antibodies were used at the following dilutions: anti-FLAG (1:2000), anti-N (1:5000), anti-C (1:1000), anti-MALEP (1:500), anti-58K Golgi (1:200), anti-GAPDH (1:1000). Alexa-fluor coupled secondary antibodies were used at 1:1000 and Alexa488-phalloidin at 1:100.
For western blotting, cells were washed with PBS and collected by directly scraping in NP-40 cell lysis buffer (50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1 mM EDTA, 1% (v/v) NP-40) containing proteases inhibitor cocktail (Sigma). For assessing the presence of BRIL at the plasma membrane, live cells were detached for 5 min at 37°C with Trypsin-EDTA (0.05%-0.5 mM) (Life Technologies, Inc., Grand Island, NY, USA), mixed with complete media with 10% FBS to inactivate trypsin, and centrifuged at 250g for 2 min at 4°C. After washing twice with PBS, the cell pellet was resuspended with the NP-40 cell lysis buffer. Total cell extracts were incubated on ice for 10 min and the insoluble material and nuclei were centrifuged at 16000g for 10 min at 4°C. The supernatant was mixed with 4X Laemmli buffer with 2-mercaptoethanol, boiled for 2 min, and separated on 16% SDS-PAGE. After transfer to 0.45 μm nitrocellulose (Protran BA85), equal amount of protein loaded across lanes was systematically verified by Ponceau S red staining. Membranes were blotted in 5% skim milk in PBS-tween (0.05%) with primary antibodies (anti-N 1:5000; anti-C 1:1500) overnight at 4°C, and then for 1 h at room temperature with an HRP-coupled goat-anti-rabbit at 1:30000 (Amersham). Detection was performed with the chemiluminescent reagent ECL Prime (Amersham).
Immunoprecipitation and fluorography
After metabolic labeling with 3H-palmitic acid, NP-40-soluble cell extracts from GFP or BRIL expressing cells were prepared as described for western blotting. An aliquot was kept as the input. Extracts were incubated overnight at 4°C with 8 μg of affinity purified anti-N antibody with gentle mixing. Twenty μl of pre-equilibrated protein G-Plus agarose 50% beads slurry (Santa Cruz) were added and incubated with mixing for 2 h at 4°C. Beads were centrifuged for 2 min at 200g at 4°C, and the supernatant recovered. The bead pellets were washed 3 times with cell lysis buffer, mixed with 1X Laemmli sample buffer with 2-mercaptoethanol, and boiled for 2 min. Equal proportion of samples (input), unbound (free), and bound were separated on duplicate 16% SDS-PAGE for western blotting and fluorography. For fluorography, the gel was soaked successively for 1 h each at room temperature in 50% methanol/10% acetic acid, En3hance (Perkin Elmer, Waltham, MA, USA), and in cold 1% glycerol (v/v in distilled H2O). After drying for 2h at 50°C, gel were exposed to autoradiographic films at −80°C for up to 4 days.
Human IFITM5 expression in OI type V
Bone fragments were obtained after surgical procedures and snap frozen in liquid nitrogen. Samples were from a 9-year old girl with OI type V who had the recurrent c.-14C > T mutation in IFITM5 and a 8-year old boy with OI type IV caused by a c.2596G > A mutation in COL1A1, respectively. The study was approved by the Institutional Review Board of McGill University and informed parental consent was provided. Bone chips were crushed to a powder in liquid nitrogen with a mortar and pestle. Total RNA and proteins were extracted with Trizol, as recommended by the manufacturer. RNA was reverse transcribed with the High Capacity cDNA synthesis kit (Applied Biosystems, Inc., Foster City, CA, USA). PCR reactions were set up in 25μl with 5μl of cDNA (corresponding to 0.1 μg of input RNA) with primers F1 and R1 located in the 5′UTR and exon 2, respectively (see Fig. 8). The 438 bp product corresponded to nucleotides 2 to 439 of the human IFITM5 cDNA (NM_001025295). The PCR reaction was separated on a 1.5% agarose gel, stained with ethidium bromide, and purified on a MinElute column (QIAGEN). Sanger sequencing was performed with an internal reverse primer. Proteins were separated on 16% SDS-PAGE and analyzed by western blotting with the anti-C antibody.
BRIL is predominantly a type II transmembrane protein
The topology of non-tagged BRIL in osteoblast cells was systematically assessed using a novel antibody raised against the C-terminal end (Fig. 1A) covering residues 114-129. The antibody reacting against the N-terminal residues 2-15 was used in parallel. On intact cells, the anti-N antibody did not systematically yield significant labeling (Fig. 1B). Occasionally, however, a punctate staining restricted to discrete focal areas at the periphery of cells was visible when using either the anti-N BRIL or anti-GAPDH antibodies (Supplemental Fig. 1). Incubation of nonpermeabilized cells with phalloidin, a marker of cytoskeletal actin fibers,[38, 39] perfectly co-localized with these BRIL- and GAPDH-positive structures, suggesting that this result was due to partial access to the intracellular compartment at the cell edges (Supplemental Fig. 1). In sharp contrast, the anti-C antibody detected BRIL uniformly over the entire cell surface and on peripheral extensions (Fig. 1B).
Immunolabeling was also carried out in cells permeabilized with digitonin, which creates pores in the plasma membrane but not in intracellular organelles (rER/Golgi). Under those conditions, the anti-N antibody yielded conspicuous BRIL labeling on the entire cell surface, including cell processes (Fig. 1B). Intracellular labeling was also detected over crescent structures juxtaposed to the nucleus, reminiscent of Golgi apparatus (Fig. 1B arrows). Co-localization with a marker of Golgi (Golgi-58K),[40, 41] testified that BRIL localizes to the Golgi apparatus (Fig. 1B arrows). The signal detected with the anti-C antibody after digitonin permeabilization was indistinguishable from that observed in intact cells except for no Golgi labeling (Fig. 1B). These data indicated that the bulk of BRIL resides at the plasma membrane, with an N-in and C-out configuration indicative of a type II topology. In the Golgi, the C-terminus is lumenal and the N-terminus is cytoplasmic. The lack of any specific staining on confluent naïve MC3T3 attested to the specificity of the antibodies used (Supplemental Fig. 2A). Very similar immunofluorescence localization data were obtained in native MC3T3 when they produce high levels of BRIL 8 days after differentiation (Supplemental Fig. 2B–2C). Localization data obtained after transfection and under native conditions also indicates that the first hydrophobic segment of BRIL (residues 39-60) is not recognized as a bona fide transmembrane passage. However, BRIL can be forced to adopt an unnatural type III configuration (N-out/C-out) when a heterologous cleavable signal sequence is fused at its N-terminus (Supplemental Fig. 3).
The type II topology of BRIL was next verified by determining its susceptibility to trypsin digestion of live transfected cells. The BRIL polypeptide contains several potential trypsin cleavage sites (Fig. 1A, arrows), of which only 3 are located in the C-terminus. We reasoned that only the BRIL polypeptide region accessible to the extracellular trypsin would be sensitive to cleavage. Forced overexpression of BRIL was achieved by transient transfection in MC3T3 cells, which do not express in proliferating and non-differentiated cells.[7, 9] Soluble extracts were prepared from cells detached either by scraping in the NP-40-detergent cell lysis buffer or after digestion with trypsin-EDTA. Proteins were separated on 16% SDS-PAGE and BRIL was detected by western blotting (Fig. 1C). Trypsin digestion caused BRIL to migrate as a single species of slightly lower molecular mass when probed with the anti-N antibody (Fig. 1C left). Assuming complete digestion and trimming of the last 19 residues of BRIL at K115, this would correspond to a loss of 2.2kDa, which is in accordance with the decreased mass observed on gel. It is important to note that there was no appreciable loss in signal intensity of the digested BRIL, suggesting that the bulk of the remaining protein is intact and that trypsin only acted on the extracellular surface. However, total loss of the BRIL signal after trypsin digestion was noted when probed with the anti-C antibody, indicative that the C-terminal epitope had been destroyed. The same results were obtained when using UMR106 osteosarcoma cells that express BRIL constitutively (Fig. 1C right).
Truncation of the C-terminal tail of BRIL does not affect localization
To corroborate that the trypsin-digested BRIL corresponded to genuine cleavage at the C-terminus, two truncated BRIL variants with a stop codon immediately following K115 or K118 were expressed (see Fig. 1A). In permeabilized cells, the two truncated BRIL appeared present at the plasma membrane and Golgi when probed with the anti-N antibody (Fig. 2A). As expected, no signal was observed in intact cells labeled with the anti-C antibody (Fig. 2A). Western blotting revealed that the K115 and K118 had masses close to the trypsin-digested wild type BRIL (Fig. 2B) and were unaffected by trypsin. These data confirmed that trypsin only cleaves the C-terminus and also that this portion is not essential for plasma membrane targeting. Although the signal observed with the K115 and K118 mutants is suggestive of plasma membrane targeting, the possibility that they did not adopt a transmembrane configuration cannot be formally excluded at present.
Mutations in the C-terminal transmembrane segment affect stability and localization
The type II configuration implies that BRIL crosses the plasma membrane only once. The trypsin digest experiments suggested that the transmembrane passage likely corresponds to the predicted region covering residues 89-111. To verify that this region is important for translocation, two leucine residues (L101 and L103) at the center of the transmembrane segment were individually mutated into charged arginine residues, which are expected to disrupt the alpha helix (Fig. 3A). When expressed in MC3T3 cells and immunolocalized with the C-terminal antibody, the mutant BRIL L101 > R or L103 > R (data not shown) were entirely absent from the cell surface of intact cells (Fig. 3B). The L101 > R mutant could be detected in the cytoplasmic compartment with the anti-C after permeabilization of the plasma membrane with digitonin (Fig. 3B). The pattern for the L101 > R appeared far more restricted than the wild type BRIL in terms of surface area. Expression levels of the mutant proteins were drastically reduced as assessed by western blotting (Fig. 3C). These data demonstrate that the single transmembrane passage of BRIL resides in its C-terminal half and that mutation of critical leucine residues abolished membrane targeting.
BRIL is targeted to the ER en route to the plasma membrane
All immunofluorescence labeling demonstrated that BRIL localizes mostly to plasma membranes and to the Golgi apparatus. However, it was unclear whether a significant proportion of BRIL localized to the rER at steady state. All proteins destined to the plasma membrane must first be targeted to the rER as the common entry point of the secretory pathway, and so should BRIL. In order to investigate whether BRIL transits in the rER en route to the surface, the last 4 C-terminal residues (EDYN) were changed to a canonical ER-retention motif (KDEL) (Fig. 4A). The BRIL-KDEL mutant was not detectable on the cell surface on intact cells with the anti-C antibody (Fig. 4B). When probed with the anti-N on cells permeabilized with digitonin, BRIL-KDEL staining looked like an ER-like reticular pattern, was more restricted in terms of surface area, and was absent from the Golgi (Fig. 4B). The trypsin digestion of live cells and western blotting also indicated that BRIL-KDEL was no longer present at the cell surface (Fig. 4C). These data indicate that the C-terminus (containing the KDEL motif) gets translocated within the lumen of the rER.
Anterograde trafficking of BRIL is dependent on COPII vesicles
To ascertain that native BRIL is targeted to the rER and to investigate whether the transport to the Golgi is mediated through COPII-coated vesicles, the SAR1-H79G mutant was utilized. SAR1 is a small GTPase essential for COPII vesicle formation and fission, and the H79G is a dominant negative form locked in a GTP-bound state which blocks this process. BRIL immunolocalization was normal in the presence of wild type SAR1, being detected at the plasma membrane and Golgi (Fig. 5A). Overexpression of SAR1-H79G caused BRIL to be retained exclusively in the rER, as depicted by the appearance of a pure reticular staining pattern and the loss of plasma membrane and Golgi labeling (Fig. 5A). Western blotting also indicated that BRIL becomes resistant to trypsin cleavage in the presence of SAR1-H79G but not with wild type SAR1 (Fig. 5B).
Palmitoylation of BRIL is essential for stability and membrane targeting
The BRIL polypeptide contains 3 cysteine residues (C52, C53, C86) that are perfectly conserved among species. The cysteine residues have been shown to be palmitoylated in IFITM3 (15) and IFITM5, but the requirements for membrane localization have not been systematically addressed. We tested the consequence of cysteine to alanine mutants on BRIL localization in MC3T3 cells. Mutation of either C52 > A (Fig. 6A) or C53 > A (not shown) caused a dramatic reduction of BRIL at the plasma membrane. Only a faint plasma membrane and Golgi-associated signal persisted for those two mutants (Fig. 6A arrowheads). The C86 > A mutant BRIL, however, was still intensely detected at the membrane (Fig. 6A), despite a lower expression level (Fig. 6B). Because all three mutant proteins accumulated equally less than the wild type BRIL (Fig. 6B), the lack of membrane staining for C52 > A and C53 > A is unlikely caused by their low level of expression. Consistent with the immunofluorescence labeling data, trypsin digestion of live cells showed that only the membrane-localized C86 > A mutant was susceptible to cleavage (Fig. 6B). The other two were insensitive to trypsin, indicating that they were not significantly present at the plasma membrane. We next tested whether BRIL is palmitoylated in vivo by performing metabolic labeling. MC3T3 cells were transfected with BRIL or GFP, as a negative control, and collected after a 4 h pulse labeling with 3H-palmitate. Detection by fluorography revealed several palmitoylated products in total lysates, with a band of stronger intensity at the expected size in BRIL expressing cells compared to GFP (Fig. 6C top, arrow). Immunoprecipitation (bound) with the anti-N antibody detected palmitoylated BRIL (Fig. 6C top, arrowhead). A duplicate gel blotted with the anti-N demonstrated expression of BRIL in the lysate, depletion in the unbound material (free) and enrichment in the bound, respectively (Fig. 6C bottom).
Localization and expression is unaltered in the OI type V mutant MALEP-BRIL
A heterozygous C to T transition at position -14 in the 5′UTR of the IFITM5 gene has been shown to be the single recurrent cause of type V OI (see Fig. 8A for a schematic). The mutation introduces a novel translation start site which adds five amino acids (MALEP) to the N-terminus of BRIL. We explored whether this N-terminal extension had any impact on the characteristics of mouse BRIL. In every aspect tested, the mouse MALEP-BRIL did not differ from the wild type BRIL (Fig. 7). MALEP-BRIL adopted a type II topology at the plasma membrane and Golgi (Fig. 7A), was sensitive to trypsin digestion (Fig. 7B), and was palmitoylated (Fig. 7C). Because of the five residue extensions at its N-terminus, MALEP-BRIL migrated slightly slower than the wild type BRIL on SDS-PAGE (Fig. 7B). The migration behavior of MALEP-BRIL was not affected when the natural methionine of BRIL was changed into an alanine (MALEP-M > MALEP-A), suggesting no internal translation (Fig. 7B). An anti-MALEP antibody displayed specificity for the MALEP-BRIL and could only detect it after permeabilization of cells, indicating that the N-terminal extension is cytoplasmic (Fig. 7A). Overexpression of the human wild type BRIL and MALEP-BRIL proteins in MC3T3 showed exactly the same characteristics as the mouse MALEP-BRIL (data not shown). In sharp contrast to MALEP-BRIL, a dramatic change in the subcellular distribution of the mouse BRIL-S42 > L was observed (Fig. 7D). In this mutant, a single point mutation converted serine 42 to a leucine (S42 > L), which corresponds to the serine at position 40 of human BRIL. The signal for BRIL-S42 > L was absent from the plasma membrane and restricted intracellularly to the Golgi apparatus (Fig. 7D). The BRIL-S42 > L was also unaffected by trypsin digestion, supporting the immunofluorescence data that it was not properly translocated to the plasma membrane (Fig. 7E). Metabolic labeling with 3H-palmitate showed that the BRIL-S42 > L is very inefficiently palmitoylated as compared to the wild type (Fig. 7F). By comparison, the KDEL and triple C52 > A/C53 > A/C86 > A mutants completely failed to incorporate palmitate, while the C52 > A/C53 > A displayed some labeling (Fig. 7F). Western blotting indicated all mutants were expressed.
The MALEP-BRIL is expressed and produced in OI type V bone
Lastly, a human bone sample obtained from a child with OI type V was analyzed to detect IFITM5 expression and production (Fig. 8). RT-PCR amplification with primers spanning the two exons (Fig. 8A) yielded the expected 438bp product (Fig. 8B). Sanger sequencing proved that the amplified IFITM5 cDNA had a doublet G/A peak (reverse strand) of equal intensity at position -14 of the 5′UTR (Fig. 8C). Proteins extracted from bone samples of OI type V and type IV patients were analyzed by western blotting with the anti-C antibody. The OI type V material had two immunoreactive bands of equal intensity exactly at the migration size of BRIL (Fig. 8D). In comparison, in proteins from bones of an age-matched child with OI type IV, only the faster migrating wild type BRIL species was detected (Fig. 8D, arrow).
In the present study, two antibodies directed against BRIL extremities were used for a systematic reassessment of its topology by immunofluorescence labeling and western blotting. Trypsin digestion of live cells was also employed as a surrogate technique to evaluate the accessibility to the extracellular compartment. We provide evidence that in transfected MC3T3 osteoblasts BRIL is a type II (N-in/C-out) transmembrane protein (schematized in Fig. 9). This type II topology was also validated in wild type differentiated MC3T3 cells and in UMR106, which express BRIL constitutively. On intact cells, the BRIL C-terminus was detected conspicuously on the plasma membrane surface. The N-terminus was localized predominantly on the cytoplasmic side of the plasma membrane and the Golgi. The N-terminus was only rarely observed at the periphery of intact cells, and the signal is due to partial permeabilization by the fixation procedure. This would potentially explain why our group had originally inferred a type III topology (N-out/C-out) from transfection studies in HEK293 cells with 3xFLAG-tagged versions of BRIL. The putative first hydrophobic region of BRIL is not recognized as a genuine transmembrane segment, and we have no direct evidence at present that this conformation bears any physiological relevance. It remains to be established, however, whether this hydrophobic stretch with the two palmitoylated cysteines in its center, forms an intramembrane configuration in the inner plasma membrane leaflet, as was proposed recently for IFITM3. Just like what we found here for BRIL, the topology of IFITM3 was recently revised from a type III to a type II. The N-terminus of IFITM3 was found in rare instances on the extracellular surface, and this was cell-type dependent, coincidentally being observed only in HEK293.
Although BRIL was not significantly detected in the rER at steady state, all proteins destined to the plasma membrane must go through the rER as the entry point to the secretory pathway. This is true also for BRIL, as introduction of a KDEL retention motif in the C-terminal end resulted in an exclusive rER staining, as was previously shown for other type II proteins such as dipeptidyl peptidase IV and IFITM3. The same result was observed after overexpression of the SAR1-H79G. These experiments confirmed that the C-terminus is lumenal in the secretory pathway and that trafficking of BRIL involves a COPII-mediated vesicular transport. Although rates of transport have not been formally investigated, the lack of significant rER BRIL at steady state implies that the rER-Golgi movement would be rapid.
Another implication of our study is the mechanism by which BRIL is being targeted to the secretory pathway. We showed that the first putative hydrophobic stretch in BRIL is not normally acting as a transmembrane passage or signal anchor. As the genuine transmembrane segment is located in the C-terminus, during BRIL translation it would only emerge out of the ribosome only once the nascent polypeptide is almost fully synthesized. It is reasonable to propose that this would prevent targeting through the canonical co-translational signal peptide recognition particle pathway, and perhaps occur in a post-translational manner. These characteristics would be compatible with targeting through the tail-anchored machinery,[42, 46] which recognizes membrane proteins having their first transmembrane domain usually more than 40 residues away from the initiator methionine. Although this was not directly addressed, some of our results would indirectly support this possibility. For instance, we have consistently observed low levels of expression for some of the mutants, particularly the L101 > R and L103 > R, which accumulated in the cytoplasm. Misfolded or improperly targeted transmembrane proteins are usually rapidly degraded in order to avoid aggregation in the cytoplasm and ensuing toxicity.[47, 48]
We also showed that BRIL is palmitoylated. Mutation of either cysteine residues at position 52 and 53 attenuated targeting of BRIL at the plasma membrane. Cellular levels of either mutant were also significantly reduced. The other highly conserved cysteine at position 86 did not significantly alter BRIL localization and production, although it is palmitoylated. Our results are in accord with and extend those of a recent report showing palmitoylation of BRIL at C52, C53, and C86. In that study, BRIL was reported to migrate as a doublet on SDS-PAGE, representing the top palmitoylated and bottom non-modified forms, respectively. Although we have observed occasional double bands on western blots, this was not consistent even though we took precaution to not overheat samples prior to loading, as this may cause a loss of the palmitate moiety. The increased mass for BRIL carrying two palmitate residues would correspond to less than 0.5 kDa, and this difference would be difficult to discriminate on gel. Tsukamoto et al. further showed by co-immunoprecipitation that palmitoylation of BRIL was required for its interaction with another transmembrane protein, FKBP19 (the product of the Fkbp11 gene), but did not provide cellular localization data. Given that FKBP19 resides predominantly in the rER (data not shown), it is at present unclear how and if this (transient) interaction would impact on BRIL structure or function. A direct peptidyl-prolyl-isomerase action of FKBP19 on the BRIL polypeptide appears unlikely as the enzymatic domain is predicted to be lumenal, facing the opposite compartment to the proline-rich N-terminal cytoplasmic end of BRIL. Whether it would serve another role is presently unknown.
OI type V is caused by a single recurrent heterozygous mutation in the 5′ UTR of IFITM5.[18, 19, 21-25] The c.-14C > T change leads to the creation of an in-frame ATG with the addition of five amino acids (MALEP) at the N-terminus of BRIL. At present it still remains unknown how the mutant MALEP-BRIL protein causes the OI type V phenotype. Because OI-V is transmitted in an autosomal dominant fashion, mutated BRIL could a priori either lead to haploinsufficiency or have a dominant negative effect. In order to shed light on this possibility, we studied the mouse version of MALEP-BRIL. However, we did not find any obvious changes in the localization, topology, production, and palmitoylation status of MALEP-BRIL compared to the wild type, at least not when tested in the “homozygote” transfection conditions. We also provide the first demonstration that BRIL is expressed and produced in OI type V bone tissue. Based on this evidence, it can be put forward that OI type V is caused by a gain of function. The N-terminal extension might generate a novel interface on the cytoplasmic side of the plasma membrane that is deleterious to osteoblast activity.
Another de novo mutation in IFITM5, substituting serine 40 by a leucine, was found in patients presenting extremely severe OI manifestations[26-28] but not the OI type V characteristics. BRIL containing this S40 > L change was poorly palmitoylated and trapped in the Golgi, at least when studied in the context of the mouse BRIL protein (S42 > L), which is a much more dramatic change than what was observed with MALEP-BRIL. Although they remain to be investigated, the mechanisms would appear to be distinct in the two subtypes of OI with IFITM5 mutations. Irrespective of the mechanisms involved, it would argue that both mutations are caused by dominant negative effects occurring in the cytosolic compartment but not by haploinsufficiency. Further studies will be required to verify the subcellular localization of the two OI BRIL mutants when produced, like in the patients, in equal (heterozygote) abundance. Our data further suggest that key residues within the cytoplasmic N-terminal domain, in addition to the cysteines, are critical for targeting BRIL to the proper sub-cellular compartment.
In conclusion, BRIL is a type II transmembrane protein, the bulk of which resides at the cell surface, with some in the Golgi. One important remaining question is about the functional role of BRIL in osteoblast activity and the impact of the BRIL mutants. Ongoing and future studies are required to shed light on the mechanisms at play, which will hopefully provide a handle to develop new and/or more targeted therapeutic strategies for OI. In that regard, our findings clearly indicate that an intracellular approach would need to be envisioned for blocking the pathogenic activity of MALEP-BRIL.
Frank Rauch: Genzyme Inc: Advisory Board member; Novartis Inc: Study grant to institution; Alexion Inc: Study grant to institution. All other authors state that they have no conflicts of interest.
We thank Ms. Liljana Lalic for performing the Sanger sequencing reactions. This work was supported in part by grants from the Shriners of North America and CIHR, and by a Recruitment Aid program from The Network of Oral and Bone Health Research of the Fonds de Recherche du Québec en Santé.
Authors' roles: Study design: MHG, PM. Study conduct: AP, MHG, PKJ, PM. Data collection: AP, MHG, PKJ, PM. Data analysis: AP, MHG, FF, FR, PM. Data interpretation: MHG, FR, PM. Drafting manuscript: FR, PM. Revising manuscript content: AP, MHG, PKJ, FF, FR, PM. Approving final version of manuscript: AP, MHG, PKJ, FF, FR, PM. PM takes responsibility for the integrity of the data analysis.