Characterization of the osteoblast-specific transmembrane protein IFITM5 and analysis of IFITM5-deficient mice
- First Online:
- Cite this article as:
- Hanagata, N., Li, X., Morita, H. et al. J Bone Miner Metab (2011) 29: 279. doi:10.1007/s00774-010-0221-0
- 584 Downloads
Interferon-inducible transmembrane protein 5 (IFITM5) is an osteoblast-specific membrane protein whose expression peaks around the early mineralization stage during the osteoblast maturation process. To investigate IFITM5 function, we first sought to identify which proteins interact with IFITM5. Liquid chromatography mass spectrometry revealed that FK506-binding protein 11 (FKBP11) co-immunoprecipitated with IFITM5. FKBP11 is the only protein it was found to interact with in osteoblasts, while IFITM5 interacts with several proteins in fibroblasts. FKBPs are involved in protein folding and immunosuppressant binding, but we could not be sure that IFITM5 participated in these activities when bound to FKBP11. Thus, we generated Ifitm5-deficient mice and analyzed their skeletal phenotypes. The skeletons, especially the long bones, of homozygous mutants (Ifitm5−/−) were smaller than those of heterozygous mutants (Ifitm5+/−), although we did not observe any significant differences in bone morphometric parameters. The effect of Ifitm5 deficiency on bone formation was more significant in newborns than in young and adult mice, suggesting that Ifitm5 deficiency might have a greater effect on prenatal bone development. Overall, the effect of Ifitm5 deficiency on bone formation was less than we expected. We hypothesize that this may have resulted from a compensatory mechanism in Ifitm5-deficient mice.
KeywordsIFITM5 Osteoblasts Protein interaction Knockout mice Bone growth
Osteoblasts originate from mesenchymal stem cells and play a crucial role in bone formation. In a previous DNA microarray analysis of the gene expression of mouse osteoblast-like MC3T3 cells, we found that the gene for interferon-induced transmembrane protein 5 (Ifitm5) is an osteoblast marker . Previously, Ifitm5 was known as fragilis4, a member of the mouse fragilis family, which consists of 5 genes and at least 3 pseudogenes . A homology search revealed that the human genome contains a similar gene family, the Ifitm family, which is involved in antiproliferative signaling and homotypic cell adhesion [3, 4, 5, 6]. Because of its similarity to the human Ifitm family, the mouse fragilis family was renamed as the Ifitm family. This family is widespread in the animal kingdom and is known to occur in other species such as rats.
Gene products of the Ifitm family have been predicted to span the membrane twice, with both the amino-terminal and carboxyl-terminal sequences positioned extracellularly. The first transmembrane domain and intracellular loop are the most highly conserved regions of the proteins produced by genes in this family . Human IFITM1 (also known as 9-27 and Leu-13) is a component of protein complexes implicated in cell-to-cell homotypic adhesion and antiproliferative signaling in leukocytes and endothelial cells [3, 4, 7, 8]. In rats, the IFITM1 homolog is known as Rat8 and is thought to be involved in the formation of domes within the maturing rat mammary gland [9, 10, 11, 12]. Recent studies of the IFITM family revealed that mouse Ifitm1–3 (also known as mil-2/fragilis2 in Ifitm1, mil-3/fragilis3 in Ifitm2, and mil-1/fragilis/fragilis1 in Ifitm3) are expressed with a highly dynamic temporal and spatial pattern in germ cells during gastrulation, and are involved in movement of germ cells from the mesoderm into the posterior embryonic endoderm [2, 13, 14]. In contrast, Ifitm5 (also known as fragilis4) and Ifitm6 (also known as fragilis5) are not known to be expressed during gastrulation, and their functions were unknown until recently.
So far, only five genes have been identified as bone-specific: two transcription factors (Runx2 and osterix), two non-collagenous proteins (osteocalcin and bone sialoprotein), and one secreted protein (osteocrin). Runx2-null embryos are characterized by an absence of differentiated osteoblasts and by a complete lack of ossification . However, Runx2 is not osteoblast-specific, because this transcription factor is required for not only osteoblastogenesis but also for chondrogenesis . Osterix acts downstream of Runx2 and is also required for osteoblast differentiation . In osterix-deficient embryos, an absence of differentiated osteoblasts resulted in a lack of cortical and trabecular bone formation via endochondral and intramembranous ossification. Histological examination indicated that osterix was involved in the expression of type I collagen α1 chain, bone sialoprotein, and osteocalcin, although osterix did not affect osteoclastogenesis . Osteocalcin is the most abundant non-collagenous protein that is produced during the late stage of osteoblast maturation, and its expression is associated with bone mineralization . However, osteocalcin-deficient mice showed an increase in bone formation but not impairments in bone resorption, which implies that osteocalcin is a negative regulator of bone formation . On the other hand, although the expression of bone sialoprotein is also highly associated with bone mineralization, bone sialoprotein-null adult mice have greater trabecular bone volume and lower bone formation rates due to reduced bone resorption by osteoclasts . Osteocrin is a vitamin D-regulated bone-specific protein that is expressed in Runx2-positive, osteocalcin-negative osteoblasts . This secreted protein appears to be a negative regulator of bone nodule formation.
The results of previous work strongly suggested that Ifitm5 expression is osteoblast-specific and plays a critical role in bone formation. To clarify the function of this gene, we first characterized its expression patterns. Next, we identified which proteins associated with IFITM5 and generated IFITM-5-deficient mice in order to examine which, if any, functions were lost. Here, we show that although IFITM5 interacts with FK506 binding protein 11 (FKBP11), which has previously been suggested to regulate bone density, mice lacking Ifitm5 did not experience significant changes in their bone formation process, but did exhibit reductions in bone size.
Materials and methods
Cell culture and real-time quantitative PCR
Osteoblast-like MC3T3 cells were provided from RIKEN Cell Bank (RCB1126). The subcultured cells were seeded at a density of 5,600 cells/cm2 in 35-mm dishes and cultured in α-Modified Eagle’s Medium (α-MEM, the proliferation medium) containing 10% (v/v) fetal bovine serum (FBS), 100 U/ml penicillin, and 100 μg/ml streptomycin. On day 3, this was replaced with differentiation medium, which was the proliferation medium supplemented with 50 μg/ml ascorbic acid and 2 mmol/l β-glycerophosphate, to promote osteoblastic differentiation. Culture medium was exchanged every 3rd day. All cultures were incubated at 37°C in a humidified atmosphere containing 5% CO2. Mineralized nodules were stained with Alizarin red.
We analyzed mRNA using real-time quantitative PCR (qPCR; Roche, Basel, Switzerland). cDNA synthesized by the PrimeScript RT reagent kit (Takara Bio Inc., Shiga, Japan) was subjected to amplification using the following gene-specific primers: Ifitm1 forward (5′-ggaagatggtgggtgatacg-3′) and reverse (5′-gggcaaatggtcaggactaa-3′); Ifitm2 forward (5′-cactgccaagtgcctgaata-3′) and reverse (5′-gcatacgcgtagggtaaagg-3′); Ifitm3 forward (5′-cctacgcctccactgctaag-3′) and reverse (5′-tgttacacctgcgtgtaggg-3); Ifitm5 forward (5′-cagctcctgggagttacagc-3′) and reverse (5′-aagtggggaaagggacaaag-3′); Ifitm6 forward (5′-agatactaagggtgggactcca-3′) and reverse (5′-aggggcagagcacatgaat-3′); Bglap2 forward (5′-atgaggaccctctctctgct-3′) and reverse (5′-ccgtagatgcgtttgtaggc-3′); and gapdh forward (5′-ttcaccaccatggagaaggc-3′) and reverse (5′-ggcatggactgtggtcatga-3′). The expression level was normalized using gapdh as an internal standard.
In situ hybridization
Embryos were fixed with Tissue Fixative (Genostaff, Tokyo, Japan), embedded in paraffin, and sectioned at 6 μm. After the tissue sections were de-waxed with xylene and rehydrated through a graded ethanol series finishing in phosphate-buffered saline (PBS), they were fixed with 4% paraformaldehyde and dehydrated with a graded series of ethanols. Hybridization was performed for 16 h at 60°C with probes at concentrations of 100 ng/ml in probe diluent (Genostaff). After treatment with 0.5% blocking reagent (Roche) in TBST, the sections were incubated for 2 h with anti-DIG AP conjugate (Roche) diluted 1:1000 with TBST. Coloring reactions were performed overnight with BM purple AP substrate (Roche). The sections were counterstained with Kernechtrot stain solution (Mutoh Chemical Co., Tokyo, Japan), dehydrated, and then mounted with malinol (Mutoh).
Recombinant DNA constructs and transfection
Forward primer sequence
Reverse primer sequence
Approximately 24 h prior to transfection, cells were seeded at a density of approximately 20,000 cells/cm2 in 35-mm dishes containing 2.5 ml of the proliferation medium. After overnight culturing, the medium in each dish was exchanged for 2.5 ml of the differentiation medium. Two hundred fifty microliters of serum-free medium and 7.5 μl of TransIT-LT1 reagent (Mirus Bio, Madison, WI) were put into a sterile tube and incubated for 20 min at room temperature before 2.5 μg of the expression vector DNA was added, after which the tube was incubated for an additional 30 min. The TransIT-LT1 reagent/vector DNA complex mixture was then added to the cells in the dishes. After 24–48 h of incubation, the cells were harvested.
Immunoprecipitation, Western blot, and liquid chromatography mass spectrometry analysis
FLAG-fused proteins were expressed in osteoblast-like MC3T3 or fibroblast NIH3T3 cells using the indicated expression vector. Total proteins were extracted from 3–5 × 106 cells using a total protein extraction kit (Millipore, Billerica, MA) and incubated with Anti-FLAG M2 affinity gel (Sigma-Aldrich, St. Louis, MO) at 4°C for 2 h. To recover the FLAG-fused proteins, the affinity gels were collected and incubated in 500 ng/ml 3× FLAG buffer at 4°C for 1 h.
For Western blot analysis, 30 μg of total protein was size-fractionated on a precast polyacrylamide gel (15% acrylamide, Atto Corporation, Tokyo, Japan) and blotted onto an Immobilon-PSQ membrane (Millipore). After 1 h of blocking at room temperature with 3% Immunoblot Blocking Reagent (Millipore), the membrane was incubated overnight at 4°C with a primary antibody. The anti-IFITM5 polyclonal antibody prepared from the amino-terminal peptide sequence (TSYPREDPRAPSSRC), anti-FLAG mAb (no. F1804, Sigma-Aldrich), anti-GFP mAb (no. 632380, Clontech), and anti-HA mAb (no. ab18181, abcam) were used as primary antibodies. The membrane was then washed three times with PBS containing 0.05% Tween-20 and incubated for 1 h at room temperature with the HRP-F(ab′) of goat anti-rabbit IgG(H + L) (Zymed Laboratories, San Francisco, CA). Proteins were detected using Immobilon Western Chemiluminescent HRP Substrate (Millipore).
For protein identification using liquid chromatography mass spectrometry (LC-MS), immunoprecipitants stained with EzStain Silver (ATTO, Tokyo, Japan) were dissected from polyacrylamide gel. The gel was dehydrated with acetonitrile and dessicated before being treated with 25 ng/μl (v/v) acetonitrile containing 0.1% trifluoroacetic acid, after which it was subjected to LC-MS. The spectrum data obtained from LC-MS/MS were identified using the Mascot program (Matrix Science, London, UK).
Generation of Ifitm5-deficient mutant mice
To construct a targeting vector, we used PCR to isolate 2.6-kb (SacII 6886-short arm-9500-NotI) and 5.5-kb (ClaI-11501-long arm-17000-SalI) fragments flanking the Ifitm5 gene from a BAC clone (ID: RP23-388N13) containing C57BL/6 genomic DNA (see Fig. 5a). The 2.6- and 5.5-kb fragments were cloned into the SacII–NotI site of pBS-NEO-DTA (Unitech, Chiba, Japan) and a ClaI–SalI site of pBluescript SK(+) (Stratagene, La Jolla, CA), respectively. The latter construct, containing the 5.5-kb fragment, was digested with SalI and then blunt-end ligated into a blunt-end EcoRI–XhoI site of pBS-DTA (Unitech Co., Ltd.). To obtain the final targeting vector, the construct containing the 2.3-kb fragment was digested with SacII and ClaI, then ligated into a SacII–ClaI site of the pBS-DTA with the 5.5-kb fragment.
Embryonic stem (ES) cells derived from C57BL/6 were electroporated with the targeting vector and selected by PCR. The forward primer (5′-CAGGAAGGAATTAAACAGAACTTGA-3′) was designed in the upstream area of the short arm, whereas the reverse primer (5′-CTTCCTCGTGCTTTACGGTATC-3′) was designed in the Neo region. The PCR was run for 35 cycles, with an annealing temperature of 60°C; this yielded a 3076-bp product.
Mutant ES cell clones were identified by southern blot analysis of HpaI-digested genomic DNA, using 5′ and 3′ probes located outside the regions of the short and long arms (see Fig. 5b). Chimeric mice were obtained by injecting mutant ES cell clones into Balb/c blastocysts and transferring them into uteri of pseudopregnant ICR females. The resulting mice were backcrossed to C57BL/6 individuals. Heterozygous mutants were identified by PCR-genotyping tail-tip DNA using the aforementioned primer set. Brother-sister mating was used to generate homozygous mutants.
For the final round of PCR genotyping, two primer sets were designed, in which the forward primer (5′-TGTAGTCAATGAGACAGCAAGTAGC-3′), which was designed on the short arm, was identical. This was paired with one of two reverse primers: one designed on the Neo region, for identification of the target allele (5′-CTTCCTCGTGCTTTACGGTATC-3′), and the other designed on the Ifitm5 gene region, for identification of the wild-type (WT) allele (5′-TACTTCACAACCACGTTCTTTCAT-3′). The PCR products were 658 and 2236 bp for the target allele and WT allele, respectively (see Fig. 5c).
All animal experiments were approved by the local Animal Testing and Ethics Committee and veterinary authorities.
Skeletal preparation and analysis of bone morphometry
Skeletal preparation for alcian blue and alizarin red staining was performed as previously described . The length of the long bones was measured with a vernier caliper. μ-CT images were obtained from a micro focus X-ray CT system (SMX-100CT, Shimadzu, Kyoto, Japan). Bone morphometric parameters were calculated from μ-CT images using TRI/3D-Bone (Ratoc System Engineering, Tokyo, Japan).
Undecalcified 3-μm paraffin-embedded sections were prepared from newborn mice and stained using von Kossa’s method and hematoxylin and eosin (HE). Tartrate-resistant acid phosphatase (TRAP) staining was performed using a TRACP staining kit (TaKaRa Bio) according to the manufacturer’s instructions. Osteocalcin was detected by immunostaining, using Anti-Mouse Osteocalcin polyclonal, rabbit (TaKaRa Bio), and Alexa Fluor 660 goat anti-rabbit IgG (H + L) (Invitrogen, Carlsbad, CA) as primary and secondary antibodies, respectively.
Results are expressed as mean ± standard error (SE) from three to five independent experiments. Student’s t tests were used to compare results between treatment groups, and significance was defined as p < 0.05.
Expression of Ifitm5 associated with bone nodule formation in osteoblasts
Bone-restricted expression of Ifitm5 transcripts in mouse embryos
Interactions between IFITM5 and other proteins
We identified three candidate IFITM5-binding proteins: FK506-binding protein 11 (FKBP11), B cell receptor-associated protein 31 (BCAP31), and hydroxysteroid (17-beta) dehydrogenase 7 (HSD17b7) (data not shown).
All members of the FKBP family are able to bind immunosuppressant compounds such as cyclosporin, FK506, and rapamycin . Thus, we investigated the effect of FK506 on the interaction between IFITM5 and FKBP11. Immunosuppressant FK506 markedly suppressed bone nodule formation, but significantly increased the expression level of IFITM5 mRNA (Fig. 4c). To examine the effect of FK506 on the binding of IFITM5 to FKBP11, we co-expressed FLAG-fused FKBP11 and IFITM5 in MC3T3 cells in the presence of FK506, then performed an immunoprecipitation with the anti-FLAG antibody. The level of co-immunoprecipitated IFITM5 slightly decreased in the presence of FK506 (Fig. 4d), but the FK506 treatment did not significantly affect the interaction between IFITM5 and FKBP11.
Effects of Ifitm5 knockout
In vitro experiments using mouse osteoblast-like MC3T3 cells, which are an in vitro model for the process of osteoblast maturation [24, 25], indicated that Ifitm5 expression was strongly associated with bone nodule formation. The Ifitm5 expression pattern observed here was similar to that documented in primary mouse osteoblasts and human osteoblasts (data not shown). Moffatt et al.  also reported the involvement of Ifitm5 in bone nodule formation in rat primary osteoblasts, rat osteosarcoma UMR106 cells, human osteosarcoma SaOS-2 cells, and human primary osteoblast cultures. These observations suggest that Ifitm5 is involved in bone formation in many species. In primary rat osteoblast cultures, IFITM5 expression co-localized with bone sialoprotein , which is associated with ALP and extracellular bone acidic glycoprotein-75 in bone matrix mineral nucleation sites [27, 28].
Both in vitro and in vivo findings suggest that IFITM5 expression is osteoblast-specific and related to bone formation. For instance, Ifitm5 expression was restricted to the skeletal system and was first observed around E14.5. This is consistent with the time when osteoblasts appear and start to form a mineralized matrix , which indicates that IFITM5 is involved in both endochondral and intramembranous ossification during embryogenesis. At E14.5, transcripts were confined mainly to the peripheral layers of the femur and tibia, though strong expression was also observed in the primary ossification center. The timing and localization of Ifitm5 expression during embryogenesis are similar to those of osterix . Ifitm5 expression was also highly restricted in adult rats , suggesting Ifitm5 expression is involved not only in bone formation in during embryogenesis, but also during postnatal development.
Many membrane proteins have been identified as playing a critical role in bone physiology, and most of these proteins are receptors for hormones and cytokines. However, IFITM5 probably does not possess receptor activity, since the extracellular and intracellular domains are too short to play a role in receptor and signal transduction. Additionally, the intracellular domain of IFITM5 does not contain any known phosphorylation motifs. In NIH3T3 cells, overexpressed IFITM5 has two forms: one with high molecular mass and the other with low molecular mass. Digestion of high-molecular-mass IFITM5 by endoglycosidase did not generate low-molecular-mass IFITM5 (data not shown), suggesting that IFITM5 does not contain a glycoform. Low-molecular-mass IFITM5 may therefore be a truncation mutant of IFITM5. In MC3T3 cells, FKBP11 interacts with IFITM5, but only in its high molecular mass form, implying that there are some major functional differences between the two forms.
FKBP11 is a member of the FK506-binding protein family. FKBPs have been implicated in protein folding, since every member of the family has demonstrated peptidyl-prolyl cis/trans isomerase (PPIase) activity. Immunosuppressant FK506 binds to the PPIase domain, resulting in PPIase inhibition . Several members of this family are known for their ability to bind to calmodulin/calcineurin. When bound to FK506, these products inhibit calmodulin/calcineurin phosphatase activity, which blocks the dephosphorylation of NFATc1 and therefore prevents its transfer to the nucleus . Recently, NFATc1 has been reported as a critical factor for bone formation  via its role as a co-factor for osterix . Although FK506 markedly inhibited bone nodule formation in MC3T3 cell cultures, it did not significantly affect the interaction between FKBP11 and IFITM5. This suggests that IFITM5 does not contribute to the calmodulin/calcineurin signaling pathway. Rather, FKBP11 may act as an adaptor molecule allowing IFITM5 to associate with other proteins.
The FK506 treatment also stimulated Ifitm5 expression, which we hypothesized was related to inhibition of NFATc1 activation. However, when we treated MC3T3 cells with an NFATc1 inhibitor, there was no significant difference in Ifitm5 expression (data not shown). This was also true when we treated MC3T3 cells with a MAPK pathway inhibitor; however, treatment with a PI3K-Akt pathway inhibitor did lead to suppression of Ifitm5 expression (data not shown).
According to the Mouse Genome Information (MGI) database, mice that are homozygous for disruption of the FKBP11 gene display increased bone density. This raises the possibility that IFITM5 and FKBP11 might cooperatively regulate bone formation. However, we did not observe any significant differences in bone morphometry parameters between heterozygous and homozygous newborns. This implies that the function of FKBP11 in bone formation may be independent of IFITM5. Also, there were no significant differences in the number of TRAP-positive cells or osteocalcin production between Ifitm5+/− and Ifitm5−/− newborn mice, suggesting that Ifitm5 deficiency does not affect either osteoclastogenesis or osteoblastogenesis. One of the characteristic features of Ifitm5−/− newborn mice was reduced skeletal size—in particular, the length of the long bones. The difference in long bone length between adult (40-, 48-, and 51-week-old mice) Ifitm5+/− and Ifitm5−/− mutants was less extreme, but still significant. On the other hand, the other bone deficiencies observed in newborns (e.g., less-calcified mandible, incisive bone and cranial incisor, as well as a thinner cranium) had disappeared after 5 weeks. These observations suggest that disruption of Ifitm5 affects bone growth, especially longitudinal length, during prenatal bone development rather than postnatal bone growth.
Newborn homozygous mutants also had bone deformities (bending of the ulna, radius, and tibia) similar to those previously observed in osterix-null newborn mice . In the osterix-null mice, bending was not observed in E15.5 embryos, where the skeletal systems consisted only of cartilage without mineralization . This implies that osterix-null mice have normal cartilaginous templates and that bending occurs during the mineralization process. Although these observations suggested that Ifitm5 might act downstream of osterix and may be directly involved in the deformity, our promoter analysis showed that osterix was not a major regulator of Ifitm5 expression (data not shown). Thus, the cause of the bending remains unknown.
It has previously been suggested that IFITM1, 2, and 3 are associated with the germ line competence of epiblast cells and that they mediate migration of early primordial germ cells [2, 13, 34]. However, previous work found that in mutant mice lacking the entire Ifitm family, including Ifitm5, germ cell development proceeded normally, and mutant adults had normal reproductive organs and were fertile . In contrast, our homozygous female mutants had smaller litters than heterozygous female mutants, suggesting that the lack of IFITM5 may be associated with some type of reproductive disorder. However, this is only one of several possibilities, and requires further exploration.
Cumulatively, our work did not indicate that IFITM5 played a critical role in bone formation, which was contrary to our hypothesis. However, in previous studies, use of small hairpin RNA to knockdown Ifitm5 in MC3T3 cells led to significant decreases in bone nodule formation . The absence of a similar result here suggests that our mutant mice benefited from an as-yet-identified compensation mechanism. Recently, it was reported that IFITM1, 2, and 3 inhibit the early reproduction of influenza A virus and flaviviruses . Our DNA microarray study revealed that overexpression of Ifitm5 in MC3T3 cells leads to increased expression of genes involved in immunity, such as Ccl5, Bst2, Irgm, Ifit3, and B2m (data not shown). In addition, according to the MGI database, mice homozygous for a disruption of the IFITM5 partner molecule, FKBP11, show increased antibody and inflammatory responses. Thus, although IFITM5 does not appear to be essential for bone formation, this osteoblast-specific protein may play an integral role in regulating the immune system in bones. However, further studies are required to explore this possibility and to provide more details on the function of IFITM5 in osteoblast maturation.
We are grateful to Ms S. Kajiwara and M. Maeda for their technical assistance. We also would like to thank Dr. T. Koda for his useful advice in knockout mice analysis. Part of this research was funded by the Research Promotion Bureau of the Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan.