NF1 Tumor Suppressor Protein and mRNA in Skeletal Tissues of Developing and Adult Normal Mouse and NF1-Deficient Embryos

Authors


  • The authors have no conflict of interest

Abstract

NF1 is a heritable disease with multiple osseous lesions. The expression of the NF1 gene was studied in embryonic and adult rodent skeleton and in NF1-deficient embryos. The NF1 gene was expressed intensely in the cartilage and the periosteum. Impaired NF1 expression may lead to inappropriate development and dynamics of bones and ultimately to the osseous manifestations of the disease.

Introduction: Neurofibromatosis type 1 is caused by mutations in the NF1 gene encoding the Ras GTPase activating protein (Ras-GAP) neurofibromin. Skeletal ailments such as short stature, kyphoscoliosis, and tibial bowing and pseudarthrosis are common osseous manifestations of NF1. These symptoms are congenital, implying a role for neurofibromin in proper bone growth. However, little is known about its expression in skeletal tissues during their development.

Materials and Methods: The expression of the NF1 gene was studied in normal and NF1+/− mouse fetuses at embryonic days 12.5-15.5 and in skeletal tissues of adult mice and rats. In situ hybridization, immunohistochemistry, and Western blot analysis were used to identify the NF1 gene expression profile.

Results: NF1 mRNA and protein were elevated in resting, maturation, and hypertrophic chondrocytes at the growth plate. Parallel studies on NF1+/− embryos showed expression patterns identical to wildtype. The periosteum, including osteoblasts and osteoclasts, and osteocytes of the cortical bone of adult mice were also intensely labeled for NF1 protein and mRNA. Western transfer analysis detected NF1 protein in the respective rat tissues. Phosphorylation of p42 and p44 MAP kinases, the downstream consequence of Ras activation, was elevated in hypertrophic chondrocytes of NF1+/− embryos.

Conclusions: The results suggest that neurofibromin may act as a Ras-GAP in skeletal cells to attenuate Ras transduced growth signals and thus play a role during ossification and dynamics of bone. Loss of NF1 function may therefore lead to dysplastic bone growth, thereby causing the debilitating osseous symptoms of NF1.

INTRODUCTION

NEUROFIBROMATOSIS TYPE 1 (NF1) is one of the most common heritable diseases, affecting 1 in 3500 live births worldwide. Neurofibromas, cafe-au-lait spots, and Lisch nodules are among the most noted diagnostic criteria for the disease. NF1, however, is also characterized by numerous skeletal manifestations, including macrocephaly, short stature, kyphoscoliosis, sphenoid wing dysplasia, and congenital bowing and pseudarthrosis of the tibia.(1–4) These symptoms afflict up to 50% of NF1 patients and are frequently debilitating.(1) Scoliosis is the most common osseous manifestation in NF1, with an incidence ranging from 10% to 30%.(4–7) Congenital bowing and pseudarthrosis are also well-characterized manifestations in NF1, with an incidence of ∼3% each.(5)

Neurofibromatosis type 1 is caused by mutations in the NF1 gene, which encodes neurofibromin, a 2818 amino acid protein with a Ras-GAP (GTPase-Activating-Protein) domain.(8–11) Calculated molecular mass of NF1 protein is 327 kDa, but the apparent molecular mass in SDS-PAGE is ∼220-280 kDa. This discrepancy is most likely caused by protein folding during the migration through denaturing polyacrylamide gels.(12) Neurofibromin is expressed ubiquitously, but its highest levels are found in tissues of the nervous system, where it acts to attenuate p21Ras activation in response to growth and proliferative signals.(13,14)

Very little is known of the natural history of the osseous manifestations in NF1. With the exception of macrocephaly, which is thought to be caused by increased brain volume, the osseous manifestations of NF1 are congenital and considered to be primary dysplasias.(7,15–19) This implies a crucial role for neurofibromin in both bone development and homeostasis. However, the molecular basis of skeletal symptoms has been difficult to study in NF1. This is in part because NF1+/− mice do not display any bone phenotype, and NF1−/− mice die in utero before bone formation can be observed.(20–23) NF1 expression has been noted in the chondrocytes of developing rodents,(14,24) but many aspects of NF1 expression in developing and mature skeletal systems have remained unsolved. As a prerequisite to molecular studies of the function of neurofibromin in bone, we therefore performed a detailed characterization of its mRNA and protein expression in both embryonic and adult skeletal tissues. Our results show that NF1 is expressed in a pattern suggestive of a central role in both embryonic skeletal formation and adult bone homeostasis.

MATERIALS AND METHODS

Tissues

Animals were obtained from Laboratory Animal Centre, University of Oulu, and treated in accordance with the guidelines of the Animal Care and Use Committee. Skeletal samples from fetal mice at embryonic days 12.5, 13.5, 14.5, 15.5 and adult mice (CD-1) were used for immunohistochemistry and in situ hybridizations. Bones of adult rat (Sprague-Dawley) were used for Western blot analyses. The muscles were dissected from the adult bones, which were cut in half to improve fixation. Fetuses and bones from adult mice were fixed in 4% paraformaldehyde (PFA). After fixation, adult bones were decalcified in 0.5 M EDTA at 4°C or in 5% formic acid at room temperature and embedded in paraffin. NF1-deficient (NF1+/− and NF1−/−) fetuses were generated as described.(25)

Immunohistochemistry

Paraffin-embedded tissues were cut to 5-μm-thick sections and plated on silanated glass slides and deparaffinized. The sections were treated with H2O2 in PBS for 5 minutes to remove endogenous peroxidase. The sections were subsequently incubated in 1% bovine serum albumin (BSA)-PBS for 30 minutes to prevent nonspecific binding. To detect NF1 protein, three primary antibodies recognizing different sequences were used: two polyclonal rabbit antibodies [NF1GRP(D); cat. Sc-67; Santa Cruz Biotechnology, Santa Cruz, CA, USA and cat. NB 300-155; Novus-Biologicals, Littleton, CO, USA] and one mouse monoclonal antibody (cat. NB 300-154; Novus-Biologicals). Polyclonal rabbit antibodies were used to detect phosphorylated p44/42 MAP kinase (Thr202/Tyr204; cat. 9101; Cell Signaling Technology, Beverly, MA, USA) and type X collagen (cat. RDI-COLL10abr; Research Diagnostics, Flanders, NJ, USA). Biotin-conjugated antibody (Dako A/S, Glostrup, Denmark) and peroxidase-conjugated streptavidin (Dako A/S) were used as secondary antibodies. DAB color reaction was obtained using Liquid DAB Substrate Bulk Kit (Zymed Laboratories, San Francisco, CA, USA). Histostain-plus kit (Zymed laboratories) was used as an alternative method to perform the stainings. Negative control immunoreactions included the following: primary antibody was replaced with 1% BSA-PBS, and primary antibody was preabsorbed with 100× M excess of synthetic peptide that was originally used for immunization for NF1GRP(D) antibody. In all negative controls, only a faint uniform background labeling was observed (Figs. 1H, 1I, 2G, 2H, and 3I–3K).

Figure FIG. 1..

NF1 protein and mRNA in mouse rib cartilage. (A-D) Hematoxylin-eosin stain. (A) 12.5 days. (B) 13.5 days. (C) 14.5 days. (D) 15.5 days. D and E represent serial sections. (E-G) Immunolabeling for NF1 protein using NF1GRP(D) antibody. (E) NF1 protein was detectable in the hypertrophic, the maturation, and the resting cartilage and in the perichondrium, but not in the proliferative cartilage. (F) High magnification of the boxed area on the left in E. F represents the border between the proliferative and the maturation cartilage. (G) High magnification of the boxed area on the right in E. (H) Negative control immunolabeling for NF1 protein without primary antibody. (I) Negative control immunolabeling for NF1 protein using preincubation of the primary antibody with the peptide originally used for immunization. (J) In situ hybridization for NF1 mRNA with a DIG-labeled antisense probe. NF1 mRNA was expressed throughout the rib cartilage at 15.5 days. The labeling was most intense in the maturation cartilage. (K) Negative control in situ hybridization for NF1 mRNA with sense probe. HY, hypertrophic cartilage; MA, maturation cartilage; PR, proliferative cartilage; RE, resting cartilage. Scale bars = 20 μm.

Figure FIG. 2..

NF1 protein and mRNA and phosphorylated p44/42 MAPK in 15.5-day wildtype and NF1+/− mouse cartilage. (A and B) Hematoxylin-eosin stain. (A) Wildtype (WT) fetus. (B) NF1+/− fetus. (C and D) Immunolabeling for NF1 protein using polyclonal antibody (NB 300-155). (C) The maturation (MA) and the hypertrophic (HY) cartilages of the WT fetus were intensely labeled for NF1 protein, while the proliferative cartilage (PR) showed a faint background labeling only. (D) The labeling for NF1 protein was similar in the cartilage of the NF1+/− fetus compared with WT. (E and F) Immunolabeling for phosphorylated p44/42 MAP kinase. (E) In WT fetus, a minority of chondrocytes locating immediately adjacent to mineralizing bone displayed intensely positive labeling for phosphorylated p44/42 MAPK (arrowhead). (F) The labeling for phosphorylated p44/42 MAPK is markedly upregulated in NF1+/− fetus compared with WT in E. (G and H) Negative control immunolabeling for NF1 protein without primary antibody. (I and J) In situ hybridization for NF1 mRNA with a DIG-labeled antisense probe. (I) In WT fetus, NF1 mRNA was expressed throughout the cartilage. The labeling was most intense in the maturation and in the hypertrophic cartilages. (J) In NF1+/− fetus, the labeling for NF1 mRNA specific antisense probe was similar as in WT fetus. (K and L) Negative control in situ hybridization with sense probe. For all panels, scale bar = 20 μm.

Figure FIG. 3..

NF1 protein and mRNA in adult mouse femur. (A-C) Hematoxylin-eosin stain. (A) Synovial cartilage (SC). (B) Growth plate (GP). (C) Cortical bone and periosteum. (D-G) Immunolabeling for NF1 protein using NF1GRP(D) antibody. (D) Most of the cells in the synovial cartilage were positive for NF1 protein. (E) Within the growth plate, the maturation (MA) and the hypertrophic (HY) cartilages were intensely labeled for NF1 protein, while the proliferative cartilage (PR) showed a faint background labeling only. (F) The periosteum (PS) and osteocytes (OC) of the cortical bone displayed a positive immunoreaction for NF1 protein. (G) Osteoclast on the surface of the cortical bone showing an intense labeling for NF1. Nuclei are marked with asterisks; RB, ruffled border. (H) Western blot analysis of the respective rat tissues for NF1protein. Antibody [NF1GRP(D)] detected a doublet of ∼250-kDa bands (arrows) representative of NF1 protein. The sciatic nerve (SN) served as a positive control. (I and J) Negative control immunolabeling for NF1 protein without primary antibody. (K) Negative control immunolabeling for NF1 protein using primary antibody preincubated with the peptide originally used for immunization (L-N) In situ hybridization for NF1 mRNA with a DIG-labeled antisense probe. (L) The synovial cartilage was faintly labeled for NF1 mRNA. (M) The growth plate expressed NF1 mRNA in the hypertrophic (HY), the maturation (MA) cartilage, and the proliferative cartilage (PR). (N) The periosteum (PS) and osteocytes (OC) of the cortical bone displayed a positive hybridization signal for NF1 mRNA. (O-Q) Negative control in situ hybridization using sense probe showing a faint background labeling only. GP, growth plate; Scale bar for A-F and I-Q is displayed in A. Scale bars = 20 μm.

Western blot analysis

Total protein lysate was created by extracting the rat tissues with RIPA buffer (1% Igepal CA-630, 0.5% sodium deoxycholate, 0.1% SDS, and 1 mM EDTA in PBS supplemented with protease inhibitors; Complete Mini; Boehringer Mannheim GmbH, Mannheim, Germany). Tissues were snap-frozen in liquid nitrogen and crushed with a tissue grinder in RIPA buffer and extracted at 4°C for 30 minutes. The lysate was subsequently centrifuged at 15,000g at 4°C for 10 minutes. Protein concentrations of soluble fractions were measured with DC Protein Assay (Bio-Rad Laboratories, Hercules, CA, USA). Laemmli buffer (3×)(26) was added to each sample to a final concentration of 1×, and 15 μg of each preparation was loaded on SDS 6% polyacrylamide gels. After electrophoresis, proteins were transferred to Immobilon-P filters (Millipore Corp., Bedford, MA, USA) and immunolabeled with anti-NF1 antibody [NF1GRP(D)]. Swine anti-rabbit IgHRP (horse radish peroxidase; Amersham Life Sciences, Little Chalfont, UK) was used as a secondary antibody. Proteins were detected with enhanced chemiluminescence (ECL; Amersham Life Sciences), and the filters were exposed to autoradiographic films (Eastman Kodak, Rochester, NY, USA). The specificity of the NF1GRP(D) antibody was tested with the preincubation of the appropriate synthetic peptide as described.(27)

In situ hybridization

In situ hybridization was carried out as described in detail.(28) The plasmid, containing a 652-bp cDNA fragment corresponding to bases 145-797 of mouse NF1 cDNA (gi:26102424), was linerized to generate templates for digoxigenin (DIG) labeled sense and antisense oriented RNA probes. Tissue sections embedded in paraffin were deparaffinized and further processed for in situ hybridization. The DIG-labeled probe was detected with a sheep anti-digoxigenin antibody coupled to alkaline phosphatase and the color substrate nitrobluetetrazole/ 5-bromo-4-chloro-3-indolyl phosphate (NBT/BCIP) according to the protocol provided by manufacturer (Roche Diagnostics, Mannheim, Germany). In situ hybridization with sense probe served as a negative control showing always only faint uniform background hybridization (Figs. 1K, 2K, 2L, and 3O–3Q).

RESULTS

Expression of NF1 in the developing embryonic skeleton

NF1 expression was studied in developing mouse ribs at embryonic days 12.5-15.5. Generally accepted morphological criteria and immunolocalization of type X collagen, a marker for hypertrophic cartilage, were used to identify the different types of cartilage over this period.(29,30) Resting cartilage, the only type present in e12.5 ribs, consists of spherical, generally nondividing cells (Fig. 1A). At e13.5, columns of flattened proliferative cartilage cells appear, which round up to become maturation cartilage (Fig. 1B). These cells enlarge to form the mineralizing hypertrophic cartilage from e14.5 on (Figs. 1C and 1D).(31) Immunohistochemical detection of neurofibromin, performed with three separate antibodies, showed intense labeling first appearing in maturation cartilage at e13.5 and continuing thereafter with a sharp border between the proliferative and maturation zones (Figs. 1E and 1F). Expression was also seen in resting and hypertrophic cartilage from e14.5 on and in the perichondrium from e13.5 on (Figs. 1E and 1G). At no stage was NF1 protein expression observed in the proliferative zone (Figs. 1E and 1F). In situ hybridization with an antisense probe to NF1 mRNA mirrored the immunohistochemical pattern with the exception of an additional faint hybridization signal in the proliferative zone (Fig. 1J).

Studies of other parts of the skeleton (data not shown) confirmed that NF1gene expression in developing cartilage is generally coincident with cessation of proliferation and progression to a mature, hypertrophic phenotype. Parallel studies on NF1+/− embryos showed expression patterns identical to wildtype, consistent with an absence of skeletal abnormalities in these mice (Figs. 2A–2D, 2I, and 2J). As expected, NF1 knockout embryos showed no NF1 mRNA or protein signal at any stage (data not shown).

NF1 expression in the adult skeleton

In addition to tracing the expression of NF1 in the cartilage of the developing skeleton, we also examined its pattern of expression in adult bone tissues (Figs. 3A–3C). In the growth plates of tibias and femora from adult mice, NF1 mRNA and protein expression followed essentially the same pattern as was observed in embryonic cartilage; that is, both were found in maturation and hypertrophic cartilages, but not, for the most part, in proliferative cartilage (Figs. 3E, 3M, and 4B). Synovial cartilage was also positive for NF1 mRNA and protein (Figs. 3D and 3L).

The vast majority of cells in the periosteum were also strongly NF1 positive, as measured by both immunohistochemistry and in situ hybridization (Figs. 3F and 3N). Because osteoblasts are the most numerous cell type in the periosteum, we conclude that these NF1 expressing cells are, in fact, osteoblasts. Additionally, both osteoclasts and osteocytes of the cortical bone were NF1 positive (Figs. 3E, 3G, and 3N).

Western analysis of skeletal tissue extracts from adult rat identified the expected doublet(32,33) of neurofibromin-specific bands at ∼250 kDa, confirming NF1 expression in synovial and growth plate cartilages and periosteum (Fig. 3H). In fact, the growth plate cartilages displayed the highest NF1 expression of any of the skeletal tissues analyzed (Fig. 3H).

Analysis of NF1 expression in adult calvariae showed the same pattern as in the periosteum of long bones (data not shown). Thus, our data show that NF1 is expressed in differentiated bone cell types in tissues undergoing both endochondral and intramembranous ossification.

Ras-MAPK activation in wildtype and NF1-deficient skeletal tissues

The known function of neurofibromin is as a Ras-GAP, acting to attenuate the activity of Ras and thereby inhibiting growth factor induced proliferation. A major end result of Ras pathway activation in this circumstance is phosphorylation of the p42 and p44 MAPK (also called Erk1 and Erk2). We therefore examined levels of phosphorylated p44/42 MAPK (p-p44/42 MAPK) in embryonic and adult mouse tissues as readout of Ras activity.(34) In both cases, the proliferative and maturation cartilages were negative for p-p44/42 MAPK, whereas approximately one-third of hypertrophic chondrocytes located immediately adjacent to mineralizing bone (identified morphologically [Fig. 1] and by collagen X immunolabeling [Fig. 4A]) were positive (Figs. 2E and 4C). In the synovial cartilage, p-p44/42 MAPK was expressed in the upper layers by some chondrocytes (data not shown). Periosteal cells (again, mostly differentiated osteoblasts) and osteocytes were devoid of p-p44/42 MAPK signal (data not shown). In general, we found that NF1 expression and p-p44/42 MAPK labeling patterns were largely complimentary, with the exception of some hypertrophic chondrocytes. Interestingly, p-p44/42 MAPK signal was markedly elevated in the hypertrophic chondrocytes of NF1+/− fetuses (Fig. 2F). This result is consistent with our hypothesized role for neurofibromin as a Ras-GAP in bone.

Figure FIG. 4..

Collagen type X, NF1 protein, and p-p44/42 MAPK in adult mouse growth plate. (A) Immunolabeling for collagen type X. The hypertrophic chondrocytes (HY) are intensely labeled for type X collagen, while the maturating chondrocytes (MA) display only a faint labeling, and the proliferative cartilage (PR) displays no immunosignal. (B) Immunolabeling for NF1 protein using polyclonal antibody (NB 300-155). The maturation and the hypertrophic cartilages are intensely labeled and the proliferative cartilage shows a faint background labeling only. (C) Immunolabeling for phosphorylated p44/42 MAP kinase. Chondrocytes located immediately adjacent to mineralizing bone displayed intensely positive labeling for phosphorylated p44/42 MAPK. For all panels, scale bar = 20 μm.

DISCUSSION

Patients afflicted with type 1 neurofibromatosis may suffer from a range of osseous disorders including short stature, sphenoid wing dysplasia, kyphoscoliosis, and tibial bowing. Review of the literature indicates that the osseous manifestations of NF1 are almost always located in the bones derived through endochondral ossification where chondrocytes are replaced by bone forming cells. Clavicles and most of the bones of the skull are formed through intramembranous ossification, a process that does not include cartilaginous anlage. The sphenoid bone at the base of the skull is an exception because it is mainly formed through endochondral ossification.(35)

The congenital nature of the osseous dysplasias seen in NF1 suggests that neurofibromin plays a critical role in bone formation and that mutation of NF1 can lead to significant structural deficiencies in bone tissue. Furthermore, tibial bowing and pseudarthrosis seen in young NF1 patients suggests that NF1 continues to be important in postdevelopmental bone dynamics. Our immunohistochemical and in situ hybridization data support such roles for NF1 in skeletal tissue. We found that NF1 mRNA and protein were upregulated both in the perichondrium and cells of the growth plate cartilage at sites of transition from a proliferative to a mature, hypertrophic state and in the osteoblasts and osteoclasts of the periosteum in mature bone. Hypertrophic cartilage plays a key role in the formation of most bones by providing the initial mineralized matrix that is remodeled into mature bone in the process of endochondral ossification. Loss of neurofibromin by NF1 mutation might therefore upset the formation of the hypertrophic zone and result in defects such as short stature. Similarly, loss of NF1 in the periosteum might cause an inappropriate failure to cease proliferation and explain the growth of periosteal-associated tissue at the expense of cortical bone that is seen in pseudarthroses in NF1 patients.(15)

Our finding that the hypertrophic cartilage of NF1+/− embryos shows elevated p-p44/42 MAPK expression suggests that the molecular basis for such a defect may lie in hyperactivation of the Ras-MAPK pathway, leading to increased p44/42 MAPK phosphorylation as is seen in neurofibromin-deficient neural tissue (i.e., neurofibromas).(36) Interestingly, patients with thanatophoric dysplasia (TD), in which a mutation in the fibroblast growth factor (FGF) receptor 3 (FGFR3) gene leads to elevated p44/42 MAPK levels and increased number of apoptotic cells in the growth plate chondrocytes, suffer from a similar set of symptoms as those seen in NF1, including short stature, kyphosis, lumbar lordosis, and macrocephaly.(37) It has been suggested that the Ras-MAPK pathway might thus play a role in chondrocyte differentiation rather than proliferation in growth plate chondrocytes.(37) It seems likely that neurofibromin is a key regulator of the Ras-MAPK pathway in chondrocytes, and loss of neurofibromin can cause dysregulation of that pathway leading to the congenital skeletal dysplasias of NF1. Furthermore, the fact that elevated p-p44/42 MAPK is seen in the bones of heterozygotes implies that these symptoms can result from haploinsufficiency for NF1 and may therefore explain the congenital nature of these bony symptoms (they do not require loss of heterozygosity to manifest themselves) and the variable expressivity of the symptoms.

In summary, the results of this study suggest that neurofibromin may play an important role in the progression of cells of skeletal tissue from a proliferative to a mature, mineralizing state. Furthermore, they are consistent with a hypothesized role for neurofibromin as a Ras-GAP in bone cells. More detailed experiments, such as conditional knockout of NF1 in specific skeletal cell types, will be needed to decipher the precise role of neurofibromin in bone biology and thereby determine the molecular nature of the osseous syndromes of NF1.

Acknowledgements

We thank Marja Paloniemi, Pirkko Peronius, and Paula Salmela for excellent technical assistance. This study was supported by Cancer Society of Finland grants, Oulu University Hospital Grant H01139, Academy of Finland grants, The Finnish Medical Foundation grants, and Maud Kuistila Foundation grants.

Ancillary