|Home | About | Journals | Submit | Contact Us | Français|
Neurofibromatosis type 1 patients develop peripheral nerve tumors (neurofibromas) composed mainly of Schwann cells and fibroblasts, in an abundant collagen matrix produced by fibroblasts. Trauma has been proposed to trigger neurofibroma formation. To test if loss of the neurofibromatosis type 1 gene (Nf1) compromises fibroblast function in vivo following trauma, skin wounding was performed in Nf1 knockout mice. The pattern and amount of collagen-rich granulation bed tissue, manufactured by fibroblasts, was grossly abnormal in 60% of Nf1+/− wounds. Nf1 mutant fibroblasts showed cell autonomous abnormalities in collagen deposition in vitro that were not mimicked by Ras activation in fibroblasts, even though some Nf1 effects are mediated through Ras. Nf1+/− skin wound fibroblasts also proliferated past the normal wound maturation phase; this in vivo effect was potentiated by muscle injury. In vitro, Nf1+/− fibroblasts showed higher proliferation in 10% serum than Nf1+/+ fibroblasts. Macrophage-conditioned media or epidermal growth factor potentiated Nf1+/− fibroblast proliferation in vitro, demonstrating abnormal response of mutant fibroblasts to wound cytokines. Thus Nf1 is a key regulator of fibroblast responses to injury, and Nf1 mutation in mouse fibroblasts causes abnormalities characteristic of human neurofibromas.
Neurofibromatosis type 1 (NF1) is one of the most common inherited autosomal dominant human diseases, affecting 1 in 3500 individuals (Huson et al, 1989; Riccardi, 1992). Patients with NF1 develop benign peripheral nerve sheath tumors called neurofibromas. These tumors contain two major cell types: fibroblasts and Schwann cells. Mutations in both Nf1 alleles are detectable in neurofibromas, suggesting that one of these cell types has sustained mutations in both copies of the Nf1 tumor suppressor gene (Colman et al, 1995; Sawada et al, 1996). Whether Nf1 mutations in Schwann cells alone, fibroblasts alone, or in both cell types contribute to neurofibroma formation is unclear (reviewed in Rosenbaum et al, 1997).
Neurofibroma fibroblasts and Schwann cells are embedded in an extensive connective tissue matrix containing a large amount of collagens, proteoglycans, fibronectin, and laminin (Uitto et al, 1986; Konomi et al, 1989). Up to 70% of the tumor dry weight of a neurofibroma is collagen (Peltonen et al, 1986; Uitto et al, 1986). Type I collagen is the major species, with collagens types III and V present in lesser amounts (Uitto et al, 1986; Konomi et al, 1989). Neurofibroma fibroblasts, which account for 20%–60% of the cells within the tumors, synthesize collagen I and III (Peltonen et al, 1988; Jaakkola et al, 1989; Sollberg et al, 1991). In comparison with normal skin fibroblasts, early passage fibroblasts from neurofibromas synthesize and secrete higher amounts of collagen (Peltonen et al, 1981, 1986; Uitto et al, 1986; Sasaki et al, 1992). These studies imply that fibroblasts might contribute to the pathogenesis of neurofibromas. It is not known, however, if fibroblasts in neurofibromas behave abnormally in response to the tumor environment or as a result of NF1 mutation.
The NF1 gene product, neurofibromin, has a Ras-GTPase activating protein (Ras-GAP) related domain (reviewed in Kim and Tamanoi, 1998). Neurofibromin functions as a Ras-GAP in some cell types (Basu et al, 1992; DeClue et al, 1992; Kim et al, 1995; Largaespada et al, 1996). Ras-GTP levels are high in neurofibromas (Guha et al, 1996) suggesting that Ras is aberrantly regulated in one or more neurofibroma cell types. Neurofibromin may also have non-Ras functions (Johnson et al, 1993; Griesser et al, 1997; Guo et al, 1997; Kim et al, 1997). Neurofibromin accounts for about 15% of the Ras-GAP activity in fibroblasts (Kobayashi et al, 1993), but the relative contributions of Ras and non-Ras functions of neurofibromin in fibroblasts are not known.
To assess fibroblast abnormalities resulting from Nf1 mutation, we have used mice with a targeted mutation in the Nf1 gene (Brannan et al, 1994; Jacks et al, 1994). We previously showed that embryonic fibroblasts cultured from the Nf1-null mutant embryos hyperproliferate and fail to form perineurium in vitro (Rosenbaum et al, 1995), suggesting that loss of neurofibromin is sufficient to lead to fibroblast abnormalities. Neurons (Vogel et al, 1995), Schwann cells (Kim et al, 1995, 1997, and myeloid cells (Largaespada et al, 1996) derived from these mutant mice have been characterized and also exhibit abnormalities.
Because Nf1-null (Nf1−/−) mice die by embryonic day 14.5 and adult heterozygous (Nf1+/−) mice do not spontaneously develop neurofibromas or skin abnormalities, we challenged Nf1+/− animals using a well-characterized skin wound-healing model to assess fibroblast function in vivo (Greenhalgh et al, 1990; Brown et al, 1994). In particular, the wound-healing model facilitates the determination of the extent and control of fibroblast proliferation and collagen deposition (Mast, 1992), two key features of neurofibromas, in an environment-dependent manner. Several striking similarities exist between the wound healing and tumor environments including the presence of similar growth factors and cytokines, proliferation and migration of multiple cell types, angiogenesis, and matrix remodeling (Dvorak, 1986; Martin, 1997). Furthermore, one hypothesis based on patient observations suggests that neurofibroma formation may be triggered after injury (Riccardi, 1992). Indeed, neurofibromin expression is upregulated in fibroblasts during skin wound healing (Yla-Outinen et al, 1998). We show here that Nf1+/− mouse skin wounds have abnormal granulation tissue (containing mostly collagen) in dermal scars after excisional wounding. We also show that Nf1−/− fibroblast cultures have an increase in deposited collagen. Furthermore, fibroblasts in the granulation bed of Nf1+/− wounds proliferate beyond the normal time frame. In culture, Nf1+/− fibroblasts also have increased proliferation that is further potentiated by macrophage-conditioned media or epidermal growth factor (EGF). These data are consistent with Nf1 mutations in fibroblasts contributing to the enhanced collagen deposition and proliferation which are characteristic of neurofibromas, and indicate that neurofibromin plays a novel role in wound healing.
MEKK inhibitor (2′-amino-3′methoxyflavone) was purchased from Calbiochem (La Jolla, CA). Alpha-[32P]-dATP was purchased from Amersham Life Sciences (Arlington Heights, IL) and 2,3,4,5-[3H]-Proline was obtained from DuPont-NEN (Boston, MA). Farnesyltransferase inhibitor L744832 was a gift of N. Kohl, Merck Sharp Dohme Research Laboratory. All purified growth factors were purchased from R&D Systems (Minneapolis, MN).
Fibroblasts were isolated from embryonic day 12.5 mice as previously described (Rosenbaum et al, 1995). Cells were maintained in DMEM (Gibco Laboratories, Gaithersburg, MD) plus 10% (vol/vol) fetal bovine serum (Harlan, Indianapolis, IN) and 1% penicillin–streptomycin (Gibco/BRL) in 7.4% CO2 at 35°C. Cells from passages 1–5 were used for all in vitro experiments. For collagen deposition assays, fibroblasts were seeded in 35 mm dishes (Fisher, Pittsburgh, PA) at a density of 2 × 105 cells per well (cells reached coNfluence in 2 d). Sixteen to 20 hours after plating, the media was supplemented with 40 g per ml ascorbate (Sigma, St. Louis, MO) and was changed every third day.
V-H-ras expressing Nf1+/+ cells were obtained by iNfecting cells with retrovirus containing V1259T v-ras (DeClue et al, 1992). Cells were selected in 300 μg per ml of Gentamycin antibiotic (Gibco). V-H-ras expressing Nf1+/+ fibroblasts hyperproliferated and exhibited transformed morphology. Clones were pooled and used for experiments.
Male C57/BL6 Nf1+/+ (n = 26) and heterozygous Nf1+/− littermates (n = 40) were used for full thickness skin wounding experiments as previously described (Greenhalgh et al, 1990; Brown et al, 1994). Mice were 8–10 wk old at the time of wounding. All procedures were approved by the Institution of Animal Care and Use Committee at University of Cincinnati. Animals had been backcrossed seven to nine generations onto the C57/BL6 background. Wounding was performed under aseptic conditions. Mice were anesthetized by methoxyflurane inhalation (Mallinckrodt Veterinary, Mundelein, IL). Back skin was shaved and prepped with povidone-iodine solution and wiped with 70% isopropyl alcohol. A full-thickness wound measuring approximately 1.5 cm × 1.5 cm was created by excising the skin on the mid-back, including the panniculus carnosus. Some Nf1+/+ and Nf1+/− mice were given the full-thickness skin wound together with a single nick in the underlying spinalis muscle with a scalpel blade (2−3 mm in depth and 1 cm in length). Tincture of benzoin (Cumberland-Swan, Smyrna, TN) was applied to the perimeter of the wound and allowed to dry. The wound was covered with a transparent, nonabsorbent dressing (OpSite; Smith and Nephew Medical Ltd, Hull, U.K.). Animals were killed by sodium pentobarbitol overdose at 1, 2, 4, 6, 8, 10, or 12 wk postwounding. The entire wound, including a 5 mm margin of unwounded skin was then excised down to the fascia. The wounds were either placed in buffered 4% paraformaldehyde, cryoprotected in 20% sucrose/0.1 M phosphate-buffered solution, and frozen for cryostat sectioning, or fixed in formalin, processed for paraffin embedding, sectioned, and processed for routine staining or as designated below.
At least four random sections spanning the length of 4 wk old skin wounds with and without muscle injury were stained with biotinylated anti-proliferating cellular nuclear antigen (PCNA) (ZYMED, San Francisco, CA) and counterstained with hematoxylin according the manufacturer's directions. Five hundred total nuclei were analyzed per section in multiple sections of each wound and the percentage of PCNA labeled nuclei was calculated. Student's t-test analysis was performed to determine the statistical significance between the different groups.
The extent of wound closure of multiple groups over time were compared using one-between, one-within repeated measures ANOVA. Statistical procedures were performed using SAS software (version 6.04, SAS Institute, Cary, NC) as described (Greenhalgh et al, 1990; Brown et al, 1994).
Cell density measurements were obtained by counting the number of hematoxylin-stained nuclei in skin wounds (with muscle injury) in at least five different fields per section, in multiple sections per wound. Cell density was calculated using NIH Image (written by Wayne Rasband at the U.S. National Institute of Health and available from the Internet by anonymous ftp from zippy.nimh.nih.gov). The nonparametric Wilcoxon rank sum test was used to determine statistical significance.
Proliferation of cultured embryonic fibroblasts was analyzed as previously described, in DMEM + 10% FBS (Rosenbaum et al, 1995). Methods for harvesting and culturing mouse peritoneal macrophages were those of Griffin and Silverstein (1974) except that DMEM media + 10% FBS + 100 U per ml penicillin + 100 g per ml streptomycin was conditioned by 0.5 × 106 macrophages per ml for 48 h and stored at −70°C. Eight day fibroblast growth assays were performed using 50% macrophage-conditioned media; 200 μl macrophage-conditioned media and 200 μl of DMEM + 10% FBS were added to triplicate cultures in 24 well plates at days 1 and 4. EGF (R&D Systems) + FBS and EGF + 250 nM AG1478 (Calbiochem) (Daub et al, 1996) + FBS were added similarly. The day of plating was designated day 0.
Quantitation of accumulated collagen matrix deposited by fibroblasts over a 14 d period was done by performing colorimetric assay of total hydroxyproline content as described (Berg, 1982), and then normalized to total amount of protein soluble in 0.1% sodium dodecyl sulfate, 1% triton X-100 (Eldridge et al, 1987).
To test the hypothesis that loss of neurofibromin function causes fibroblast dysfunction, we challenged Nf1+/− fibroblasts in vivo using a full-thickness skin wound-healing model. Before wounding, Nf1+/+ and Nf1+/− skin appeared normal by gross examination and in histology sections (data not shown). The overall rate of reepithelialization over 0–18 d in Nf1+/+ (n = 12) and Nf1+/− (n = 15) murine wounds did not differ significantly (data not shown). Histologic analysis of wounded skin was used to determine key spatial and temporal features of the healing process. Random sections spanning the wound were stained with either hematoxylin and eosin or Gomori's trichrome. Sections were examined for gross changes in the degree of cellular infiltration, granulation tissue formation, vascularity, re-epithelialization, and collagen deposition. We did not find any obvious differences in the degree of inflammatory cell infiltration, angiogenesis, or re-epithelialization (with the exceptions noted below); subtle changes cannot be excluded. In contrast to Nf1+/+ wounds (Fig 1A), many Nf1+/− wounds, however, had a profoundly abnormal pattern of granulation tissue distribution in which the granulation bed of the regenerated dermis invaded the underlying muscle fibers (Fig 1B black arrowheads). This unusual phenotype was not observed in any of the Nf1+/+ wounds examined in this study, nor in any of the more than 100 mice analyzed previously after excisional wounding (Greenhalgh et al, 1990; Brown et al, 1994).
To assess the frequency and timing of the abnormality, wounds were examined between 1 and 12 wk postwounding. Wounds were scored as abnormal if the pattern of fibrotic tissue distribution was unorganized and penetrated underlying muscle fibers in at least one section per wound. Sectioning of half the wound beds of a total of 27 Nf1+/− mice from various time points showed the granulation bed penetrating the underlying muscle in 50% of the wounds. This phenotype was evident as early as 2 wk postwounding. To determine if abnormal histology was present in all wounds, complete 4 wk old wound beds of 13 additional Nf1+/− mice were serially sectioned and examined. Sixty per cent of wound beds were scored as histologically abnormal. We conclude that a majority of Nf1+/− wounds have an abnormal pattern of granulation tissue. To examine the possibility that accidental wounding of muscle contributes to the wound-related phenotype observed in Nf1+/− mice, we performed full-thickness skin wounding and nicked the underlying muscle of Nf1+/+ and Nf1+/− mice. Wounds were harvested at 4 wk after wounding and serially sectioned. All of the Nf1+/+ (three of three) mice showed complete, normal skin regeneration (Fig 1C); however, three of five of Nf1+/− wound beds had an abnormal pattern of matrix in the granulation bed (Fig 1D). The collagen matrix in the Nf1+/− mouse wound beds penetrated into the muscle layer and encircled muscle fascicles (Fig 1D). We conclude that muscle nicking is not sufficient to elicit in (Nf1+/+) mice or exacerbate in the (Nf1+/− mice) the granulation bed phenotype of Nf1+/− wounds.
Using higher magnification, the dermal scar consists of elongated nuclei in the collagen-rich matrix of both Nf1+/+ (Fig 1E) and Nf1+/− (Fig 1F) scars. Most of these cells are believed to be fibroblasts, which produce the surrounding matrix. Collagenous matrix penetrating the muscles in Nf1+/− wounds tended to be less compacted (Fig 1G) than the underlying regenerated dermis of Nf1+/+ scars, but also contained elongated fibroblast nuclei.
In addition to the excessive granulation tissue in the Nf1+/− wounded skin, rarer phenotypes were detected. At 4 wk postwounding, failure to re-epithelialize was observed in three of 40 Nf1+/− mice versus none of 26 Nf1+/+ mice. In one Nf1+/− mouse, the collagen matrix erupted above the skin layer (Fig 2A). In two Nf1+/− mice, the wound had clear signs of delayed wound contracture and delayed healing (Fig 2B, C), excessive angiogenesis (Fig 2C), as well as failure to re-epithelialize completely (Fig 2C).
In order to quantitate the amount of granulation tissue, the area and the depth of the regenerated dermis were analyzed in paraffin sections from 12 Nf1+/− and six Nf1+/+ mice 4 wk after wounding (Fig 3). The area of the regenerated dermis was measured using Metamorph software in four to seven different sections at intervals spanning each wound, by measuring from the neoepithelium to the muscle layer in height, and to the normal skin interface on either side of the wound in width. The depth of each wound was measured by calculating the distance from the neoepithelium to the muscle layer in the same sections. In comparison with Nf1+/+ wounds, most but not all Nf1+/− wounds showed an increase in the area and depth of the regenerated dermal scar even though sections without overt granulation bed distortion were included in the analysis (Fig 3). The depth and the area of the dermal scar in the Nf1+/− group was significantly larger than the Nf1+/+ wounds (p < 0.001 for depth measurements and p < 0.05 for area measurements). Statistical analysis performed on both the groups even after excluding the two animals which had failed to re-epithelialize (Fig 3, right two animals) also showed that Nf1+/− wounds are significantly different from Nf1+/+ wounds (p < 0.05 for depth measurements). The area and depth of the regenerated dermal scar in the Nf1+/− skin wounds with muscle injury (Fig 3A, B) was not significantly different from the Nf1+/− skin wounds without muscle injury (Fig 3A, B).
The abnormal amount and pattern of granulation tissue observed in Nf1+/− wounds suggested that increased numbers of fibroblasts might be present in the Nf1+/− wound beds. To begin to test this possibility, Nf1+/+ and Nf1+/− wound sections were stained with a biotinylated anti-PCNA antibody, which labels proliferating cells. Immunoreactivity was visualized as a brown precipitate after DAB reaction. Fibroblast nuclei were identified based on their morphology and location in the granulation bed. As compared with Nf1+/+ wounds (Fig 4A), a significantly higher percentage of fibroblasts in the granulation bed of most Nf1+/− (n = 9/11) wounds were labeled by anti-PCNA (Fig 4B). Quantitation of the percentage fibroblast nuclei labeled by the antibody revealed that wounds from Nf1+/− mice had a significantly higher percentage of PCNA positive cells than those from Nf1+/+ mice (p < 0.0001) (Fig 4C). Heterozygous mouse skin wounds in which the muscle was also injured had an even higher percentage of PCNA positive cells than Nf1+/− skin wounds alone (Fig 4C). Cell density measurements in the dermal scars of skin wounds with muscle injury also revealed a small but significant difference between the Nf1+/+ and Nf1+/− wounds (p < 0.0001) (data not shown) consistent with accumulation of fibroblasts within wound beds.
To determine if Nf1+/− fibroblasts have inherent abnormal regulation of proliferation, in vitro growth assays were performed. Over 8 d, in media with serum, Nf1+/− and Nf1−/−embryonic fibroblasts proliferated significantly more than Nf1+/+ fibroblasts (Fig 5). We tested if factors found in wounds could augment proliferation of mutant fibroblasts. Macrophages are the major source of cytokines that regulate fibroblast function in wounds (Diegelman et al, 1981). Therefore, conditioned media of activated macrophages was added to fibroblast cultures. Nf1+/+ fibroblasts were not affected by the addition of macrophage-conditioned media to the culture medium (Fig 5) or by the addition of fibroblast-conditioned medium (not shown). Nf1+/− fibroblasts were also unaffected by the addition of fibroblast-conditioned medium (not shown). In contrast, Nf1+/− fibroblasts from four of seven embryos tested showed augmentation of proliferation in response to activated macrophage-conditioned media (Fig 5). Fibroblasts from the remaining embryos may already have reached maximal response in the assay. In an attempt to identify the growth factors that contribute to the macrophage-conditioned media activity, individual purified growth factors that are secreted by macrophages were used in the proliferation assay. Basic fibroblast growth factor (0.3 or 4 ng per ml), transforming growth factor β1 and β2 (5 and 165 pg per ml), platelet-derived growth factor BB (5 and 10 ng per ml), and midkine (5 μg per ml) did not potentiate proliferation of Nf1+/+ and Nf1+/− fibroblasts (data not shown). EGF (0.3–3 ng per ml) mimicked the macrophage-conditioned media effect on Nf1+/− fibroblasts proliferation (Fig 5B). EGF stimulated proliferation was blocked by the EGF receptor kinase inhibitor, tyrphostin AG1478 (Fig 5B). AG1478 did not inhibit proliferation of fibroblasts in normal growth conditions (Fig 5B). Addition of AG1478 to macrophage-conditioned media did not alter its activity, suggesting that EGF is not the predominant factor in the macrophage-conditioned medium that potentiated Nf1+/− fibroblast proliferation. Thus, mutation in a single Nf1 allele causes abnormal fibroblast proliferation in response to specific environmental signals including macrophage produced growth factor(s) and EGF.
Excess granulation tissue observed after in vivo wounding suggested that Nf1+/− fibroblasts, the chief producers of collagen matrix in the wound bed (Mast, 1992), might produce abnormal amounts of collagen, as well as showing increased proliferation. Fibroblasts of all three genotypes were tested for their ability to deposit collagen in vitro. We tested the amount of collagen deposited by Nf1+/+ (13 individual embryos) and Nf1+/− (11 individual embryos) fibroblast cultures and failed to detect a significant difference between the two genotypes (Fig 6A). Next we investigated if complete loss of neurofibromin altered the amount of collagen deposited. Nf1−/− fibroblast cultures (13 individual embryos) had 2–2.5-fold higher deposited collagen (p < 0.0001) in comparison with Nf1+/+ fibroblasts (Fig 6A). Fibroblasts from three individual Nf1−/− embryos tested did not demonstrate an increase in deposited collagen (data not shown). The robust alteration in amount of collagen deposited by Nf1−/− fibroblasts coupled with the increased collagen observed in Nf1+/− mice after skin wounding prompted analysis of the mechanism underlying alteration in collagen deposition.
Ras activation decreases collagen transcription in fibroblast cell lines (Hatamochi et al, 1991; Kenyon et al, 1991; Slack et al, 1992). It therefore seemed unlikely that Ras activation caused the increased collagen deposition observed in Nf1−/− fibroblasts, even though Nf1 regulates Ras activation in some cells. Consistent with the literature for fibroblast cell lines, v-H-ras iNfected primary wild-type fibroblasts deposited extremely low levels of collagen (Fig 6B). We also tested if Ras pathway inhibitors could inhibit the enhanced collagen deposition in Nf1−/− cultures. MEKK Inhibitor (MEKKI), which inhibits the Raf-MAPK arm of the Ras pathway was used (Dudley et al, 1995) and farnesyl protein transferase inhibitor L744832 at up to 10 μM also did not alter collagen deposition in Nf1+/+ or Nf1−/− cultures (data not shown). These data are consistent with the idea that increases in collagen deposited by Nf1−/− fibroblasts are not mediated through Ras.
Our in vivo studies reveal a role for Nf1 in wound healing. Skin wound healing is a complex process consisting of inflammatory, proliferative and maturation phases that occur in a specific sequence. Several cell types contribute to the repair process in a coordinated manner (reviewed in Clark, 1993; Gailit and Clark, 1994). In the first 7 d following an excisional skin wound, a fibrin clot provides a provisional matrix for inflammatory cell invasion and creates the environment for the subsequent proliferative phase of multiple cell types. In Nf1+/− mice, this early phase of wound repair appears to proceed normally. Seven to 21 d after an excisional wound, epithelialization is completed, fibroblasts proliferate, and collagen is synthesized in the granulation bed. In the majority of Nf1+/− wounds we analyzed, epithelialization occurred on schedule; in only three of 40 wounds was epithelialization delayed, when disrupted by a large dermal scar. Indeed, an increase in dermal scar area and depth was observed in most Nf1+/− wounds, indicating an abnormal increase in the amount and persistence of granulation tissue in the Nf1+/− wound beds. Dermal connective tissue components may differ in the skin in Nf1+/+ and Nf1+/− mice and contribute to these abnormalities, but some putative changes have not been studied. Many Nf1+/− wounds also had a highly abnormal pattern of granulation tissue in the wound bed, in which the granulation bed actually penetrated underlying muscle bundles. Thus, a specific in vivo defect in granulation tissue formation has been identified in Nf1+/− mice.
Based on our in vivo and in vitro studies, we suggest that abnormal fibroblast proliferation in wound beds is due to abnormal fibroblast response to wound growth factors and cytokines. Nf1+/− wounds showed an abnormally high percentage of PCNA-positive fibroblasts; fibroblast proliferation was exacerbated by muscle injury. Potent mitogens for fibroblasts are released by injured skeletal muscle (reviewed in Goldring and Goldring, 1991; Husmann et al, 1996). Macrophage secreted factors are the chief effectors of fibroblast behavior in the wound bed (Diegelman et al, 1981). Indeed, many of the same factors released by injured muscle are also released from activated macrophages. Activated macrophages in vitro contain a similar profile of factors and cytokines as the wound bed (Rappole et al, 1988; Grotendorst et al, 1992). Nf1+/− fibroblasts in vitro showed increased proliferation in the presence of macrophage-conditioned media or recombinant EGF. Because an EGFR antagonist failed to reverse effects of macrophage-conditioned media, additional effective factors must be contained in the macrophage-conditioned media. Neurofibromin expression is upregulated in fibroblasts after human skin wounding, and by exposure of human fibroblasts to PDGF and transforming growth factor-β (Ylä-Outinen et al, 1998). In mouse fibroblasts with Nf1 mutations, mitogenic effects subsequent to injury may be prolonged because neurofibromin upregulation is missing.
We postulate that wound factors also contribute to collagen deposition by Nf1+/− fibroblasts. Increased numbers of proliferating fibroblasts in vivo may cause the excess in granulation tissue in Nf1+/− wound beds, or individual Nf1+/− fibroblasts may synthesize increased amounts of collagen in the wound environment. Our in vitro studies show a 2–2.5-fold increase in deposited collagen in Nf1−/− fibroblast cultures. Nf1+/− fibroblasts failed to show abnormalities in collagen deposition in culture. Fibroblasts from human neurofibromas have increased collagen mRNA and secretion (Peltonen et al, 1981, 1986; Uitto et al, 1986; Sasaki et al, 1992), but the genetic status of NF1 gene in neurofibroma fibroblasts is not known. Preliminary in vitro studies have failed to demonstrate alterations in collagen deposition by Nf1+/− fibroblasts in the presence of macrophage-conditioned media (data not shown). Proliferation and collagen deposition phenotypes may be regulated differently or may require different concentrations of wound factors.
Our data suggest that fibroblasts use Ras-independent NF1 pathways for abnormal collagen deposition, because Nf1+/+ fibroblasts expressing activated v-H-ras did not mimic the collagen deposition phenotype of the Nf1−/− fibroblast in vitro. In contrast, Schwann cells expressing the same v-H-Ras allele mimicked Nf1-mutant Schwann cell phenotypes (Kim et al, 1995). Indeed, primary mouse fibroblasts expressing activated Ras diminished all tested parameters of collagen manufacture and deposition, consistent with previous studies in fibroblast cell lines (Hatamochi et al, 1991; Kenyon et al, 1991; Slack et al, 1992). Non-Ras functions for NF1 in fibroblasts are consistent with recent data demonstrating that Nf1 mutant fibroblasts do not show elevated Ras-GTP (Sherman, L. Atit, R. Cox, A. & Ratner, N, submitted). The mechanism underlying the increased collagen deposition by Nf1−/− fibroblasts remains unknown. Preliminary examination of key steps in collagen biosynthesis did not reveal detectable differences in collagen mRNA levels, secretion, processing, or gelatinase activity of fibroblasts from various genotypes (Atit et al, unpublished observations).
Increases in wound depth and area, and abnormal pattern of granulation tissue formation were evident in 60% of wounds from Nf1+/− mice. In vitro abnormalities in collagen deposition also occurred in cells derived from most, but not all, embryos. Because the environment remains constant in in vitro assays, yet responses differed, signals present at different levels in the environment of individual mice may not completely explain variable expressivity and penetrance. Mice used for our studies had been backcrossed seven to nine generations on to the C57BL/6 background, so genetic variation is unlikely to account for the observed differences among mice. Our in vivo studies were all performed on male mice so sex-specific differences are also unlikely to explain differences among animals. It is possible that neurofibromin acts in cooperation with one or more molecules to produce the collagen-deposition phenotype. This may occur through a non-NF1 alternate pathway (controlled by one or more genes) that is stochastically used to compensate for the mutation (e.g., see Horan et al, 1994). Alternatively, the single active Nf1 allele may not generate enough functional protein to rise above a threshold needed to achieve an appropriate biologic response in all target cells (e.g., see Dunn et al, 1997).
Fibroblasts from Nf1+/− mice tested proliferated more than fibroblasts from wild-type mice, both in vivo after wounding and in vitro. Rosenbaum et al (1995) found no differences in proliferation of Nf1+/− fibroblasts as compared with wild-type cells, whereas Nf1−/− fibroblasts did hyperproliferate. A difference between the two sets of experiments is the number of backcrosses onto the C57BL/6 background. Recessive modifier(s) in C57BL/6 may be required together with decreased Nf1 gene dosage to evidence this difference among animals. It is believed that modifier genes influence expressivity of human NF1 (Easton et al, 1993).
Our examination of the wound healing process in Nf1+/− mouse skin suggests that abnormalities in skin fibroblasts could contribute to skin abnormalities in NF1 patients. These include hyperpigmentation (café-au-lait macules, axillary and inguinal freckling) (Friedman and Birch, 1997) and localized areas of skin atrophy and hypoplasia (Norris et al, 1985). Neurofibromin is expressed in developing and adult human skin (Malhotra and Ratner, 1994; Hermonen et al, 1995) and is upregulated in fibroblasts after wounding (Ylä-Outinen et al, 1998). The healing process in human NF1 skin has not yet been studied.
NF1 patients develop neurofibromas, benign tumors associated with peripheral nerves that contain Schwann cells, fibroblasts, and collagen-rich matrix. It has been suggested that wounding could facilitate the formation of neurofibromas (Riccardi, 1992). We have shown that injury promotes, in mice, specific features of neurofibromas found in human patients including fibroblast hyperplasia and collagen accumulation. Our data support the view that trauma and dysfunction of Nf1 mutant fibroblasts could contribute to human neurofibroma formation.
We thank J. Florer and B. Ling for technical assistance, T. Rosenbaum and D. Brown for initiating experiments to measure wound closure, L. James for statistical analysis, and S. Newman for teaching us how to collect activated macrophage medium. T. Suh, L.Sherman, and T. Bugge gave helpful comments on the manuscript. RA is the recipient of the Univeristy of Cincinnati Dean's Distinguished Dissertation Award. This work was supported by NIH-R01-NS28840 and a grant from the DOD to N.R..