| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Regular Article |
From the Divisions of Neurology*
and
Genetics,
the Department of Pediatrics, and
the Department of Pathology and Laboratory
Medicine,
University of Florida Brain
Institute and College of Medicine, Gainesville, Florida
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Unlike schwannomas, which consist predominantly of Schwann cells (SCs), neurofibromas show marked cellular heterogeneity. Nevertheless, SCs are the major cell type amplified in neurofibromas and typically comprise 40 to 80% of the tumor cells. Additionally, there is a substantial population of interspersed fibroblastic or perineurial cells, along with various vascular and inflammatory elements embedded in an extensive extracellular matrix.3 Because of this cellular heterogeneity, the histogenesis of neurofibromas has been controversial. Although there is increasing evidence for a SC origin, some studies suggest the contribution of SCs and fibroblastic cells.4-6 On the other hand, an emerging view proposes that all of the major cellular elements of neurofibromas are of SC lineage and that the fibroblastic/perineurial-like elements may be immature or variant SCs.7-10
NF1 is caused by disruptive mutations in the NF1 gene, which encodes the GAP-related protein neurofibromin. Thus, all cells in an NF1 individual are initially haplo-insufficient for neurofibromin activity(s). However, consistent with the tumor suppressor gene two-hit model, it seems that loss of function of the remaining NF1 allele is associated with neurofibroma formation, as first observed by our lab as loss of heterozygosity.11 Although there are several approaches to animal models of NF1, presently none exist in which neurofibromas can be readily induced using defined human cell populations deficient in neurofibromin. Gene targeting has been used to construct mouse strains harboring mutations in the Nf1 gene.12,13 The mouse knockout model, similar to the human NF1 condition, involves only a single constitutional mutation and homozygous mice (Nf1-/-) die during gestation. Despite the high level of conservation between mouse and human neurofibromin, it is clear that Nf1 knockout mice are not prone to the formation of neurofibromas. Thus, the inadequacy of the heterozygous mouse model may be attributed to a low mutation rate of the remaining wild-type Nf1 allele within the relatively short murine life span. Original studies by Martuza and co-workers14 demonstrated the growth of minced human neurofibromas in the subrenal capsule and sciatic nerve of immunodeficient mice that retained their morphological features and genomic identities. Thus, the use of defined neurofibroma cell populations in animal models will greatly enhance efforts to understand the histogenesis of neurofibromas.
Neurofibroma SCs have invasive and angiogenic properties, suggesting that these are genetically altered cells with tumorigenic properties.15,16 Additionally, cytogenetic studies show that plexiform neurofibromas harbor genetically abnormal SCs and strongly implicate these cells as the central component in the development of these potentially progressive tumors.17 Recently, somatic loss of heterozygosity was found in SCs, but not fibroblasts cultured from a neurofibroma, suggesting that genetic alterations of the NF1 gene in SCs are involved in the development of neurofibromas.18 In a more comprehensive study, Rutkowski and co-workers19 further demonstrated that neurofibroma-derived SCs typically lacked NF1 mRNA whereas fibroblasts isolated from neurofibromas expressed the NF1 transcript. In the present study, SCs subcultured from numerous neurofibromas were examined for neurofibromin expression and tumorigenic properties in vitro and after intraneural engraftment. Our findings strongly implicate neurofibromin-deficient SCs in the histogenesis of at least a subset of neurofibromas.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Normal Adult Nerve SCs
All specimens included in this study were obtained in accordance with protocols approved by the University of Florida Institutional Review Board. Human SCs were isolated from normal adult sural nerves by modifications of methods described previously.20,21 Briefly, segmented nerve fascicles were cultured for 10 days in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum, 5% calf serum, 2 µm forskolin, 25 ng/ml human recombinant glial growth factor-2 (GGF-2), and antibiotics (expansion medium). The tissue was then dissociated for 18 hours in medium 15% calf serum, 1.25 U/ml Dispase (Collaborative Research Inc., Bedford, MA), 300 U/ml collagenase (type XI; Sigma Chemical Co., St. Louis, MO) and antibiotics. The digested tissue was dispersed by trituration, passed through a 30-µm mesh nylon screen, and centrifuged (200 x g, 10 minutes). The cell pellet was resuspended (50 segments/2 ml) in medium containing N2 supplements22 and 2 ml of the cell suspension was spread across the surface of a 75-cm2 flask precoated sequentially with polyornithine (0.1 mg/ml) and laminin (10 µg/ml) (prepared as described by Muir23 ). After a 6-hour incubation, the medium was supplemented by the gentle addition of expansion medium (10 ml). The cultures were grown to near confluency and the SCs were isolated by differential detachment using mild trypsinization and gentle shaking. The highly enriched cultures were expanded in expansion medium in dishes coated with laminin only. All cultures were withdrawn from treatment with forskolin and GGF-2 before storage or use.
Neurofibroma SCs
All patients met recognized diagnostic criteria for
NF124
and tumor specimens were characterized as
neurofibromas by histopathological study. Patient ages ranged from 4 to
69 years (the majority were young to middle-aged adults) and the
reasons for surgery included cosmetic, functional deficit, and tissue
donation on NF1-related fatality. These factors were unrelated to the
tissue culture outcome and phenotype. Any capsular material was removed
and viable tumor isolated from surgically resected neurofibromas. Tumor
tissue (1 cm2) was finely minced and incubated at
37°C overnight in 10 ml of L-15 medium containing 15% calf serum,
1.25 U/ml Dispase, 300 U/ml collagenase, and antibiotics. The tissue
was dispersed by trituration and strained through a 30-µm mesh nylon
screen. The filtrate was diluted with L-15 and centrifuged (400 x
g, 5 minutes). The cell pellet was resuspended in
Dulbecco’s modified Eagle’s medium supplemented with 10% fetal
bovine serum, 5% calf serum, and antibiotics (standard medium), and
cells were seeded into tissue culture flasks
(
106
cells/75 cm2. After
4 days, cultures were detached with trypsin/ethylenediaminetetraacetic
acid and passaged 1:4. Half of the passaged cells were grown in
standard culture conditions; these were later harvested and stored in
liquid nitrogen. The other half of the cells were seeded in flasks
precoated with laminin (10 µg/ml) and grown in standard medium
containing GGF-2 (25 ng/ml). GGF-2 treatment caused rapid proliferation
of Schwann-like cells (SLCs). During subsequent passage, SLCs were
enriched further by differential detachment using mild trypsinization
and shaking. For the specified cultures, the combination of
preferential laminin attachment, differential detachment, and selective
mitogen treatment with GGF-2 yielded highly enriched (95 to 99.5%) SLC
cultures within 3 to 4 passages. All tumorigenic and protein expression
assays were performed using cultures at passage 3 to 4 that had been
withdrawn from GGF-2 for at least 2 days. GGF-2 was generously provided
by M. Marchionni (Cambridge Neuroscience, Cambridge, MA).
Anchorage and Serum Requirements
The growth of SLC-enriched cultures was assessed in serum-free and unattached conditions. Early (passage 2 to 4) cultures were maintained in standard medium and detached from the culture dish with 0.5 mmol/L ethylenediaminetetraacetic acid in phosphate-buffered saline (PBS). To examine anchorage dependency, cells were seeded in poly(HEMA)-coated culture wells at a density of 105 cells/well in Joklik medium (to minimize cell aggregation) supplemented with 10% serum. To examine serum dependency, cells were seeded in culture wells coated with laminin-1 (10 µg/ml) at a density of 105 cells/well and grown in serum-free N2 medium supplemented with 1% heat-inactivated bovine serum albumin. Cell viability was assessed at 0, 24, and 72 hours using a Trypan blue dye-exclusion assay. Counting chambers of a hemocytometer were filled with the cell suspension and viable cells (dye-excluding) as well as nonviable cells (dye-absorbing) were counted.
Growth in Soft Agarose
Anchorage-independent colony formation was determined by growing cells in soft agarose as described by Neugut and Weinstein.25 A thin base-layer of 0.9% agarose was allowed to solidify for 1 hour in culture wells. A single-cell suspension in standard medium containing 0.4% agarose was layered over the solid agarose base and allowed to solidify. The cultures were grown for 2 to 8 weeks and viable colonies consisting of >25 cells were scored by phase-contrast microscopy. Percent colony formation was calculated as: (number of viable colonies/total viable cells seeded) x 100%.
Subcutaneous Engraftment
All animal procedures were performed in accordance with approved IACUC protocols. NF1 SC cultures were tested for their ability to form tumors after subcutaneous injection in immunodeficient nude mice. Cells grown on laminin in medium containing GGF-2 were harvested by trypsinization and resuspended in Hanks’ balanced salt solution. Numerous subcutaneous injections were made using 2 x 106 cells/site. Mice were examined for development of palpable tumors for at least 3 months. Thereafter, animals were euthanized and the injection sites were surgically exposed and examined for signs of tumor formation. Because there were no signs of tumor growth no histology was performed.
Nerve Engraftment
Human neurofibroma-derived SC cultures (Table 1)
from cryopreserved stocks were grown
on laminin for 4 days in medium supplemented with GGF-2. Dissociated
cells were collected, rinsed thoroughly, and resuspended as a dense
slurry in Hanks’ solution. Young adult scid mice were
anesthetized and sciatic nerves of both legs were exposed at mid-thigh.
A cell suspension (5 x 105
in 4 µl) was
gradually injected intrafascicularly in both nerves through a fine
needle (35 gauge) attached to a Hamilton syringe. The site was closed
in layers with sutures and the revived mouse returned to specific
pathogen-free housing. At 1 to 8 weeks after implantation, the animals
were sacrificed under anesthesia and the nerves
(n = 4 for each culture) were removed and
fixed by immersion in 4% paraformaldehyde. Nerve segments were
embedded in paraffin and sectioned for immunohistochemical staining.
|
Neurofibroma Cultures
Monolayer cultures were examined for immunoreactivity with antibodies to the SC antigens S-100 (DAKO, Carpinteria, CA) (1/300) and the low-affinity nerve growth factor receptor (p75) (4 µg/ml, hybridoma 200-3-G6-4; American Tissue Culture Collection, Rockville, MD). Cultures grown on laminin-coated chamber slides were fixed with 2% paraformaldehyde in 0.1 mol/L phosphate buffer (pH 7.2) for 20 minutes, then washed with PBS containing 0.5% Triton X-100. Nonspecific antibody binding was blocked with PBS containing 0.1% Triton and 10% normal serum (blocking buffer) for 1 hour. Primary antibodies were diluted in blocking buffer and applied to wells for 2 to 4 hours at 37°C. Bound antibodies were labeled with peroxidase-conjugated secondary antibodies for 1 hour at 37°C and chromogenic development was accomplished with 3,3'-diaminobenzidine-(HCl)4 (0.05%) and hydrogen peroxide (0.03%) in PBS. Bromodeoxyuridine (BrdU) incorporation in vitro and immunolabeling of BrdU-DNA were performed as described previously.26
Nerve Grafts
Sciatic nerves engrafted with neurofibroma-derived SC cultures
were fixed by immersion in 4% paraformaldehyde in 0.1 mol/L phosphate
buffer (pH 7.2), sectioned in paraffin, and stained with hematoxylin
and eosin for routine light microscopic examination. To identify
transplanted human neurofibroma-derived SCs, nerve sections were
immunostained with polyclonal anti-GST
(DAKO) (1/100) (a
human-specific antiserum to the ubiquitous cellular protein,
glutathione S-transferase) and a monoclonal antibody to p75
(4 µg/ml, hybridoma 200-3-G6-4) (a primate-specific antibody to the
low-affinity nerve growth factor receptor). Deparaffinized sections
were pretreated with methanol containing 1% hydrogen peroxide for 30
minutes to quench endogenous peroxidase activity. Nonspecific antibody
binding was blocked with 10% normal serum in PBS containing 0.3%
Triton X-100 for 60 minutes at 37°C. Primary antibodies were diluted
in blocking buffer and applied to sections overnight at 4°C. Bound
antibodies were labeled with biotinylated secondary antibodies for 4
hours at 37°C followed by the avidin-biotin-peroxidase reagent (DAKO)
for 2 hours. Chromogenic development was accomplished with
3,3'-diaminobenzidine-(HC1)4 (0.05%) and
hydrogen peroxide (0.03%) in PBS.
Neurofibroma Tissue Specimens
Portions of the primary tumor used for cell culture were fixed by immersion in 4% paraformaldehyde in 0.1 mol/L phosphate buffer (pH 7.2), sectioned in paraffin, and stained with hematoxylin and eosin for routine light microscopic examination. Sections were immunostained for neurofibromin with the NF1GRP(N) antibody (1 µg/ml) (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) raised against a peptide corresponding to residues 509 to 528 of the predicted NF1 gene product. The specificity of antibody to this neurofibromin peptide was reported previously.27 Serial sections were immunolabeled with polyclonal anti-S-100 (1:300, DAKO). Immunoperoxidase labeling with the avidin-biotin-peroxidase reagent was performed as described above, except to enhance neurofibromin staining, antigen retrieval was achieved by pretreating sections in 0.1% trypsin for 20 minutes at 37°C. Immunostained sections were lightly counterstained with hematoxylin. Negative controls used no primary antibody. Additionally, for the NF1GRP(N) antibody, preadsorption with a 10-fold molar excess of peptide antigen (SC-67P, Santa Cruz) was used to achieve complete blocking of neurofibromin immunoreactivity.
Western Immunoblotting
SC cultures were scraped from dishes and cell pellets were homogenized in ice-cold extraction buffer consisting of 50 mmol/L Tris-HCl (pH 7.4), 250 mmol/L NaCl, 1% Nonidet P-40, 0.25% sodium deoxycholate, and complete protease inhibitors (Boehringer-Mannheim, Indianapolis, IN). The soluble fraction was collected by centrifugation (10,000 x g, 20 minutes) and then was made 2 mol/L in urea. The extract was concentrated and fractionated by ultrafiltration using a 100-kd cut-off membrane. Total protein content of the high molecular mass retentate was determined using Bradford Reagent (Bio-Rad Laboratories, Hercules, CA). Samples were mixed with sodium dodecyl sulfate-containing electrophoresis sample buffer containing 2 mol/L urea and 5% 2-ME and then heated to 80°C for 2 hours. Samples (100 µg) were electrophoresed on 4 to 15% polyacrylamide gels and electroblotted to nitrocellulose sheets in transfer buffer containing 0.1% sodium dodecyl sulfate. Blots were rinsed in water and fixed in 25% isopropanol/10% acetic acid. Nitrocellulose sheets were washed with 0.05 mol/L Tris-HCl (pH 7.4) containing 1.5% NaCl and 0.1% Triton X-100 and then blocked in the same buffer with the addition of 5% dry milk (blocking buffer). The blots were incubated for 2 hours with anti-NF1GRP(N) antibody (1 µg/ml) in blocking buffer. Bound antibody was detected by peroxidase conjugated swine anti-rabbit IgG (affinity purified, DAKO) diluted 1/2,000 in blocking buffer. Immunoreactive bands were developed by chemiluminescent methods (Pierce Chemical Co., Rockford, IL) according to the manufacturer’s instructions. Relative molecular mass was determined using prestained markers including myosin (205 kd). Control samples were similarly processed from cell pellets obtained from normal human nerve SC cultures and SC cultures derived from embryonic homozygous Nf1 knockout mice.12
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
SCs do not proliferate in response to standard serum-supplemented
medium and, until recently, methods for isolating normal human SCs were
mostly unreliable. Based on recent advances,20,21
we
investigated various means to enrich and expand SCs from adult human
nerve segments. Successful enrichment of SC cultures and depletion of
fibroblastic cells was readily achieved by treatment with the SC
mitogens GGF-2 and forskolin combined with differential cell detachment
and preferential growth on a laminin-coated substratum. After 3 to 4
passages under these conditions, hundreds of millions of SCs were
obtained nearly free of fibroblast contamination from several
centimeters of adult tibial nerve. These cultures were homogeneous and
contained highly elongated SCs that stained intensely for S-100 and
p75, recognized markers for cells committed to a SC
lineage.28
The specificity of S-100 as a marker for
cultured human SCs is shown in Figure 3A
. After enrichment and
expansion, SC division rapidly decreased in the absence of GGF-2 and
forskolin. Growth on a laminin substratum was required at all stages to
improve attachment and to minimize cell attrition. SC expansion was
limited to 
10 population doublings before senescence regardless of
mitogen stimulation, confirming the earlier findings by Rutkowski and
colleagues.29
|
Our goal was the enrichment and characterization of SLCs from
dermal and plexiform neurofibromas. Monolayer cultures of neurofibromas
were initiated by enzyme dissociation under standard culture
conditions. The most frequent primary culture obtained from dermal and
plexiform neurofibromas contained a sublayer of fibroblast-like cells
(FLCs) admixed with numerous (20 to 60%) spindle-shaped SLCs (Figure 1)
. Numerous procedures to enrich and
expand SLCs from the primary cultures were tested. It is notable that
enrichment of SLCs was not improved by combined treatment with GGF-2
and agents that elevate cAMP. In particular, forskolin caused
considerable heterogeneity in the SC population and hampered the
development of the SLC cultures described below. Thus, unlike normal SC
cultures, neurofibroma cultures were treated only with GGF-2 and were
not exposed to forskolin.
|
A first type of neurofibroma culture (type-1) was distinguished by a poor response to the SC enrichment treatment with GGF-2 and laminin. These cultures (from dermal and plexiform tumors) were rapidly dominated by FLCs that often displayed abnormal growth characteristics (eg, rapid and protracted proliferation and loss of density-limited growth). The SLCs (immunopositive for S-100 and p75) failed to proliferate sufficiently and became increasingly diluted with repeated passage. Type-1 cultures accounted for 14 of 24 dermal and eight of 16 plexiform cultures, were deemed intractable to SLC enrichment, and were excluded from the analyses to follow.
A second subset, type-2 neurofibroma cultures, were amenable to
enrichment of the SLCs. In most of these cultures the SLCs proliferated
rapidly in the presence of GGF-2 and soon formed confluent islands on
top of underlying FLCs (Figure 1A)
. Cultured on native tissue culture
plastic, the SLCs attached exclusively to the FLCs. Despite the limited
surface area of the underlying the FLCs, proliferation by the SLCs
continued in the presence of GGF-2. As a result, the SLCs formed dense
cellular masses or culture tumors (Figures 1B and 2A)
, indicating the absence of
contact inhibition (density-limited growth) by neurofibroma SLCs.
Immunolabeled sections of these culture tumors showed an extensive
laminin-rich extracellular matrix (Figure 2B)
. These observations
potentially relate to the adhesive mechanisms involved in the growth
and development of neurofibromas.30
|
. Once outnumbered by the
mitogen-driven SLCs, the FLCs were effectively diminished by
differential detachment. Thereafter, the SLC population was readily
enriched and expanded in large numbers within 3 to 4 passages (Figure 1D)
. Type-2 SLC cultures stained for S-100 and p75 and generally
resembled highly spindled SCs obtained from normal nerve (Figure 3, A and B)
. However, purified SLCs from
some neurofibromas, particularly those from plexiform tumors, were
stubby and less elongated (Figure 3C)
. The SLCs derived from dermal
neurofibromas had a limited proliferative capacity similar to that of
normal adult SCs (10 doublings) and became senescent thereafter. By
comparison, the plexiform SLCs were less restricted and often were
passaged >20 times before showing signs of senescence. All type-2
SLC-enriched neurofibroma cultures grew very slowly when withdrawn from
GGF-2, but were stable for months on a laminin substratum. SLC-enriched
type-2 cultures accounted for eight of 24 dermal and two of 16
plexiform cultures.
A third type of neurofibroma culture was obtained exclusively from
plexiform tumors and contained SLCs with several preneoplastic
properties. Type-3 SLCs expanded rapidly without close association with
the FLC sublayer and were easily enriched to near homogeneity. Five
type-3 cultures were established from plexiform tumors; SLCs in two
cultures had a stubby, spindle shape whereas the other three cultures
were multipolar or polygonal (Figure 3D)
. In each culture, nearly all
cells were stained for S-100, whereas p75 expression varied. Three
multipolar/polygonal type-3 cultures grew particularly well (doubling
times 2 to 4 days) in response to serum and GGF did not further
increase their growth rates. These cultures also showed protracted
expansion (>20 to 50 passages). The occurrence of these preneoplastic
properties is almost certainly indicative of genetic abnormalities
originating in vivo because growth factor-independent
proliferation and expansion beyond 10 passages were never observed in
cultures of normal human SCs (also see Rutkowski et al29
).
Furthermore, the three GGF-independent type-3 cultures were derived
from sizable recurrent plexiform neurofibromas.
A fourth subset of neurofibroma culture, at low density, were pleomorphic, lacy and phase-pale, and stained faintly, or not at all, for S-100 and p75. However, with increasing density they became elongated spindle cells and grew in parallel arrays. Remarkably, when grown to confluency, these cultures formed dense ridges of S-100-expressing SLCs (not shown). The appearance of these S-100-positive SLCs did not decrease after numerous passages but, instead, continued to increase in number with increasing culture density. These cultures were expanded extensively (>50 passages) before showing signs of senescence. These observations suggest that these cultures contained a poorly differentiated cell type capable of giving rise to a reversibly differentiated SLC component. This type of pleomorphic property also was observed in several neurofibrosarcoma cultures (not shown).
In summary, 40 primary cultures were established from 24 dermal and 16
plexiform neurofibromas and subcultured for enrichment of SLCs.
Twenty-two of 40 cultures did not respond to SLC enrichment or were
predominantly fibroblastic (type-1). Ten dermal and eight plexiform
tumor cultures yielded enriched SLC cultures (type-2, -3, and -4)
suitable for additional genetic and biological studies (listed in Table 1
). The type-2 and type-3 cultures (eight from dermal and seven from
plexiform tumors) were the focus of subsequent studies.
Neurofibromin Expression by SCs in Neurofibromas
Before neurofibroma specimens were prepared for cell culture,
representative specimens of tumor were fixed and processed for routine
histology and immunohistochemistry. Paraffin sections were
immunostained with an anti-neurofibromin peptide antiserum.
Neurofibromin immunoreactivity in normal control tissues was similar to
that reported by other laboratories.31
In neurofibromas
the SC elements were identified by their nuclear characteristics
(elongated, wavy nuclei with pointed ends) and positive
immunoreactivity for S-100 protein (Figure 4)
. Four basic patterns of neurofibromin
immunoexpression in S-100-positive regions were observed: 1)
neurofibromin-negative (-/-); 2) predominantly
neurofibromin-negative with focal areas of distinct positive staining
(-/+); 3) predominantly immunoreactive tumor
with focal areas of distinct negativity (+/-);
and 4) neurofibromin-positive (+/+).
Immunoreactivity patterns for the surgical resection
specimens corresponding to the SC cultures from which they were derived
are summarized in Table 1
. The majority (14 of 17) of the neurofibromas
that gave rise to SC cultures were predominately neurofibromin-negative
(designated -/- and -/+
in Table 1
). Moreover, more than one-half of these tumors were
completely negative (-/-), showing no
neurofibromin immunostaining in any SC elements (Figure 4A)
.
Furthermore, even in the few predominately neurofibromin-positive
tumors (+/-), many individual SCs were negative
for neurofibromin (Figure 4C)
. In areas of positive immunoreactivity,
tumor cells showed discrete, granular staining in the perinuclear
cytoplasm and in delicate elongated processes. In tumors with
positively and negatively stained regions, the regions of neurofibromin
nonreactivity were histologically similar to immunoreactive areas.
Overall, neurofibromin immunoexpression in most neurofibromas that gave
rise to a SC culture was very low (see Discussion). The finding of both
positive and negative areas of neurofibromin immunoreactivity in some
neurofibromas could not be attributed to artifacts of fixation, tissue
preparation, or regional differences in antibody concentration. Dermal
and vascular elements were uniformly positive for neurofibromin in both
immunoreactive and nonreactive regions of a given tumor (Figure 4)
.
Also, tumors that were negative or that contained areas of
neurofibromin nonreactivity were all strongly and widely immunoreactive
for S-100 protein (Figure 4B)
.
|
We hypothesized that the abnormal growth properties of NF1 SC
cultures, particularly the type-3 cultures, are the result of a severe
deficiency in neurofibromin expression. Normal SCs and type-2, -3, and
-4 NF1 SC cultures were examined for the expression of neurofibromin by
Western immunoblotting. Results are shown in Figure 5
. Antibody specificity was examined on
extracts of SC-enriched cultures from embryonic Nf1 knockout
mice. Full-length neurofibromin was absent from homozygous cultures
(Nf1-/-) (lane 1), whereas
heterozygous cultures (Nf1+/-)
(lane 2) expressed a predominant immunoreactive band-pair with a
Mr = 225 to 250 kd (slightly smaller than that
produced by human cells). A band at 140 kd appeared in each sample
that, although undefined, indicated the consistency of total protein
loaded in each lane. Extracts of normal human SC cultures (lane 3)
contained a predominant neurofibromin-immunoreactive band-pair with a
Mr = 240 to 260 kd. In contrast, full-length
neurofibromin was absent in extracts of the 10 dermal SC cultures
(lanes 4 to 13) and eight plexiform SC cultures (lanes 14 to 20 and
22). As an example, we also show the mixed primary culture pNF95.11b
(lane 21) from which the SC culture, SC+(pNF95.11b) (lane 22), was
derived. The first passage of the culture pNF95.11b contained an
admixture of cells (including numerous FLCs and perhaps diverse SC
lineages). A modest band-pair corresponding to full-length
neurofibromin was observed in this culture extract (lane 21),
indicating the genetic heterogeneity in this early culture. This
finding is consistent with the pattern of neurofibromin immunostaining
(+/-) observed in the originating tumor sections
(Table 1)
. However, neurofibromin expression was undetectable in the
derived type-3 SC culture (lane 22). The pNF95.11b cultures were
established from a resection of a recurrent plexiform neurofibroma. Two
years earlier, we established a culture from a specimen obtained from
the initial resection. This culture, pNF95.11a, was type-1
(dominated by FLCs and intractable to SC enrichment) and expressed
abundant neurofibromin (lane 23). Neurofibromin content in this culture
extract was presumably contributed by the large population of
fibroblasts, but we cannot exclude the contribution by NF1
heterozygous SCs. From these results we conclude that
neurofibromin-deficient SCs have the best long-term growth advantages
and that our subculture methods to enrich for SCs from neurofibromas
are highly selective for neurofibromin-deficient SCs.
|
Neoplastic and tumorigenic properties of type-2 and -3
neurofibroma SC cultures were examined in several classical tests,
including assays of serum and anchorage dependence, colony formation in
soft agarose, and subcutaneous tumor formation in immunodeficient mice.
The proliferative properties (normal and abnormal) of these cultures
were described in a previous section. Although not all of the NF1 SC
cultures listed in Table 1
were tested repeatedly, findings of
tumorigenic properties were consistent for all type-2 and -3 cultures
from dermal and plexiform tumors. Notably, all type-2 and type-3 NF1 SC
cultures showed properties similar to normal human SCs. First, the
survival of the NF1 SC cultures and normal SCs was not highly growth
factor-dependent, as 85 to 95% of cells remained viable in serum-free
medium for at least 72 hours (compared to 95% survival of the highly
tumorigenic RN22 schwannoma and C6 glioma cell lines). Proliferation
was not observed in any of the SC cultures in the absence of serum.
Second, the survival of the SC cultures was anchorage-dependent. Under
nonadherent conditions the percentage of viable normal SCs dropped
below 40% after 72 hours. Similarly the survival of NF1 SC cultures
was 20 to 40% after 72 hours in suspension culture (compared to >95%
in the RN22 and C6 lines).
Colony formation in soft agarose is an indicator of high tumorigenicity and anchorage-independent growth, properties of transformed cells and some, but not all solid tumor cultures. Suspended as single cells in soft agarose, 72% of RN22 and 88% of C6 cells proliferated and rapidly formed colonies that become visible to the eye within 2 weeks. In contrast, multicellular foci of normal SC or NF1 SC cultures were rarely observed and none reached the 25-cell colony criterion, even after cultured for 2 months in agarose.
NF1 SC cultures showed no subcutaneous tumorigenic growth. Millions of cells per site were injected subcutaneously in nude mice. No palpable tumors were found after 3 months and there were no subcutaneous foci visible in postmortem examinations. Taken together, these observations indicate that the neurofibroma SC cultures, even the type-3 cultures, had low tumorigenic potential in these classical assays.
Neurofibroma SC Cultures Grafted in the Mouse Nerve
The tumorigenic growth of selected neurofibroma SC cultures was
examined as xenografts in the sciatic nerves of adult immunodeficient
scid mice. Each NF1 culture was engrafted into four nerves
and six nerves were engrafted with an equal number of normal human SCs.
Engrafted nerves were examined by immunostaining with an antibody
specific to human glutathione S-transferase (Figure 6)
. First, transplantation of normal SCs
resulted in transient occupancy (Figure 6C)
and survival appeared to be
severely limited because in four of six of the nerves glutathione
S-transferase labeling was absent after 8 weeks. In
contrast, all neurofibroma SC cultures (seven of seven) showed
persistent and diffuse intraneural growth throughout the same period.
Typically, neurofibroma SCs emanated from glutathione
S-transferase-positive foci and grew in extensive
longitudinal streams that intermingled with the host nerve elements
(Figure 6A)
. Tumor cell migration also was associated with the nerve
sheaths. Some NF1 SC cultures developed sizable masses that displaced
nerve elements and caused significant regional enlargement of the nerve
diameter (Figure 6B)
. Tumor masses varied in size for the different
culture grafts, but additional work is required to quantitate the
size and distribution of the tumor grafts. It will also be important to
determine whether growth rates of the cultures, grafts, and originating
tumors are related. Nevertheless, all observations indicated that tumor
development in the mouse nerve, like that of human neurofibromas, was
relatively slow and benign. There were only sporadic signs of
functional impairment associated with the largest tumors and no
mortality was associated with the transplants for up to 8 weeks. These
results demonstrate reliable and sustained tumor growth by
neurofibroma-derived human SCs implanted in the mouse nerve.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Loss of NF1 gene expression has been reported in malignant and benign SC tumors and pheochromocytomas from patients with NF1.32,33 Based on the two-hit hypothesis of tumor suppressor genes such as NF1, it is expected that neurofibromas should contain a supernumerary population of neurofibromin-deficient cells that are tumorigenic. Recent findings of cytogenetic alterations, loss of heterozygosity, and the absence of NF1 mRNA expression in SCs cultured from neurofibromas strongly implicate SCs as the major neoplastic elements of dermal and plexiform neurofibromas.17-19 In addition, in the present study we found that SCs isolated from numerous neurofibromas lack neurofibromin expression, confirming that NF1 is inactivated in these SCs. Despite these persuasive findings, immunohistochemical evidence for a dominant population of neurofibromin-deficient cells has been highly ambiguous and scarcely reported. In an earlier study of numerous NF1 tumors, we concluded that a majority of neurofibromas consist mainly of SCs that express neurofibromin.34 This antithetical observation was based on a large number of random archival specimens and, moreover, held true for a significant proportion of neurofibrosarcomas as well. However, on closer scrutiny, to examine the two-hit hypothesis at the single-cell level, it became evident that neurofibromin-negative SCs were present in most, if not all, neurofibromas. In the present study we focused on an independent group of neurofibromas from which SC subculture was successful. In most of the tumors in this group, neurofibromin labeling of tumoral elements was particularly sparse or absent. The remainder contained mixed populations of neurofibromin-negative and neurofibromin-positive SCs. The neurofibromin antibody used in these studies was raised against a peptide corresponding to amino acids 509 to 528 of the predicted NF1 gene product. Directed against an N-terminal epitope, the antibody should bind to known neurofibromin splice variants as well as highly truncated (abnormal) forms. These results provide strong evidence that neurofibromin-negative cells of SC lineage contribute centrally to the formation of at least a significant subset of both dermal and plexiform neurofibromas. Nevertheless, there was notable variability in neurofibromin expression by SCs within many of the neurofibromas. The presence of neurofibromin expressing (NF1 heterozygous) SCs suggests they too may perpetuate tumor formation, perhaps driven by the paracrine influence of the neurofibromin-deficient cell population. In this regard, Gutmann and co-workers35 reported that neurofibromin expression by SCs in benign tumors may be down-regulated by factors produced within the tumor. Thus, paracrine influences may represent a novel mechanism for inactivating growth-suppressing genes and allowing for increased cell proliferation in tumors even in nonclonal cells. It is interesting that proliferation of neurofibromin-deficient SCs in response to GGF-2 was not enhanced by forskolin. Because forskolin increases SC expression of growth factor receptors including the GGF receptors erbB2 and erbB3,36,37 this suggests that neurofibroma SCs express high levels of GGF receptors. Hyperexpression of erbB receptors has been reported in NF1 tumors and an inverse expression pattern of erbB2 and neurofibromin was shown for human SCs.38,39 Additionally, SCs can express GGF and, at doses submaximal for proliferation, GGF-2 increases and directs the migration of SCs.40-42 Taken together, these findings raise the possibility that GGF may function in an autocrine/paracrine mechanism that supports the continued growth of SCs in neurofibromas.
Despite considerable advances in the molecular genetics of NF1, the histogenesis of neurofibromas remains enigmatic. Dermal and plexiform neurofibromas contain a variety of cell types including SCs, perineurial cells, and fibroblasts. It is commonly held that despite their cellular complexity the histological features of neurofibromas are monotonously consistent.43 In contradistinction, our studies of numerous neurofibromas and their derivative cell cultures indicate there are several levels of cellular and genetic diversity in this class of benign peripheral nerve sheath tumor. Cytogenetic abnormalities were identified in one of five of the dermal (our unpublished observation) and four of six of the plexiform SC cultures.17 There were no consistent chromosomal regions involved in the abnormal karyotypes, suggesting that originating tumors are heterogeneous and may bear a variety of primary and/or secondary genetic changes. Additionally, the two plexiform cultures that displayed no cytogenetic rearrangements showed GGF-independent growth, suggesting that they either contain underlying genetic abnormalities not yet detected, or have expression abnormalities because of epigenetic influence. Clearly, neurofibromin deficiency did not confer GGF-2-independent growth on all of the developed SC cultures. Despite some differences in morphology, GGF-2 dependence, and karyotype, all NF1 SC cultures showed similarly low tumorigenic potential in several classical in vitro assays. However, the neurofibroma SC cultures showed a strong propensity to aggregate and form culture tumors. A similar growth pattern was observed for suspension cultures whereby NF1 SCs readily grew in aggregates reminiscent of tumor spheroids (an in vitro model for tumorigenic growth). Neurofibroma culture tumors contained an extensive laminin-rich extracellular matrix similar to that observed in neurofibromas.30 These observations attest to a tumorigenic property of neurofibromin-deficient SC cultures that may be related to the adhesive mechanisms involved in the growth and development of neurofibromas. Sheela and co-workers15 first demonstrated that NF1 SCs are angiogenic and invasive. In subsequent studies, which included two of the NF1 SC cultures used in the present report, we also concluded that NF1 SCs have a high invasive potential and a loss of negative autocrine growth control.16 Despite these tumorigenic properties, our NF1 SC cultures, as well as the neurofibroma cultures used by Sheela and colleagues,15 failed to form subcutaneous tumors in immunodeficient mice. Taken together, these findings imply that NF1 SCs have a tumorigenic potential that was not fully expressed in the in vivo model systems used previously.
Inceptive studies demonstrated the growth of implanted human neurofibroma tissue or SC preparations into the sciatic nerves of immunodeficient mice and the potential of this xenograft model for studying the tumorigenesis in NF1.14,44 In the present study neurofibroma-like tumors resulted from the transplantation of neurofibromin-deficient NF1 SC cultures into the nerves of scid mice. Extensive migration was consistently observed and many tumors were sizeable and substantially enlarged in the mouse nerve. There were apparent differences in the growth by the different transplanted NF1 SC cultures but overall tumor expansion was slow, suggesting that the growth rate of the developed tumors may reflect that of human neurofibromas. Additional studies are required to determine whether the growth patterns of the engrafted SC tumors correlate with those of the originating human tumors. This intraneural engraftment model is the first to achieve tumorigenic growth in vivo by human neurofibromin-deficient SCs and provides the means to study the histogenesis of neurofibromas in a relevant cellular environment. A further enhancement to this NF1 tumor model will be to transplant these neurofibroma SCs in the nerves of immunodeficient mice that are also heterozygous for Nf1. This highly relevant model of neurofibroma will also provide the opportunity to observe the interactions and contributions of engrafted (NF1-/-) cells and (Nf1-/+) nerve elements with the same genetic background as those found in NF1 patients.
|
Acknowledgements |
|---|
|
Footnotes |
|---|
Supported by the National Institutes of Health (grants R01NS34780 and R29NS31550) and the U.S. Department of Defense (grants DAMD17-98-1-8609 and DAMD17-00-1-0549).
Accepted for publication October 4, 2000.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  S. Liu, Y. Tian, A. Chlenski, Q. Yang, P. Zage, H. R. Salwen, S. E. Crawford, and S. L. Cohn Cross-Talk between Schwann Cells and Neuroblasts Influences the Biology of Neuroblastoma Xenografts Am. J. Pathol., March 1, 2005; 166(3): 891 - 900. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Plexiform neurofibromas in NF1: Toward biologic-based therapy Neurology, May 28, 2002; 58(10): 1461 - 1470.
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
