Identifying, Diagnosing, and Grading Malignant Peripheral Nerve Sheath Tumors in Genetically Engineered Mouse Models

Summary

We have developed a methodology for assessing whether nervous system neoplasms in genetically engineered mice accurately recapitulate the pathology of their human counterparts. Here, we apply these histologic techniques, defined pathologic criteria, and culture methodologies to neurofibromas and malignant peripheral nerve sheath tumors arising in the P₀-GGFβ3 mouse model.

Abstract

Patients with the autosomal dominant tumor susceptibility syndrome neurofibromatosis type 1 (NF1) commonly develop plexiform neurofibromas (PNs) that subsequently transform into highly aggressive malignant peripheral nerve sheath tumors (MPNSTs). Understanding the process by which a PN transforms into an MPNST would be facilitated by the availability of genetically engineered mouse (GEM) models that accurately replicate the PN-MPNST progression seen in humans with NF1. Unfortunately, GEM models with Nf1 ablation do not fully recapitulate this process. This led us to develop P₀-GGFβ3 mice, a GEM model in which overexpression of the Schwann cell mitogen neuregulin-1 (NRG1) in Schwann cells results in the development of PNs that progress to become MPNSTs with high frequency. However, to determine whether tumorigenesis and neoplastic progression in P₀-GGFβ3 mice accurately model the processes seen in NF1 patients, we had to first prove that the pathology of P₀-GGFβ3 peripheral nerve sheath tumors recapitulates the pathology of their human counterparts.

Here, we describe the specialized methodologies used to accurately diagnose and grade peripheral nervous system neoplasms in GEM models, using P₀-GGFβ3 and P₀-GGFβ3;Trp53^+/- mice as an example. We describe the histologic, immunohistochemical, and histochemical methods used to diagnose PNs and MPNSTs, how to distinguish these neoplasms from other tumor types that mimic their pathology, and how to grade these neoplasms. We discuss the establishment of early-passage cultures from GEM MPNSTs, how to characterize these cultures using immunocytochemistry, and how to verify their tumorigenicity by establishing allografts. Collectively, these techniques characterize the pathology of PNs and MPNSTs that arise in GEM models and critically compare the pathology of these murine tumors to their human counterparts.

Introduction

Over the last three decades, numerous laboratories have attempted to create mouse models of human cancers by introducing human cancer-associated mutations into the mouse genome or by overexpressing a gene product that is overexpressed in human cancers. The resulting genetically engineered mouse (GEM) models can be used for a variety of purposes such as establishing that the newly introduced genomic modification initiates tumorigenesis, identifying other subsequently occurring genetic or epigenetic changes that contribute to tumor progression, and defining the key signaling pathways that drive tumor initiation and progression. Unlike orthotopic xenograft models, which rely on the use of immunodeficient mice, GEM cancer models have a fully functional immune system and so more accurately model responses to candidate therapeutic agents. However, when using GEM cancer models for purposes such as these, it is essential that investigators confirm that observations made with GEM neoplasms are relevant to their human counterparts. This validation should include a thorough assessment of the pathology of the GEM neoplasms and a determination as to whether the pathologic features of the GEM neoplasms recapitulate the pathology of the corresponding human tumor type.

The tumor susceptibility syndrome neurofibromatosis type 1 (NF1) is the most common genetic disease affecting the human nervous system, occurring in approximately 1 in every 3,000-3,500 live births^1,2,3. Individuals afflicted with NF1 develop multiple benign peripheral nerve sheath tumors known as neurofibromas in their skin (dermal neurofibromas) and in large nerves and nerve plexuses (plexiform neurofibromas). While both dermal and plexiform neurofibromas worsen the patient's quality of life by producing physical, behavioral, and/or social impairment, plexiform neurofibromas (PNs) are particularly dangerous^4,5. This is because PNs frequently transform into malignant peripheral nerve sheath tumors (MPNSTs), which are aggressive spindle cell neoplasms with an exceptionally low survival rate^1,2. In large part, this low survival rate is because the radio- and chemotherapeutic regimens that are currently used to treat MPNSTs are ineffective. However, developing new, more effective therapies has been challenging. This is because, despite how commonly MPNSTs occur in NF1 patients, they are still rare neoplasms. As a result, it is very difficult to obtain large numbers of human tumors for study; it is also challenging to recruit enough patients with MPNSTs for clinical trials. To overcome these limitations, several GEM models have been generated with the goal of gaining further insights into the abnormalities driving neurofibroma pathogenesis and PN-MPNST progression and to facilitate preclinical trials with candidate therapeutic agents.

NF1 patients have inactivating mutations in one copy of the NF1 gene. Neurofibroma pathogenesis is triggered when an inactivating mutation in the remaining functional NF1 gene occurs in a cell in the Schwann cell lineage. Surprisingly, however, when mice were generated with germline inactivating Nf1 mutations, they did not develop neurofibromas^6,7. The subsequent demonstration that mice with Nf1-null Schwann cells and Nf1 haploinsufficiency in all other cell types (Krox20-Cre;Nf1^flox/- mice) developed plexiform neurofibromas suggested that reduced Nf1 gene dosage in additional cell types was required for neurofibroma pathogenesis⁸. Even then, the plexiform neurofibromas in Krox20-Cre;Nf1^flox/- mice did not progress to become MPNSTs and so only partially mimicked the biology of their human counterparts. MPNST pathogenesis did occur when Nf1 mutations were partnered with mutations in additional tumor suppressor genes such as Trp53⁹ or Cdkn2a¹⁰, but MPNSTs in these GEM models developed de novo or from atypical neurofibromatous neoplasms of uncertain biologic potential (ANNUBPs)^11,12, rather than from preexisting benign plexiform neurofibromas (see^13,14 for excellent reviews of these models as well as other models introducing additional MPNST-associated loss of function mutations in genes such as Suz12 and Pten¹⁵).

These mouse models have been invaluable for establishing the role that genes such as NF1, TP53, and CDKN2A play in the pathogenesis of NF1-associated peripheral nervous system neoplasms and for preclinical trials testing candidate therapeutic agents. However, we still have an incomplete understanding of the process by which plexiform neurofibromas progress to become atypical neurofibromatous neoplasms of uncertain biologic potential (ANNUBPs¹⁶) and then MPNSTs. Some progress has recently been made in understanding this process with the recent report that mice with deletions in Nf1 and Arf develop ANNUBPs that progress to become MPNSTs¹¹. However, Nf1 mutation-based mouse models that fully recapitulate the process of plexiform neurofibroma-MPNST progression seen in humans do not yet exist. In addition, it is not clear whether there are multiple distinct pathways that lead to the development of MPNSTs. Given this, it is possible that the GEMs described above only model a subset of several different pathways that lead to neurofibroma-MPNST progression and MPNST pathogenesis. This point is emphasized by the fact that MPNSTs also occur sporadically and that some sporadic MPNSTs apparently do not have NF1 mutations^17,18.

Although this latter point has been challenged by Magollon-Lorenz et al.'s recent suggestion that at least some sporadic MPNSTs lacking NF1 mutations are melanomas or a different type of sarcoma¹⁹, we have recently reported a sporadic MPNST and a cell line derived from this tumor (2XSB cells) that was NF1 wild-type²⁰. During the characterization of the parent tumor and the 2XSB cell line, we systematically ruled out alternative diagnostic possibilities, including melanoma and the multiple other sarcoma types that are routinely considered in the differential diagnosis of a sporadic malignant peripheral nerve sheath tumor²⁰. In addition, we note that Magollon-Lorenz et al. acknowledged that their findings in the three sporadic MPNST cell lines that they studied could not be generalized to indicate that all tumors identified as sporadic MPNSTs are not MPNSTs.

To construct a GEM model in which neurofibroma and MPNST pathogenesis were not necessarily dependent upon specific tumor suppressor gene mutations, we generated transgenic mice in which overexpression of the potent Schwann cell mitogen neuregulin-1 (NRG1) was driven by the Schwann cell-specific myelin protein zero (P₀) promoter (P₀-GGFβ3 mice)²¹. We have previously shown that human neurofibromas, MPNSTs, and MPNST cell lines express several NRG1 isoforms together with the erbB receptor tyrosine kinases (erbB2, erbB3, and erbB4) that mediate NRG1 signaling and that these erbB receptors are constitutively activated²². We have also demonstrated that pharmacologic inhibitors of the erbB kinases potently inhibit MPNST proliferation²², survival²³, and migration²⁴. In keeping with our observations in humans, P₀-GGFβ3 mice develop plexiform neurofibromas²⁵ that progress to become MPNSTs at a high frequency^21,25. We have shown that P₀-GGFβ3 MPNSTs, like their human counterparts, commonly develop mutations of Trp53 and Cdkn2a, as well as numerous other genomic abnormalities that potentially contribute to tumorigenesis²⁵. MPNSTs arising in P₀-GGFβ3 mice do not have inactivating Nf1 mutations. However, using genetic complementation, we showed that NRG1 promotes tumorigenesis in P₀-GGFβ3 mice predominantly through the same signaling cascades that are altered by Nf1 loss²⁶; this conclusion is based on our finding that substituting NRG1 overexpression for Nf1 loss in the presence of Trp53 haploinsufficiency (P₀-GGFβ3;Trp53^+/- mice) produces animals in which MPNSTs develop de novo, as is seen in cis-Nf1^+/-;Trp53^+/- mice²⁷.

To obtain this and other information demonstrating that P₀-GGFβ3 mice accurately model the processes of neurofibroma pathogenesis and neurofibroma-MPNST progression seen in humans with NF1, we have developed specialized methodologies for processing tissues from these animals, accurately diagnosing their tumors, grading the MPNSTs arising in these mice, establishing and characterizing early-passage P₀-GGFβ3 and P₀-GGFβ3;Trp53^+/- MPNST cultures and critically comparing the pathology of P₀-GGFβ3 PNs and MPNSTs and P₀-GGFβ3;Trp53^+/- MPNSTs to that of their human counterparts. Many of these methodologies are generalizable to other GEM models of nervous system neoplasia. Additionally, several of these methodologies are more broadly applicable to GEM models in which neoplasms arise in other organ sites. Consequently, here we present a detailed description of these methodologies.

Protocol

The procedures described here were approved by the Medical University of South Carolina's IACUC and were performed by properly trained personnel in accordance with the NIH Guide for the Care and Use of Laboratory Animals and MUSC's institutional animal care guidelines.

1. Determining tumor penetrance and survival in P₀-GGFβ3 mice and identifying tumors in these animals for further characterization

1. Generate the cohort of mice that will be assessed for tumorigenesis. The number of mice required depends on the penetrance of the tumor phenotype. To compensate for losses such as these, start with a cohort that is 10-15% higher than the desired final number of animals.
   NOTE: We determined tumor penetrance in P₀-GGFβ3 mice empirically in an initial cohort of 50 mice (six of which were lost for reasons unrelated to neoplasia (e.g., fighting))²⁵.
2. Assess the health of animals three times weekly and record body condition scores, including weight, and behavior changes at each examination. Terminate tumor-bearing or moribund mice (see note below) using carbon dioxide euthanasia followed by cervical dislocation. Record age at death in days. If tumors are externally visible, record tumor dimensions (length, width, and height).
   NOTE: It is essential that animals are not allowed to proceed past the IACUC-approved maximum allowable tumor size and/or humane endpoint.
3. Using the age at death, establish a Kaplan-Meier survival curve to determine the mean age at death and range of survival.
   NOTE: P₀-GGFβ3 mice had a mean age at death of 261.5 days, with a range of 74-533 days; 91% of our cohort had multiple neurofibromas and 71% had MPNSTs²¹.
4. Determine whether grossly visible tumors are present and, if they are, dissect them under sterile conditions to establish whether the tumor is associated with a peripheral nerve and then excise the tumor. In P₀-GGFβ3 and P₀-GGFβ3;Trp53^+/- mice, peripheral nerve sheath tumors are most commonly found in the trigeminal and sciatic nerves. Even if grossly visible tumors are not seen in association with these nerves, fix and embed these nerves as described below so that they can be examined for the presence of micro-tumors. Micro-tumors typically occur within the ganglia associated with these nerves.
5. Cut grossly visible tumors into three pieces (Figure 1A). Use piece 1 for histology (paraffin sections, immunohistochemistry, and histochemistry). Use piece 2 to establish early-passage cultures. Snap-freeze piece 3 (using liquid nitrogen) for possible future experiments (e.g., immunoblots, RNA and DNA isolation).
6. Fix piece 1 in 4% paraformaldehyde in phosphate-buffered saline (PBS) overnight at 4 ^oC. Fix the remainder of the animal's body by immersion in 4% paraformaldehyde overnight at 4 °C.

2. Paraffin-embedding of grossly visible tumors and preparation of hematoxylin and eosin-stained sections for initial diagnostic assessment

1. Paraffin embedding of grossly visible tumors
   1. After fixing tumor piece 1 overnight in 4% paraformaldehyde, rinse the tissue by placing it in a volume of PBS that is at least 10x greater than the volume of the tissue for 15 min; use the same volume for all subsequent rinses. Repeat two more 15 min PBS rinses, using a fresh volume of PBS for each rinse.
   2. Transfer tumor piece 1 to 70% ethanol. Hold the tissue in 70% ethanol for 24 h at 4 °C. If desired, store the tissue long-term under these conditions.
      NOTE: Steps 2.1.1 through 2.1.7 are provided for the benefit of investigators who do not have access to a commercial tissue-processing machine. If the investigator does have access to a tissue processor, the tissue will be processed through graded ethanols and xylenes per the programmed instrument protocol.
   3. Transfer tumor piece 1 to 80% ethanol for 30 min at room temperature. Repeat the incubation for an additional 30 min in a fresh volume of 80% ethanol.
   4. Transfer tumor piece 1 to 95% ethanol for 75 min at room temperature. Repeat the incubation twice more, each time for an additional 75 min in a fresh volume of 95% ethanol.
   5. Transfer tumor piece 1 to 100% ethanol and incubate for 60 min at room temperature. Repeat the 60 min incubation twice more, each time using a fresh volume of 100% ethanol.
   6. Transfer tumor piece 1 to a d-limonene-based solvent and incubate for 30-60 min at room temperature. Transfer tumor piece 1 to a fresh volume of the d-limonene-based solvent and incubate for an additional 30-60 min at room temperature.
   7. Transfer tumor piece 1 into 1:1 paraffin/d-limonene-based solvent for 1 h at 60 ^oC. Discard the 1:1 paraffin/d-limonene-based solvent and transfer tumor piece 1 to 60 ^oC paraffin and incubate for 20 min at 60 ^oC. Repeat the incubation in a fresh volume of 60 ^oC paraffin once more for 20 min. Incubate tumor piece 1 in paraffin overnight at 60 ^oC.
   8. Pour a base layer of paraffin into a steel histology mold and let it solidify. Transfer tumor piece 1 to the surface of the base paraffin and immediately after positioning, cover the tissue with 60 ^oC paraffin and place the tissue cassette on top of and in contact with the molten paraffin. Transfer the mold to the cold side of the embedding station and let it solidify for 10-15 min.
   9. Remove the paraffin block with the attached tissue cassette from the mold. Use the attached tissue cassette to clamp the block into the microtome during sectioning.
      NOTE: The paraffin block can now be stored indefinitely at room temperature.
2. Prepare hematoxylin and eosin-stained sections of grossly visible tumors.
   1. Cut 4-5 µm sections of tumor tissue using a microtome and mount them on positively charged glass slides.
   2. To perform hematoxylin and eosin (H&E) staining, deparaffinize the tissue sections by incubating the slides in a d-limonene-based solvent for 10 min at room temperature. Repeat the incubation in a fresh volume of d-limonene-based solvent for 10 min.
   3. Transfer the slides to 100% ethanol and incubate for 5 min at room temperature. Transfer the slides to a fresh volume of 100% ethanol and incubate for another 5 min at room temperature.
   4. Transfer the slides to 95% ethanol and incubate for 5 min at room temperature. Transfer the slides to a fresh volume of 95% ethanol and incubate for another 5 min at room temperature.
   5. Transfer the slides to 70% ethanol and incubate for 5 min at room temperature. Then, transfer to 50% ethanol and incubate for 5 min at room temperature.
   6. Transfer the slides to distilled water and incubate for 5 min at room temperature.
   7. Stain the slides by immersing them in hematoxylin for 5 min at room temperature. Rinse with running tap water for 1-2 min. Differentiate by dipping slides 2-5x in 0.5% hydrochloric acid in 70% ethanol. Rinse in a running tap water bath for 1-2 min. Place the slides in 95% ethanol with 0.5% acetic acid for 10 s.
   8. Stain the slides with eosin-Y for 30 s at room temperature and then transfer the slides to 100% ethanol for 5 min. Clear the samples in a d-limonene-based solvent for 5 min.
   9. Mount coverslips using a xylene-based mounting medium while the slide is still wet with the d-limonene-based solvent. Allow the mounting medium to set overnight.
      NOTE: Do not use a water-based mounting medium.
3. Prepare tissues without grossly visible tumors to identify plexiform neurofibromas and micro-MPNSTs. Embed the tissues in paraffin, prepare histologic sections, and stain these sections with hematoxylin and eosin.
   1. Remove the internal organs and embed representative portions of each organ in paraffin to prepare H&E-stained sections that will be examined for microscopic evidence of tumors (Figure 1B). Remove the head, forelimbs, hindlimbs, and tail (Figure 1C). To examine the spinal cord and nerve roots in situ for evidence of nerve root tumors, decalcify the vertebral column and associated ribs and soft tissue by immersion in 0.3 M EDTA/4% paraformaldehye (pH 8.0) for 48-72 h at 4 ^oC; then, rinse the samples with 1x PBS. At the end of decalcification, ensure that decalcification is complete by attempting to pierce the vertebral column with a needle. Decalcification is successful if the needle easily penetrates the bone without crunching.
   2. Cut the decalcified vertebral column and associated structures into blocks that will fit in a tissue cassette. Dehydrate through graded ethanols followed by a d-limonene-based solvent as described in steps 2.1.2-2.1.7. Embed in paraffin and prepare 4-5 µm sections as described in steps 2.1.7-2.2.1. Perform H&E stains as described in steps 2.2.2-2.2.9; allow the mounting medium to set overnight.

3. Identify potential plexiform neurofibromas and perform special stains to confirm diagnoses

NOTE: We strongly recommend including an experienced human or veterinary pathologist in the evaluation of H&E and special stains of GEM tumor sections.

1. Examine the H&E-stained slides prepared as described above using brightfield microscopy to identify potential peripheral nerve sheath tumors. Since neurofibromas and MPNSTs arise within peripheral nerves, it is important to determine whether microscopic tumors are associated with a peripheral nerve. Identify blocks with potential peripheral nerve sheath tumors so that immunostains and other special stains can be performed to confirm or refute diagnoses.
2. Distinguish potential PNs from MPNSTs/other high-grade neoplasms based on histologic features seen during brightfield examination of H&E-stained sections. Human PNs are mildly hypercellular neoplasms predominantly composed of cells with elongated, often wavy nuclei. In many areas, the cells are loosely packed and may be separated by myxoid extracellular material; PNs in P₀-GGFβ3 mice have a similar appearance. Mitoses are typically not present; if they are and the tumor is moderately to highly hypercellular, the tumor is an MPNST or another type of high-grade neoplasm (see below).
3. PNs are composed of a mixture of neoplastic Schwann cells and other non-neoplastic cell types such as mast cells. To confirm the identity of a PN, perform immunostains for the Schwann cell markers S100β and Sox10 and the mast cell marker CD117 (c-Kit) on sections from blocks containing potential PNs identified on H&E examination (see section 3.4 for alternative mast cell marker staining options). Perform Ki67 immunostains to confirm low proliferation rates.
   NOTE: Table 1 lists the antigen markers used for routine identification of GEM plexiform neurofibromas (S100β, Sox10, CD117, Ki67), the diagnosis of MPNSTs, and for a complete assessment of other cell types present in neurofibromas; we stain for these latter antigens only when first describing a new GEM model. Consult the antibody manufacturer's datasheet for recommended conditions. Note whether the antibody manufacturer's insert recommends antigen retrieval; not all antigens require retrieval to be detectable.
   1. Prepare 4-5 μm sections from the selected paraffin blocks and mount them on positively charged slides. Deparaffinize sections as described in steps 2.2.2-2.2.6.
   2. Rinse sections with 1x PBS for 3-5 min.
   3. If the antibody manufacturer recommends antigen retrieval, prepare citrate buffer. Combine 9 mL of citric acid solution (10.5 g of citric acid in 500 mL of distilled water) and 41 mL of sodium citrate solution (14.7 g of sodium citrate in 500 mL of distilled water).
   4. Perform antigen retrieval by placing slides in citrate buffer for 20 min in a heated rice cooker.
   5. Transfer slides into 3% hydrogen peroxide/1x PBS for 10 min to eliminate peroxidase activity in the tissue.
      NOTE: Make this solution fresh each time and do not use the stock solution of hydrogen peroxide if it is older than 3-4 months.
   6. Rinse sections in 1x PBS.
   7. Block sections using blocking buffer (2% BSA, 0.1% Triton X-100 in 1x PBS) for 1 h. Rinse slides briefly in 1x PBS.
   8. Add primary antibody to tissue sections. Prepare primary antibodies in diluted blocking buffer. Incubate slides for 2 h at 37 ^oC or overnight at 4 ^oC.
   9. Wash slides in 1x PBS for 5 min. Repeat 3x.
   10. Incubate tissue with biotinylated secondary antibody for 1-2 h.
   11. Prepare diaminobenzidine (DAB) solution based on the manufacturer's protocol and immerse slides in the solution for 2-10 min. Wash slides in water for 5 min.
   12. Counterstain sections by immersing them in hematoxylin for 10-15 min. Expose to running water until it runs clear. Quickly dip slides in alcohol and rinse slides in water.
   13. Dehydrate slides in graded ethanols by quickly dipping slides, 4-5x per solution, in 70% ethanol, 95% ethanol, and then 100% ethanol. Incubate slides in a d-limonene-based solvent for 2 min. Incubate slides in a second batch of d-limonene-based solvent for 2 min.
   14. Mount slides with a xylene-based mounting medium.
   15. Examine slides by brightfield microscopy.
   16. For immunofluorescence, omit steps 3.3.10-3.3.15. Instead, incubate tissue with species-specific secondary antibody diluted in blocking buffer for 1-2 h.
   17. Rinse slides in 1x PBS for 5 min. Repeat 3x.
   18. Mount slides using 1x PBS/glycerol.
   19. Examine slides using fluorescence microscopy.
4. Alternative method for staining mast cells: toluidine blue staining
   1. Prepare toluidine blue stock solution by dissolving 1 g of toluidine blue O in 100 mL of 70% ethanol.
   2. Prepare 1% sodium chloride (NaCl) solution by dissolving 0.5 g of NaCl in 50 mL of distilled water. Mix to dissolve. Adjust pH to approximately 2.0-2.5 using HCl.
      NOTE: This NaCl solution must be made fresh each time stains are performed.
   3. Prepare toluidine blue working solution by adding 5 mL of toluidine blue stock solution to 45 mL of the 1% NaCl solution. Mix well and ensure that the pH is approximately 2.3.
      NOTE: A pH higher than 2.5 will result in staining with less metachromasia. Make this solution fresh every time staining is performed.
   4. Deparaffinize 4-5 μm sections mounted on positively charged slides as described in steps 2.2.2-2.2.6. Wash deparaffinized sections in running water for 2 min.
   5. Stain sections in toluidine blue working solution for 2-3 min. Wash in distilled water for 2 min.
   6. Dehydrate sections by dipping 10x in 95% ethanol. Dip slides in 100% ethanol 10x. Dip slides in fresh 100% ethanol 10 more times.
   7. Incubate slides in a d-limonene-based solvent for 2 min. Incubate slides in a second batch of d-limonene-based solvent for 2 min.
   8. Mount coverslips using a xylene-based mounting medium while the slide is still wet with a d-limonene-based solvent.
      NOTE: Do not use a water-based mounting medium.
   9. Examine slides by brightfield microscopy. Identify mast cells by the presence of dark purple granules in their cytoplasm. Other cell types are stained light blue.

4. Special stains to diagnose MPNSTs and rule out alternative diagnoses

1. The histologic appearance of human MPNSTs is highly variable and several different tumor types have an appearance similar to that of MPNSTs²⁸. Since MPNSTs are derived from the Schwann cell lineage, determine whether the tumor arises from a peripheral nerve or within an existing benign PN by gross examination or brightfield microscopic examination of H&E-stained sections. Although MPNSTs are occasionally found not in association with a peripheral nerve or PN (usually because of inadvertent separation from the nerve during dissection, destruction of the pre-existing PN via MPNST overgrowth, or because the lesion is metastatic), look for the association of the tumor with a nerve or PN as the diagnosis is less likely if the tumor is not associated with a nerve or PN.
2. Prepare 4-5 μm sections from paraffin blocks containing potential MPNSTs and mount them on positively charged slides as described in step 2.2.1. Deparaffinize sections as described in steps 2.2.2-2.2.6.
3. Perform immunohistochemistry with the panel of antibodies indicated in Table 2 as described in steps 3.3.1-3.3.15.
4. Examine the immunostains by brightfield microscopy. Since MPNSTs demonstrate schwannian differentiation, look for uniform or patchy immunoreactivity for S100β, nestin, and Sox10. If the tumors are not positive for these three markers, refer to Table 2 to establish the diagnosis.
   ​NOTE: Human and mouse MPNSTs can demonstrate divergent differentiation. For example, some human MPNSTs demonstrate focal rhabdomyoblastic differentiation (these variants are known as malignant Triton tumors); although these neoplasms will stain for the Schwannian markers, they also express muscle markers such as desmin, MyoD1, and myogenin. Immunoreactivity for these latter markers should not dissuade researchers from diagnosing the tumor as an MPNST. Although multiple mouse models conditionally expressing the SS18-SSX fusion gene have been established to model synovial sarcoma pathogenesis²⁹, to the best of our knowledge, synovial sarcoma models that spontaneously generate the equivalent fusion transcript are not known. Consequently, we do not look for SS18-SSX fusion transcripts in GEM tumors and rely instead on immunohistochemical profiles.

5. Grading of the MPNSTs

1. Determine whether tumor necrosis, a hallmark of WHO grade IV MPNSTs, is present. For Grade IV MPNSTs, look for prominent hypercellularity, brisk mitotic activity (≥4 mitoses per 10 high power (40x) fields), and cytologic atypia.
   1. If necrosis is not present, assess the tumor to determine whether hypercellularity, brisk mitotic activity, and cytologic atypia are present. If these features are all present in the absence of necrosis, grade the tumor as a WHO grade III MPNST.
   2. WHO grade II tumors are characterized by cellularity that is increased, but to a lesser degree than WHO grade III and IV MPNSTs. The nuclei of tumor cells in a WHO grade II MPNST demonstrate increased size (more than 3x the size of the tumor cell nuclei in neurofibromas) and hyperchromasia. Increased mitotic activity (<4 mitoses per 10 high power fields) is associated with, but not a requirement for, WHO grade II tumors.
      NOTE: Since higher-grade tumors typically progress from lower grades, low-grade and high-grade regions may be present in the same MPNST. In this circumstance, the tumor's grade is determined by the highest-grade region present.
      NOTE: There is controversy among human pathologists as to whether the WHO grading system described above is predictive of patient outcomes and some of these pathologists prefer instead to simply classify MPNSTs as low grade or high grade. Under this scheme, high-grade MPNSTs have prominent cytologic atypia, brisk mitotic activity (>5 mitoses per 10 high power fields), and marked hypercellularity with or without necrosis. Low-grade MPNSTs lack necrosis but show mitotic activity, cytologic atypia, and cellularity that is intermediate between a high-grade MPNST and an atypical neurofibromatous neoplasm with unknown biological potential (ANNUBP). We prefer the system described above because distinguishing between an ANNUBP and a low-grade MPNST can be challenging for investigators that do not have long experience with these entities.

6. Preparation of cultures of early passage P₀-GGFβ3 MPNST cells

1. Culture early-passage P₀-GGFβ3 tumor cells from fresh tumor piece 2 (step 1.5, Figure 1A). Maintain sterile conditions from this point forward.
   1. Dissociate the tumor by cutting it into small (2-4 mm) pieces and mincing in DMEM10 (Dulbecco's minimal essential medium) containing 10% fetal calf serum, 2 µmol/L forskolin, and 10 nmol/L neuregulin 1β (NRG1β) in 100 mm tissue culture dishes.
   2. Allow cells to grow out from the minced tissue fragments at 37 ^oC in 5% CO₂ until plates are confluent. Change media every 3-4 days.
   3. Split the cell culture. Remove the media from the 100 mm dish and wash cells with PBS; remove and discard minced tissue. Remove PBS and add 1 mL of 0.25% trypsin for 2-5 min at room temperature. To promote cell detachment, gently tap the plate and check for detachment using a light microscope; extend trypsin incubation as needed.
   4. Neutralize trypsin by adding 2 mL of DMEM10. Transfer the cell suspension to a centrifuge tube and pellet cells at 500 × g for 5 min. Remove the media and add 5 mL of fresh DMEM10 to the pellet.
   5. Distribute the cell suspension into two to four 100 mm dishes containing DMEM10. Do not split the cells for more than five passages (steps 6.1.3 and 6.1.4) and maintain them in DMEM without forskolin and NRG1β. Remember to freeze down cells at passage 2 for future use.

7. Verifying the identity of early-passage MPNST cells by immunocytochemistry

1. Place sterile 18 mm #1.5 round glass coverslips in the wells of 6-well sterile tissue culture dishes.
   1. Place poly-L-lysine/laminin solution into the wells in a volume sufficient to cover the coverslip. Seal the plate with plastic wrap to minimize evaporation. Incubate at 4 °C overnight. The next morning, wash the coverslips 3x with sterile PBS.
      NOTE: We have attempted to plate early passage MPNST cells on uncoated coverslips and have found that the cells do not adhere well in the absence of poly-L-lysine/laminin coating.
   2. Plate 10,000 cells onto the coverslip by gently adding 2 mL of the cell suspension in growth media into each well containing the coverslip. Incubate the plates overnight at 37 °C with 5% CO₂ in a tissue culture incubator.
   3. The next morning, rinse the coverslips for 2 x 5 min with PBS.
   4. Fix cells with 4% paraformaldehyde in PBS for 18 min at room temperature.
   5. Rinse the coverslips 3 x 5 min with PBS. If desired, seal the plate using plastic film and store it at 4 °C at this point.
   6. Make fresh 50 mM NH₄Cl in PBS. Quench aldehydes by incubating coverslips in 50 mM NH₄Cl in PBS for 10 min at room temperature.
   7. Permeabilize the cells by incubating coverslips in 0.3% Triton X-100 in PBS for 15 min at room temperature. Wash coverslips for 3 x 5 min with PBS.
   8. Block non-specific binding by incubating coverslips in blocking buffer (1x PBS containing 1% bovine serum albumin, 0.2% nonfat dry milk, and 0.3% Triton X-100) for 1 h at room temperature.
   9. Remove the blocking buffer and add primary antibodies recognizing the MPNST markers present in the parent tumor (typically S100β, Nestin, and Sox10) diluted to the predetermined optimal dilution in the blocking buffer. Wrap the plate with plastic film and incubate at 4 °C overnight in a humidified chamber.
   10. Wash 3 x 5 min with PBS to remove unbound primary antibody. Add fluorescently labeled secondary antibody diluted in blocking buffer and incubate for 1 h at room temperature. Protect from light.
   11. Wash 3 x 5 min with PBS. Incubate coverslips in 1-5 µg of Hoechst dye for 10 min. Continue to protect from light.
   12. Wash coverslips with PBS for 5 min. Mount slides using 1:1 1x PBS/glycerol. Image with a fluorescent microscope.

8. Allograft of early-passage tumor cells to demonstrate tumorigenicity

1. Grow early-passage P₀-GGFβ3 MPNST cells in DMEM10 to 80% confluence. Rinse cells once with room temperature Hanks' balanced salt solution (HBSS). Treat cells for 30 s to 1 min with a non-enzymatic cell dissociation solution to remove them from the substrate. Add 5 mL of DMEM10 per 1 mL of a non-enzymatic cell dissociation reagent.
   NOTE: Cellular dissociation may take a great deal longer, up to 10-20 min. Please refer to the manufacturer's protocol.
   NOTE: Do not grow cells to confluence as this will reduce graft effectiveness.
   1. Count cells using a hemocytometer. Spin cells down at 5 × g for 5 min and resuspend them at a concentration of 1-2 × 10⁶ cells per 100 µL of DMEM10. Keep cells on ice until ready for injection.
   2. Anesthetize NOD-SCIDγ (NOD.Cg-Prkdc^scidIl2rg^tm1wjl/SzJ) mice in the induction chamber of an isoflurane vaporizer. Place the animal on its stomach and sterilize the injection site with 70% ethanol. Allow the site to air dry prior to injection. Prior to injection, allow cells to come to room temperature.
   3. Inject 1-2 x 10⁶ cells subcutaneously in the right flank. Place the mouse alone in a bedding-free cage and allow it to recover. Make sure that only part of the cage is on a heating pad to prevent hypothermia. This provides a zone that the mouse can move to if it becomes overheated.
      NOTE: We graft a minimum of three mice with each early-passage culture as some grafts may not take.
   4. Allow the graft to grow for 15-60 days; assess animal health 3x weekly and record body condition scores, including weight, and behavior changes at each examination. Terminate mice that have reached maximally allowable graft size or moribund mice using carbon dioxide euthanasia followed by cervical dislocation. Record age at death in days. As tumors become externally visible, record tumor dimensions (length, width, and height).
      ​NOTE: In our experience, different early passage MPNST cultures graft with varying efficiency and differ in the time required for graft growth. Animals must not be allowed to proceed past the IACUC-approved maximum allowable tumor size and/or humane endpoint.
2. Harvest the tumor, fix it in 4% paraformaldehyde, embed it in paraffin, and perform H&E stains as described in sections 2.1-2.2.9 to verify that the allograft is a tumor rather than a reactive tissue. If necessary, immunostain the graft for the markers that were present in the parent tumor.

Results

Figure 2 illustrates examples of grossly evident neoplasms arising in P₀-GGFβ3 mice. Tumors that are easily identifiable with the naked eye may be seen as masses distending body regions as shown in Figure 2A (arrow). When determining whether the neoplasm is potentially a peripheral nerve sheath tumor, it is essential to establish that the tumor is associated with a peripheral nerve. In this instance, an MRI scan (Figure 2B) demonstrates that the tumor is associated with the sciatic nerve (arrowhead); this association was confirmed after euthanizing the mouse and dissecting the tumor. It should be noted that a large size does not necessarily indicate that the tumor is malignant. In this case, a histologic examination of the tumor (Figure 2C) demonstrated that it was a neurofibroma. Most commonly, however, grossly evident tumors are not identified until performing the necropsy of the mouse. Figure 2D illustrates a large, fleshy MPNST that arose within the brachial plexus of this animal.

Figure 3 illustrates representative examples of MPNSTs and neurofibromas from P₀-GGFβ3 mice prepared according to the procedures outlined in protocol section 1. Figure 3A-J illustrates ten examples of independently arising MPNSTs from our P₀-GGFβ3 mouse colony. Note that the histologic appearance of MPNSTs can be highly variable, even between MPNSTs arising independently in the same animal. This histologic variability illustrates why we routinely confirm the diagnosis of P₀-GGFβ3 MPNSTs with immunohistochemical stains. We would also point out that other, apparently unrelated tumor types occur sporadically at a low frequency in some inbred mouse strains (e.g., we have several times encountered lymphomas in P₀-GGFβ3 mice carrying the transgene on a C57BL/6J background), further emphasizing the importance of using immunohistochemistry to confirm tumor diagnoses. Despite the variability of their histologic appearance, all ten of the tumors illustrated in Figure 3A-J were immunoreactive for S100β and nestin and negative for markers of other tumor types. Figure 3K illustrates a representative image of a properly decalcified vertebral column and associated tissues. Note that the spinal cord, nerve roots, vertebral body, and skeletal muscle all maintain their proper anatomic relationship with each other. Figure 3L is a higher-power image of the vertebral body and overlying spinal cord. Since this tissue is properly decalcified, the bone has cut easily without shredding or folding, and bone marrow is readily identifiable in the marrow spaces. If the tissue had not been decalcified properly, it would have caught on the microtome blade and been torn out of the section, doing significant damage to adjacent tissues (spinal cord, spinal nerve roots, and skeletal muscle. Figure 3M presents a representative image of a neurofibroma arising within a dorsal nerve root in a P₀-GGFβ3 mouse. Note that this tumor is less cellular than the MPNSTs shown in Figure 3A-J. The key diagnostic for neurofibromas is that they are composed of a complex mixture of neoplastic Schwann cells and non-neoplastic mast cells, macrophages, fibroblasts, and perineurial-like elements that infiltrate the nerve and spread axons apart.

Figure 4 illustrates examples of the stains that are most useful for the initial identification of a plexiform neurofibroma and distinguishing between a plexiform neurofibroma and an MPNST. Figure 4A illustrates S100β immunoreactivity in a P₀-GGFβ3 plexiform neurofibroma. Note that S100β immunoreactivity is only evident in a subpopulation of cells, which is in keeping with the fact that neurofibromas are composed of a mixture of neoplastic Schwann cells and other non-neoplastic elements (fibroblasts, mast cells, macrophages, perineurial-like cells, a poorly defined CD34-immunoreactive cellular population, and vasculature). Unfortunately, patchy S100β staining does not distinguish between plexiform neurofibromas and MPNSTs as S100β staining can be patchy in MPNSTs (H&E staining is useful for this purpose; however, since MPNSTs are typically more cellular than plexiform neurofibromas [see Figure 3]). Neoplastic Schwann cells are also immunoreactive for the intermediate filament nestin as demonstrated in the P₀-GGFβ3 MPNST presented in Figure 4B. Neoplastic Schwann cells also often demonstrate nuclear immunoreactivity for the transcription factor Sox10, as shown in an MPNST in Figure 4C. Features useful for distinguishing between plexiform neurofibromas and MPNSTs are the presence of mast cells and prominent immunoreactivity for the Ki67 proliferation marker. Figure 4D illustrates an Unna stain performed on a P₀-GGFβ3 plexiform neurofibroma to highlight the presence of mast cells, which are easily identifiable by the prominent metachromatic violet staining of their cytoplasmic granules. Mast cells are not present in MPNSTs. In contrast, Ki67 immunoreactivity is virtually non-existent in plexiform neurofibromas as seen in the P₀-GGFβ3 tumor shown in Figure 4E. Nuclear Ki67 labeling is typically present in a very high fraction of tumor cells as seen in the microscopic MPNST arising in the trigeminal ganglion of a P₀-GGFβ3 mouse (Figure 4F).

Figure 5 illustrates examples of the stains that we perform to fully characterize the cellular composition of neurofibromas in a newly developed GEM model. Although the stains shown in this figure were obtained in human dermal neurofibromas, they are identical in appearance to what we have seen in GEM tumors. Immunoreactivity for CD117 (c-Kit) is present in mast cells within neurofibromas and thus, has a distribution highly similar to what is seen with Unna stains (see Figure 3A). Macrophages are also present scattered throughout neurofibromas, as seen with the pan-macrophage marker Iba1 (see Figure 5D); this includes subclasses of macrophages that are immunoreactive for CD163 and CD86 (see Figure 5B and Figure 5C, respectively). A fraction of the Schwannian element in neurofibromas also demonstrates nuclear immunoreactivity for Sox10. Fibroblasts can be highlighted by their immunoreactivity for TCF4. In Figure 5G, CD31 labels the vascular elements within the neurofibroma, while CD34, demonstrated in Figure 5H, labels an enigmatic dendritic population of cells that have been suggested to be either a subpopulation of resident tissue macrophages³⁰ or a novel population of nerve sheath cells that are neither Schwann cells nor fibroblasts³¹.

Figure 6 is included to enable the comparison of human plexiform neurofibromas and MPNSTs to the tumors seen in P₀-GGFβ3 mice and to provide representative examples of some of the human tumor types that can be confused with MPNSTs. Figure 6A illustrates a plexiform neurofibroma that arose in the brachial plexus of an NF1 patient, while Figure 6B presents the WHO grade IV MPNST that arose within this same plexiform neurofibroma. Figure 6B shows some of the characteristic features of a WHO grade IV MPNST, including marked hypercellularity and cellular atypia, brisk mitotic activity, and tumor necrosis. For comparison, Figure 6C shows a WHO grade II MPNST that has significant cellular atypia but is less hypercellular than the grade IV MPNST and, although mitoses are present, demonstrates less mitotic activity. Figure 6D illustrates a fibrosarcoma with its characteristic "herringbone" pattern of interweaving sheaths of tumor cells. This pattern does not necessarily distinguish between fibrosarcomas and MPNSTs, however, because some MPNSTs will have a similar pattern. Further, the higher-power view presented in Figure 6E shows cellular morphology that is similar to that seen in the WHO grade IV MPNST presented in Figure 6B. Figure 6F illustrates a leiomyosarcoma. Unlike most MPNSTs, leiomyosarcomas are immunoreactive for muscle markers such as smooth muscle actin and desmin. Smooth muscle actin immunoreactivity can be variable from tumor to tumor, though, with some tumors demonstrating intense uniform immunoreactivity (Figure 6G) and others showing immunoreactivity that shows cellular variability within the tumor (Figure 6H). Desmin immunoreactivity can also be patchy in leiomyosarcomas (Figure 6I). Melanomas are highly variable in morphology, with some tumors being composed of polygonal cells (Figure 6J) and others being composed of spindled cells that can mimic the morphology of MPNST cells. Melanomas can be distinguished from MPNSTs by immunoreactivity for melanosome markers such as MART1 (Figure 6K). However, melanomas, like MPNSTs, frequently are positive for S100β and Sox10 (Figure 6L).

Figure 7 illustrates the pathologic features of WHO grade II, III, and IV MPNSTs isolated from P0-GGFβ3 mice. Figure 8 shows representative images of early-passage P₀-GGFβ3 MPNST cells at low (Figure 7A) and high (Figure 7B) power. The tumorigenicity of these cells is demonstrated both by their ability to form colonies when suspended in soft agar (Figure 7C) and to form grafts when allografted subcutaneously in immunodeficient mice (Figure 7D).

Figure 1: Workflow used to process tumor and other tissues from P₀-GGFβ3 mice. (A) Grossly visible tumors are harvested and segmented into three portions for 1) fixation in 4% paraformaldehyde followed by immunohistochemistry and histochemistry, 2) establishment of early passage tumor cell culture and/or genomic analyses, and 3) snap-freezing using liquid nitrogen for protein, DNA or RNA isolation. (B) After the excision of the tumor, the body of the mouse is fixed in 4% paraformaldehyde and the internal organs are removed. These organs are sampled for histologic examination performed to identify microscopic evidence of neoplasm and other pathologic processes. (C) Following removal of internal organs, the extremities (head, limbs, tail) and skin are removed from the carcass. The vertebral column, adjacent ribs, and adjacent skeletal muscle are decalcified using 0.3 M EDTA/4% paraformaldehyde (pH 8.0). The decalcified tissues are then paraffin-embedded and sections of the tissues prepared for immunohistochemical and histochemical examination. Please click here to view a larger version of this figure.

Figure 2: Representative images of grossly evident neurofibromas and MPNSTs in P₀-GGFβ3 mice. (A) P₀-GGFβ3 mouse with a large grossly evident tumor on the right flank (arrow). (B) An MRI scan of this mouse shows that the tumor is connected to the sciatic nerve (arrowhead) and that it has grown through the overlying fascia to expand within the subcutaneous (arrow, bulk tumor mass). (C) Microscopic examination of this tumor shows that, despite its large size, the tumor is a neurofibroma. (D) Large fleshy MPNST that arose in the brachial plexus of a P₀-GGFβ3 mouse. Scale bar = 100 µm. (C). Abbreviations: MPNSTs = malignant peripheral nerve sheath tumors; MRI = magnetic resonance imaging. Please click here to view a larger version of this figure.

Figure 3: Representative images of MPNSTs, decalcified vertebral column and neurofibromas prepared as described in protocol section 1. (A-J) Excised and H&E-stained P₀-GGFβ3 MPNSTs show histologic variability. All images are independently arising MPNSTs. Despite this histologic variability, these tumors all showed appropriate labeling for the MPNST markers indicated in Table 1. Scale bars = 200 µm. (K) Representative image of an H&E-stained cross-section of the decalcified vertebral column. In this image the following structures are easily visualized: the spinal cord; vertebral bone; the dorsal root ganglia on the dorsal spinal nerve root; and paravertebral skeletal muscle. Magnification 4x. (L) A higher-power image of the spinal cord and vertebra shown in K demonstrates the proper appearance of bone following decalcification with this methodology. Magnification 10x. (M) Representative image of a dorsal nerve root neurofibroma in a P₀-GGFβ3 mouse. Magnification 40x. Scale bars = 100 µm. Abbreviations: MPNSTs = malignant peripheral nerve sheath tumors; H&E = hematoxylin and eosin. Please click here to view a larger version of this figure.

Figure 4: Diagnostic stain used for the initial identification of plexiform neurofibromas and their distinction from MPNSTs. (A) Immunostains for S100β in a P₀-GGFβ3 plexiform neurofibroma. Note that intense brown staining for this antigen is only present in a subset of cells in this tumor, consistent with the fact that neurofibromas are composed of neoplastic Schwann cells and multiple other non-neoplastic cell types. (B) Immunofluorescence image of a P₀-GGFβ3 MPNST stained for the intermediate filament nestin (red) and counterstained with bisbenzimide (blue, nuclear stain). (C) MPNST stained for the transcription factor Sox10. Intense immunoreactivity (brown) is evident in the nuclei of a subset of tumor cells. (D) High-power view of a P₀-GGFβ3 plexiform neurofibroma after an Unna stain. This stain produces metachromatic (violet) staining of the granules in mast cells. Plexiform neurofibromas can be distinguished from MPNSTs because the latter lack mast cells. (E,F) Ki67 immunohistochemistry in (E) a P₀-GGFβ3 plexiform neurofibroma and (F) a P₀-GGFβ3 MPNST. Both sections have been counterstained with hematoxylin (blue nuclear staining), with Ki67 immunoreactivity being evident as brown nuclear staining in these immunoperoxidase stains. Note that no brown nuclear staining is evident in the plexiform neurofibroma, whereas most of the tumor cell nuclei are positive in the MPNST. Scale bars = 100 µm. Abbreviation: MPNST = malignant peripheral nerve sheath tumor. Please click here to view a larger version of this figure.

Figure 5: Representative images of immunostains used to identify the subpopulations of cells that compose neurofibromas. These immunofluorescent images from a human dermal neurofibroma have been stained for (A) CD117 (c-Kit; a marker of mast cells), (B) CD163 (M2 macrophages), (C) CD86 (M1 macrophages), (D) Iba1 (pan-macrophage marker), (E) Sox10 (Schwann cell marker), (F) TCF4 (fibroblast marker), (G) CD31 (marker of vasculature), and (H) CD34 (marks a distinct, poorly understood subpopulation of cells in neurofibromas). Magnification 60x, scale bars = 200 µm. Please click here to view a larger version of this figure.

Figure 6: Representative images of plexiform neurofibromas, MPNSTs, and some other human tumor types that are considered in the differential diagnosis of an MPNST. (A) Plexiform neurofibroma arising in the brachial plexus of an NF1 patient, showing the overall lower cellularity and benign appearance of this neoplasm. Although not easily visualized in H&E-stained sections, a complex mixture of cell types is present. Mitoses are not seen. Magnification: 40x. (B) A WHO grade IV MPNST that arose from the plexiform neurofibroma illustrated in A. Note the much higher degree of cellularity. The arrow indicates a mitotic figure, and the asterisk denotes a region of tumor necrosis in the upper right portion of this microscopic field. Magnification: 40x. (C) A WHO grade II MPNST. This tumor has a lower degree of cellularity than the WHO grade IV MPNST illustrated in B. However, there is more nuclear atypia and hyperchromasia than is evident in the plexiform neurofibroma illustrated in A. The arrow indicates one of the occasional mitotic figures that were encountered in this neoplasm. Magnification: 63x. (D) A low-power view of an adult-type fibrosarcoma illustrating the "herringbone" pattern (the interweaving sheaths of tumor cells) typically seen in this tumor type. Unfortunately, herringbone architecture is also encountered in some human MPNSTs and so cannot be used to distinguish between fibrosarcomas and MPNSTs. Magnification: 20x. (E) A higher-power view of the fibrosarcoma illustrated in D. Note the similarity between the cellular morphology in this fibrosarcoma and the cellular morphology evident in the WHO grade IV MPNST shown in B. Magnification: 40x. (F) High-power image of a leiomyosarcoma, demonstrating cellular morphology that falls within the range of variation seen in MPNSTs. Magnification: 40x. (G) Immunostains for smooth muscle actin in the leiomyosarcoma illustrated in F. Note that the tumor cells show uniform intense immunoreactivity for this antigen. Magnification: 40x. (H) Immunostains for smooth muscle actin in a different leiomyosarcoma. In this tumor, there is greater variability in the degree of immunoreactivity, with some cells staining more intensely than others. It is not uncommon for immunoreactivity for the same antigen to be uniformly present in one tumor and to be present only in a subset of tumor cells in a different neoplasm. Magnification: 40x. (I) Immunostain for desmin in a third leiomyosarcoma. In this tumor, only a subset of tumor cells is intensely immunoreactive for this antigen. (J) Metastatic melanoma. Melanomas are notorious for having highly variable morphology that can range from cuboidal to spindled; tumors with the latter morphology are most likely to be confused with MPNSTs, particularly given that both melanomas and MPNSTs can demonstrate S100β and Sox10 immunoreactivity. Magnification: 40x. (K) Immunoreactivity for the melanoma marker MART1 in the tumor illustrated in J. Magnification: 40x. (L) Nuclear immunoreactivity for the transcription factor Sox10 in the melanoma shown in J. Magnification: 40x, scale bars = 100 µm. Abbreviation: MPNST = malignant peripheral nerve sheath tumor. Please click here to view a larger version of this figure.

Figure 7: Representative images of WHO grade II-IV P₀-GGFβ3 MPNSTs. (A) Low- and (B) high-power photomicrographs of a WHO grade II MPNST. Note that the cellularity is lower than the WHO grade III MPNST illustrated in panel C and that the nuclei of the tumor cells in D are more hyperchromatic (darker) than those seen in panel B. (C) Low- and (D) high-power photomicrographs of a WHO grade III MPNST. This tumor demonstrated >4 mitotic figures per 10 high-power fields, with cells that were more densely packed and more hyperchromatic, atypical nuclei. (E) Low- and (F) high-power views of a WHO grade IV MPNST. Note the focus of necrosis in the bottom center portion of panel F. Low power = 20x. High Power = 40x, scale bars = 100 µm. Please click here to view a larger version of this figure.

Figure 8: Representative images of early-passage P₀-GGFβ3 MPNST cells and their tumorigenicity as demonstrated by growth in soft agar and their ability to grow as allografts. (A) Low- and (B) high-power phase contrast images of early-passage P₀-GGFβ3 MPNST cells. (C) P₀-GGFβ3 MPNST cells grown in soft agar and stained with Sudan black; the colonies are evident as black puncta in the agar. (D) Hematoxylin and eosin-stained image of a tumor that formed after P₀-GGFβ3 MPNST cells were allografted subcutaneously in an immunodeficient mouse as described in the protocol. (E) Proliferation of early-passage P₀-GGFβ3 MPNST cells over a 5-day period as determined by a Celigo Image Cytometer. Low power = 10x, high power = 40x; scale bar = 100 µm for all panels Please click here to view a larger version of this figure.

Name	Usage	Species Reactivity/Class
CD117	Prediluted	rabbit monoclonal
CD163	1:200	mouse monoclonl
CD31	1:50	rabbit polyclonal
CD34	1:2000	rabbit monoclonal
CD86	1:1000	rabbit monoclonal
Cytokeratin	1 µg/mL	mouse monoclonal
Desmin	1:50	mouse monoclonal
Iba1	1:500	polyclonal rabbit
Ki-67	1:50	rabbit monoclonal
MART1	1 µg/mL	mouse monoclonal
Nestin	1:1,000	mouse monoclonal
PMEL	1:100	rabbot monoclonal
S100B	1:200	rabbit polyclonal
SMA	1:100	mouse monoclonal
Sox10	1:10	mouse monoclonal
TCF4/TCFL2	1:100	rabbit monoclonal

Table 1: Antibodies used for the diagnosis of plexiform neurofibroma and malignant peripheral nerve sheath tumors. Antibodies used for routine identification of GEM plexiform neurofibromas (S100β, Sox10, CD117, Ki67), the diagnosis of MPNSTs, and a complete assessment of other cell types present in neurofibromas. Abbreviation: MPNST = malignant peripheral nerve sheath tumor.

		S100β	Nestin	Sox10	MART1	PMEL	Desmin	Smooth muscle actin	Cytokeratin	SS18-SSX Fusion
MPNST		50-90%, usually focal	Positive1	~30%	Negative	Negative	Negative	Negative	Negative	Negative
Fibrosarcoma		Negative	Negative	Negative	Negative	Negative	Negative	Negative	Negative	Negative
Leiomyosarcoma		Rare	Negative	Negative	Negative	Negative	50-100%	Positive	~40%	Negative
Epitheloid sarcoma		Negative	Negative	Negative	Negative	Negative	Negative	Negative	Positive	Negative
Melanoma		Positive	Positive	85%	Positive	Positive	Negative	Negative	Negative	Negative
Monophasic synovial sarcoma		~30%	Negative	Negative	Negative	Negative	Negative	Negative	Positive	Positive

Table 2: Markers used to establish tumor identity in humans. These immunohistochemical and histochemical markers, together with an assessment of tumor microscopic morphology, are used to distinguish MPNSTs from other neoplasms that mimic them. The differential diagnosis typically considered for human MPNSTs includes adult-type fibrosarcoma, epitheloid sarcoma, leiomyosarcoma, monophasic synovial sarcoma, and melanoma. Adult-type fibrosarcomas have a herringbone pattern microscopically and stain for vimentin, but not S100β. Leiomyosarcomas, but not MPNSTs, are immunoreactive for desmin; leiomyosarcomas also have nuclei with a notably blunt-ended morphology. Epitheloid sarcomas, but not MPNSTs, are immunoreactive for cytokeratin. Melanomas, like MPNSTs, may be S100β-positive, but are also immunoreactive for MART-1 and PMEL. Abbreviation: MPNST = malignant peripheral nerve sheath tumor. ¹Combination of S100β and nestin highly predictive of MPNST.

Discussion

The histological and biochemical methods presented here provide a framework for diagnosing and characterizing GEM models of neurofibroma and MPNST pathogenesis. Over the years, we have found these methodologies to be quite useful for assessing the pathology of peripheral nerve sheath tumors arising in GEM models^21,25,26. However, while the protocols outlined here are useful for determining how accurately tumors in the GEM models recapitulate the pathology of their human counterparts, there are some limitations to these strategies that should be appreciated. To begin with, P₀-GGFβ3 mice show almost complete penetrance of their tumor phenotype, which makes it relatively easy to obtain large numbers of mice with neurofibromas and MPNSTs^21,25,26 that can be used for genomic studies and genome-scale shRNA screens. The high penetrance of the phenotype in this model also makes P₀-GGFβ3 mice quite useful for preclinical trials. It should be recognized, though, that this will not be the case for all GEM cancer models. This is why we have emphasized the importance of determining the fraction of animals that can be predicted to develop tumors. When working with an animal model that less commonly develops tumors, we have found it advantageous to use imaging modalities such as magnetic resonance imaging or positron emission tomography to definitively identify animals carrying tumors that can be entered into preclinical trials or other studies. This approach can be cost-prohibitive, however, if repeated scans are required. This is why we also emphasized the importance of establishing the average survival time of the GEM model of interest²⁶-understanding when an animal can be expected to develop a tumor reduces the number of scans that are required to identify animals with the desired phenotype and associated costs.

It is also important to recognize that a large tumor in a mouse is still actually a rather small amount of tissue. In this protocol, we recommend a process in which the tumor tissue is divided into thirds, with portions devoted to histology/immunohistochemistry, establishing early passage cell cultures, and isolation of biomolecules (proteins, RNA, DNA) of interest. However, tumor size will vary and if the tumor is quite small, this can preclude having enough tissue to divide in this fashion. In that case, the investigator must determine whether tumor diagnosis or another use of tumor tissue is prioritized³². Under those circumstances, our bias is that we must have a proper diagnosis to understand what we are working with before embarking on other experiments. We have found that tumor diagnoses can usually be readily established using less than a third of the neoplasm. When dealing with small tumors, we instead will typically remove a small fragment of tumor tissue, wrap it in histology tissue and place it in a tissue cassette to ensure that the tissue is not lost during processing. The disadvantage of this approach is that it may cause the investigator to undergrade the neoplasm if WHO grading is desired; this is because lower and higher-grade regions commonly coexist in cancers and, if a very small sample is taken, the grade obtained will depend on where the tumor sample was taken from.

The protocols we describe for establishing peripheral nerve sheath tumor identity rely heavily on immunohistochemistry. While there are occasional alternative approaches that can be used for some cell types (e.g., performing Unna stains to identify mast cells), that is not uniformly true for most of the cell types present in neurofibromas and MPNSTs. Consequently, great care must be taken to establish that the immunohistochemical procedures that will be used have been properly optimized. In our experience, several issues can compromise diagnostic immunohistochemistry. It is critically important that the investigator perform a dilutional series with a primary antibody that they have not used before; in addition, a dilutional series should be performed when using a new batch of an antibody that has been previously optimized³³. Care must be taken as well to ensure that fresh batches of reagents are on hand. In our experience, when they age, hydrogen peroxide and DAB are particularly prone to not performing properly, which may lead the investigator to inappropriately decide that their stains are negative.

The immunohistochemical profiles that we describe for plexiform neurofibromas, MPNSTs, and the diverse types of neoplasms that are considered in the differential diagnosis of MPNSTs are those that we have most commonly encountered^21,25,26,34. However, as divergent differentiation is not uncommon in MPNSTs and these other malignancies, the investigator may encounter staining profiles that differ somewhat from what we describe here. These nuances are why we strongly recommend that the team characterizing a new GEM model of tumorigenesis include an experienced human or veterinary pathologist. In this protocol, we have applied WHO grading to GEM and human MPNSTs. We prefer this grading system because the grading criteria are relatively well defined and are easy to compare between human and mouse MPNSTs^2,28. We would note, though, that WHO grading criteria are not uniformly accepted by pathologists. This is because it is not clear that the three WHO grades that are applied to MPNSTs are predictive of clinical outcomes in patients. Because of that, many pathologists prefer to classify MPNSTs simply as low-grade or high-grade malignancies. Using this approach, approximately 85% of MPNSTs are considered high-grade malignancies characterized by brisk mitotic activity, marked cellular atypia, and hyperchromasia that may be accompanied by tumor necrosis. Low-grade MPNSTs show less cellularity, hyperchromasia, and mitotic activity.

We have found that allografting early passage MPNST cells is a straightforward means of verifying the tumorigenicity of these cultures. However, some issues should be considered when the cells do not establish an allograft³². In our experience, while the overwhelming majority of early-passage MPNST cells will establish an allograft, we have occasionally encountered early passage cultures that do not. Despite this failure, array comparative genomic hybridization (aCGH) and whole exome sequencing demonstrated that these early-passage cultures had multiple genomic abnormalities consistent with them being tumor cells^25,26. We have also found early-passage cultures that are not healthy, usually because the cultures were allowed to become confluent before harvesting them for grafting. Cells damaged by rough handling (e.g., incubated overly long with non-enzymatic cell dissociation solution pipetted up and down an excessive number of times) also perform poorly when grafted. We would also note that the times we have indicated for graft establishment are an estimate based on our experience. On occasion, we have encountered early-passage cultures that will successfully establish allografts but take a longer time to do so.

The approaches outlined above provide a comprehensive means of characterizing key aspects of a newly developed GEM model of peripheral nervous system neoplasia and properly diagnosing any neurofibromas and MPNSTs that these animals develop. The methods that we outline for establishing early passage MPNST cultures and using these cultures for allografting also provide clear evidence for the tumorigenicity of neoplasms identified in GEM models. We would emphasize that we are not performing the selection of individual clones when establishing early passage cultures. Although we recognize that some selection is unavoidable when placing tumor cells in cultures, these initial findings suggest that the mutations identified in early-passage GEM MPNST cultures remain highly like those present in the parent tumor. However, using aCGH, we have also found that continuing to carry these early passage cultures for more extended passages results in genomic changes that begin to diverge from what is seen in the earlier passages of these tumor cells.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
100 mm Tissue Culture Plates	Corning Falcon	353003
3, 3'- Diaminobensidine (DAB)	Vector Laboratories	SK-400
6- well plates	Corning Costar	3516
Acetic Acid	Fisher Scientific	A38-212
Alexa Fluor 488 Secondary (Goat Anti-Mouse)	Invitrogen	A11029
Alexa Fluor 568 Secondary (Goat Anti-Mouse)	Invitrogen	A21043 or A11004
Alexa Fluor 568 Secondary (Goat Anti-Rabbit)	Invitrogen	A11036
Ammonium Chloride (NH4Cl)	Fisher Scientific	A661-500
BCA Protein Assay Kit	Thermo Scientific	23225
Bovine Serum Albumin	Fisher Scientific	BP1600-100
Caldesmon	ABCAM	E89, ab32330
CD117	Cell Marque	117R-18-ASR
CD163	Leica	NCL-L-CD163
CD31	ABCAM	ab29364
CD34	ABCAM	ab81289
CD86	ABCAM	ab53004
Cell Scraper	Sarstedt	83.183
Cell Stripper	Corning	25-056-CI
Circle Coverslip	Fisher Scientific	12-545-100
Citrisolve Hybrid (d-limonene-based solvent)	Decon Laboratories	5989-27-5
Critic Acid	Fisher Scientific	A104-500
Cytokeratin	ABCAM	C-11, ab7753
Desmin	Agilent Dako	clone D33 (M0760)
Diaminobensizdine (DAB) Solution	Vector Laboratories	SK-4100
DMEM	Corning	15-013-CV
Eosin Y	Thermo Scientific	7111
Ethanol (200 Proof)	Decon Laboratories	2716
Fetal Calf Serum	Omega Scientific	FB-01
Forksolin	Sigma-Aldrich	F6886
Glycerol	Sigma-Aldrich	G6279
Hank's Balanced Salt Solution (HBSS)	Corning	21-022-CV
Harris Hematoxylin	Fisherbrand	245-677
Hemacytometer	Brightline-Hauser Scientific	1490
Hydrochloric Acid	Fisher Scientific	A144-212
Hydrogen Peroxide	Fisher Scientific	327-500
Iba1	Wako Chemicals	019-19741
ImmPRESS HRP (Peroxidase) Polymer Kit ,Mouse on Mouse	Vector Laboratories	MP-2400
ImmPRESS HRP (Peroxidase) Polymer Kit, Horse Anti-Rabbit	Vector Laboratories	MP-7401
Incubator	Thermo Scientific	Heracell 240i CO2 incubator
Isoflurane	Piramal	NDC 66794-017-25
Isopropanol	Fisher Scientific	A415
Ki-67	Cell Signaling	12202
Laminin	Thermo Fisher Scientific	23017015
Liquid Nitrogen
MART1	ABCAM	M2-9E3, ab187369
Microtome
Nestin	Millipore	Human: MAD5236 (10C2), Human:MAB353 (Rat-401)
Neuregulin 1 beta	In house	Made by S.L.C. (also available as 396-HB-050/CF from R&D Systems)
Neurofibromin	Santa Cruz Biotechnology	sc-67
NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ mice	Jackson Laboratory	5557
Nonfat Dry Milk	Walmart	Great Value Brand
P0-GGFβ3 mice	In house
Paraffin Wax	Leica	Paraplast 39601006
Parafilm M	Sigma-Aldrich	PM-999
Paraformaldehyde (4%)	Thermo Scientific	J19943-K2
Permount (Xylene Mounting Medium)	Fisher Scientific	SP15-100
pH Meter	Mettler Toldedo	Seven Excellence, 8603
Phosphate Buffered Saline (Dulbecco's)	Corning	20-031-CV
PMEL	ABCAM	EP4863(2), ab137078
Poly-L-Lysine Hydrobromide	Sigma-Aldrich	P5899-5MG
Portable Isoflurance Machine	VetEquip Inhalation Anesthesia Systems
PVA-DABCO (Aqueous Mounting Medium)	Millipore Sigma	10981100ML
Rice Cooker	Beech Hamilton
S100B	Agilent Dako	Z0311  (now GA504)
SMA	Ventana Medical Systems	clone 1A4
Sodium Chloride	Fisher Scientific	S640
Sodium Citrate (Dihydrate)	Fisher Scientific	BP327-1
Sox10	ABCAM	ab212843
Steel histology mold
Superfrost Plus Microscope Slides	Fisher Scientific	12-550-15
TCF4/TCFL2	Cell Signaling	(CH48H11) #2569
Tissue Cassette
Toluidine Blue	ACROS Organics	348600050
Triton X-100	Fisher Scientific	BP151-500
TRIzol	Invitrogen	15596026
Trypsin	Corning	25-051-31