Proof-of-principle of NF1 gene therapy in plexiform neurofibroma xenograft mouse models

Abstract

Neurofibromatosis type I is a rare neurocutaneous syndrome characterized by the development of disfiguring neurofibroma tumors with unmet clinical needs. As Neurofibromatosis Type I is a monogenic disease, the development of gene therapy is highly attractive, but it is currently unknown if rescuing the NF1 gene in established neurofibroma is sufficient for tumor regression. Here, we test this hypothesis by building two novel NF1 mouse models with reversible NF1 expression. In the first model, the human NF1 ^−/− Schwann cells named ipNF95.11b C/T are genetically modified with a doxycycline-inducible full-length mouse Nf1 gene. One month after cells implantation in the sciatic nerve, mice are split into two treatment groups and received either doxycycline or control water. Strikingly, most sciatic nerves from mice allowed to drink doxycycline water for one month do not display neurofibroma histologically (2 out of 16 sciatic nerves, 13%) whereas 75% (12 out of 16 sciatic nerves) develop or maintain a neurofibroma when drinking control water. In the second model, the human NF1 ^+/− Schwann cells named ipNF95.11 C are genetically modified with a doxycycline-inducible potent shRNA against the NF1 mRNA transcript. Strikingly, doxycycline withdrawal after neurofibroma establishment allowed limited neurofibroma maintenance (3 out of 14 sciatic nerves, 21%), whereas most sciatic nerves showed evidence of neurofibroma when kept on doxycycline (12 out of 14 sciatic nerves, 86%). Finally, intrathecal injection of the full-length Nf1 lentivirus in established 1-month ipNF95.11b C/T xenograft pNF significantly reduced tumor burden. All mice are female. Thus, we provide proof-of-principle of the efficacy of NF1 Gene Therapy in plexiform neurofibroma mice models, paving the way to the development of therapeutic gene therapy solutions for NF1 patients.

Using two novel xenograft mouse models of plexiform neurofibroma type I, the authors provide a proof-of-principle for NF1 gene therapy of this disease.

Introduction

Neurofibromas are tumors arising in the peripheral nervous system that occur in nearly all patients with Neurofibromatosis Type I (NF1), a condition caused by mutations in the NF1 tumor suppressor gene within Schwann cells^1,2. Neurofibromas are disfiguring tumors that can severely affect daily life. Surgical options are limited, especially when numerous tumors are present or when they affect vital organs. Genetic analysis of Schwann cells from neurofibromas has shown that NF1 is the sole gene consistently mutated across multiple samples, making NF1 a prime candidate for gene therapy. Gene therapy holds promise for NF1 patients, but its efficacy may take extensive research and significant funding to establish, potentially taking a decade to reach conclusions from phase II clinical trials. Despite NF1 mutations being implicated in neurofibroma formation, it remains untested whether restoring NF1 expression could disrupt tumor maintenance and serve as a definitive cure. This gap in knowledge is partly due to current NF1 mouse models, which involve permanent NF1 inactivation. Current models include mice with tissue-specific Nf1 ablation using Cre recombinase under a Schwann cell promoter (e.g., Krox-20, Dhh, Plp1, Hoxb7, Postn, Sox10), leading to para-spinal neurofibroma with high penetrance within 4–12 months^3–8. Alternatively, graft models use human or mouse cells with biallelic Nf1 or NF1 inactivation in the vicinity of a sciatic nerve of immunocompromised mice. These models confirm neurofibroma development through histological analysis after 1–4 months^8–11. However, these models are unsuitable for testing the conceptual feasibility of NF1 gene therapy due to permanent gene inactivation. To address this, novel NF1 mouse models were developed with reversible NF1 expression to investigate whether the loss of NF1 is essential for neurofibroma maintenance. We expect that re-expression of Nf1 in established neurofibroma will lead to tumor regression/tissue normalization.

Results

We began by developing a xenograft recapitulating plexiform neurofibroma (pNF) using human NF1 ^−/− Schwann cells. Briefly, 2-months old triple-immunocompromised NOD SCID mice were implanted with the NF1 ^−/− ipNF95.11b C/T cells^10,12 in the vicinity of injury-induced sciatic nerves. After three months, sciatic nerves were harvested and submitted to histological evaluation following guidelines from the literature^13,14. Compared to normal nerves, we consistently observed higher cellularity, disorganized neural architecture characteristic of neurofibroma and presence of thick collagen bundles (Fig. 1A, Supplementary Fig. 1A). These neurofibroma characteristics were not consistently observed when performing a sham surgery or xenograft with normal or NF1 ^+/− Schwann cells (Fig. 1A, Supplementary Fig. 1B-D). To ensure that the developed pNFs originate from the implanted human Schwann cells, we performed immunohistochemistry using Schwann cell-specific (S100) and human-specific (Ku80) markers (Fig. 1A, Supplementary Fig. 1A–D). Overall, pNF were only and consistently observed when ipNF95.11b C/T cells were implanted (n = 4 sciatic nerves per condition) (Fig. 1B). As athymic nude mice are more cost effective and convenient because they are hairless, we performed a side-by-side comparison of NOD SCID vs athymic nude mice. Of note, there was a similar high penetrance of pNF even at 1 month in NOD SCID mice (5 out of 5 sciatic nerves) and nude mice (19 out of 22 sciatic nerves) (Fig. 1C, Supplementary Fig. 1E–G). Since both left and right sciatic nerves of the same mouse were used for xenograft, we performed a sub-analysis to determine if there was any tumor penetrance difference between right and left sciatic nerve but we didn’t find any significance (Supplementary Fig. 1Fa). Thus, we decide to pursue with nude mice for the remaining of this study.

Fig. 1. Development of the ipNF95.11b C/T xenograft assay.

A Representative H&E (upper), S100 [(Schwann cell marker) (middle)] and Ku80 [(human-specific marker) (bottom)] and immunohistologies of sciatic nerve from NOD SCID mice implanted with ipNF95.11b C/T cells (first column), no cells (2^nd column), ipn02.3 2λ (3rd column) or ipNF95.11 C (4^th column) and harvested after 3 months. The scale bar equals 100 µm. B Tumor penetrance of the experiment in A (n = 4 sciatic nerves per condition). C Bar graph representing the comparison of tumor penetrance of the ipNF95.11b C/T xenograft assay in NOD SCID (n = 5 sciatic nerves) and nude mice (n = 22 sciatic nerves) at 1 month. Fisher’s exact test was used to measure statistical significance (* means p ≤ 0.05). ns means non-significant.

Recently, Meittinen and collaborators reported histological criteria to characterize atypical neurofibromatous neoplasm with uncertain biologic potential (ANNUBP) an intermediate tumor between the benign neurofibroma and the malignant peripheral nerve sheath tumor¹⁵. According to them, a neurofibroma must presents with nuclear atypia and at least 2 of the following criteria: loss of neurofibroma architecture, high cellularity, and/or mitotic activity (1/50 but < 3/10 high-power fields). In our study, all the neurofibroma present at least with loss of neurofibroma architecture, and high cellularity. After review, we could not find nuclear atypia in any of the 19 neurofibroma developing at 1 month in nude mice (Supplementary Fig. 1F, G). Therefore, they do not classify as ANNUBP based on Meittinen criteria and ipNF95.11b C/T graft into nude mice represents a rapid and robust pNF assay.

Next, we assessed whether the re-expression of NF1 in neurofibroma tumor cells was sufficient to alter tumor maintenance. We genetically modified the ipNF95.11b C/T cells with a lentiviral vector expressing the mouse Nf1 gene under the control of the inducible Tet-ON system (pLVX-TetOne-Nf1). To ensure that the system was working as expected, we validated Nf1 expression increased upon doxycycline induction in vitro by qPCR (Fig. 2A) and western blot (Supplementary Fig. 2A). We next performed an experiment where 2-months old immunocompromised mice were implanted with ipNF95.11b C/T TetOne-Nf1 cells. After one month (the time to develop neurofibroma), mice were allowed or not to drink doxycycline water to respectively induce or not Nf1 expression (Fig. 2B). After 2 months, all mice were euthanized, sciatic nerves were harvested and processed for histology. Strikingly, very few (2 out of 16 sciatic nerves) tumors were found in the mice re-expressing exogenous Nf1 (group 2) while sciatic nerves (12 out of 16 nerves) from control mice (group 1) still exhibited neurofibroma (Fig. 2C, D, Supplementary Fig. 2B–G). These results demonstrate that neoplastic nerve tissue can be normalized by re-expression of NF1 in a pNF xenograft model. It also indicates that doxycycline by itself have minimal pro-tumorigenic properties to NF1 ^−/− Schwann cells grafted to sciatic nerves. Importantly, we tested the capacity of ipNF95.11b C/TTetOne-Nf1 cells to form pNF within 1 month (Fig. 2B, group 3). 85% (17 out of 20 sciatic nerves) classified histologically as pNF (Fig. 2D, Supplementary Fig. 2H, I). The fact that we obtained similar results using the ipNF95.11b C/T (Fig. 1C, Supplementary Fig. 1F, G) and ipNF95.11b C/T TetOne-Nf1 without doxycycline (Fig. 2D, Supplementary Fig. 2H, I) confirms minimal leakage of the Tet-ON system. Again, we performed a sub-analysis to determine if there was any tumor penetrance difference between right and left sciatic nerve but we didn’t find any significance (Supplementary Fig. 1Fb–d). As the Ras signaling amplitude is related to the expression level of NF1, we performed immunohistochemistry using p-ERK as a surrogate to evaluate any change in Ras signaling upon re-expressing Nf1. Whereas p-ERK staining was observed in most samples evaluated from group 1 (Supplementary Fig. 2B, C) and group 3 (Supplementary Fig. 2H), not much signal was found in samples from group 2 (Supplementary Fig. 2E, F) as expected. Thus, it demonstrates that neoplastic nerve tissue can be normalized by re-expression of NF1 in a pNF xenograft model.

Fig. 2. Re-expressing mouse Nf1 normalize neurofibroma in the ipNF95.11b C/T TetOne-Nf1 xenograft model.

A Validation of the Tet-ON inducible Nf1 expression system in vitro by qPCR. Error bars represent standard deviation. A t-test was used to measure statistical significance (*** means p ≤ 0.01). B Schematic illustrating the experimental in vivo design to proof-of-principle NF1 gene therapy. C Representative H&E for groups 1, 2, 3. Dash lines circumscribed the cellular and neurofibroma architecture area. Arrows point to collagen bundles. The scale bar equals 100 µm. D Bar graph representing the comparison of tumor penetrance of group 1 (n = 16), 2 (n = 16) and 3 (n = 20 sciatic nerves). Fisher’s exact test was used to measure statistical significance (*** means p ≤ 0.01).

However, these encouraging results rely on the exogenous expression of mouse Nf1. To demonstrate the re-expression of human NF1, we developed a second xenograft strategy using the human NF1 ^+/− Schwann cell line ipNF95.11 C. We modified these cells to express a potent shRNA against NF1 (shNF1) under the inducible Tet-ON promoter. To ensure that the system was working as expected, we verified reduced NF1 expression upon doxycycline induction in vitro by qPCR (Fig. 3A). We then performed an experiment where 2-months-old immunocompromised mice were implanted with ipNF95.11 C TetON-shNF1 and allowed to drink doxycycline for one month to knockdown NF1 expression to stimulate neurofibroma development (Fig. 3B, group 3). These mice developed tumors with high penetrance (14 out of 16 nerves) (Fig. 3C, D, Supplementary Fig. 3E–G). In group 2, when tumors were fully established after one month, withdrawal of doxycycline restores normal nerve histology (3 out of 14 nerves) (Fig. 3C, D, Supplementary Fig. 3C, D). In contrast, most nerves (12 out of 14 sciatic nerves) from control mice (group 1) exhibited neurofibroma at the same time point (Fig. 3C, D, Supplementary Fig. 3A, B). Again, we performed a sub-analysis to determine if there was any tumor penetrance difference between right and left sciatic nerves but we didn’t find any significance (Supplementary Fig. 1Fe–g). Similarly to the first model developed in Fig. 2, we performed immunohistochemistry using p-ERK as a surrogate to evaluate any change in Ras signaling upon re-expressing human NF1. Whereas p-ERK staining was observed in most samples from group 1 (Supplementary Fig. 3A) and group 3 (Supplementary Fig. 3E, F), not much signal was found in samples from group 2 (Supplementary Fig. 3C) as expected. Thus, we proof-of-principle NF1 Gene therapy for pNF by re-expressing human NF1 in a genetic model.

Fig. 3. Re-expressing human NF1 normalize neurofibroma in the ipNF95.11 C TetON-shNF1 xenograft model.

A Validation of the Tet-ON inducible shNF1 expression system in vitro by qPCR. Error bar represent standard deviation. A t-test was used to measure statistical significance (*** means p ≤ 0.01). B Schematic illustrating the experimental in vivo design to proof-of-principle NF1 gene therapy. C Representative H&E for groups 1, 2 and, 3. Dash lines circumscribed the cellular and neurofibroma architecture area. Arrows point to collagen bundles. The scale bar equals 100 µm. D Bar graph representing the comparison of tumor penetrance of group 1 (n = 14), group 2 (n = 14) and group 3 mice (n = 16 sciatic nerves). Fisher’s exact test was used to measure statistical significance (*** means p ≤ 0.01).

Beyond our genetic demonstration in xenografts, the next step is to validate a therapeutic NF gene therapy option in pre-clinical models. In this sense, we hypothesize that the lentiviral particles used to generate our ipNF95.11b C/T TetOne-Nf1 cell line can be delivered to mice by an intrathecal injection and restore Nf1 expression in the peripheral nervous system including Schwann cells. In this sense, we performed an experiment where immunocompromised mice were implanted with ipNF95.11b C/T cells. After one month (the time to develop neurofibroma), an intrathecal injection of pLVX-TetOne-Nf1 (n = 16 nerves) or control pLVX-TetOne-GFP (n = 16 nerves) was performed, and mice were allowed to drink doxycycline water to respectively induce Nf1 or GFP expression (Fig. 4A). One month after lentivirus injection, all mice were euthanized, sciatic nerves were harvested and processed for histology. As expected, only a minority of sciatic nerves (43%, 7 out of 16 nerves) show tumors in the pLVX-TetOne-Nf1 lentivirus treated mice (Fig. 4B, C, S4C-D). Importantly, we obtained similar neurofibroma penetrance in mice treated with pLVX-TetOne-GFP lentivirus and doxycycline (group 1) (83%, 13 out of 16 nerves) (Fig. 4B, C, S4A, B) compared to mice grafted with the same cells but without any lentivirus treatment (Fig. 1C, Supplementary Fig. 1F, G). Again, we performed a sub-analysis to determine if there was any tumor penetrance difference between right and left sciatic nerve but we didn’t find any significance (Fig. Supplementary Fig. 1Fh-i). Altogether, these results demonstrate that neoplastic nerve tissue can be normalized by intrathecal introduction and re-expression of Nf1 lentivirus in a pNF xenograft model.

Fig. 4. Intrathecal injection of a Nf1 lentivirus normalize neurofibroma in the ipNF95.11b C/T xenograft model.

A Schematic illustrating the experimental in vivo design to proof-of-principle NF1 gene therapy. B Bar graph representing the comparison of tumor penetrance of group 1 and group 2 mice (n = 16 sciatic nerves per condition). Fisher’s exact test was used to measure statistical significance (* means p ≤ 0.05). C Representative H&E (upper row) and GFP (lower row)] immunohistology of sciatic nerves from nude mice implanted with ipNF95.11b C/T cells for groups 1 and 2. Dash lines circumscribed the cellular and neurofibroma architecture area. Arrows point to collagen bundles. The scale bar equals 100 µm.

Discussion

There is significant enthusiasm in the NF1 scientific community for developing a gene therapy to reverse the clinical manifestation of NF1. However, proof-of-principle for NF1 gene therapy in animals for any NF1 clinical manifestation has yet to be demonstrated, partly due to the absence of suitable models. The workhorses of the field are tissue-specific Nf1 knockout mouse models generating para-spinal neurofibroma with high penetrance, a rare human neurofibroma subtype. All these models rely on the same DNA alteration in Nf1 exon 31¹⁶. Here, we took advantage of the commercially available ipNF95.11b C/T and ipNF95.11 C to build novel mouse models developing pNF in the peripheral nerves as generally occurring in NF1 patients. Leveraging the Tet-On system, we have shown convincingly that re-expressing full-length human or mouse neurofibromin is sufficient to normalize peripheral nerves. Although we used immortalized human cells that retained the characteristics of primary cells in vitro¹⁰, and the resulting tumor faithfully recapitulated pNF, we are aware the grafted ipNF95.11b C/T may not fully represented the corresponding primary cells in our xenograft assay. To make sure tumors developed from the grafted human cells, we successfully performed staining using the human-specific antibody against Ku80. Intriguingly, staining intensity was consistently lower when the resulting sciatic nerves were examined one month after implantation. The reason behind this observation is unclear but does not impair the conclusion that re-expression of Nf1 by three different means (Figs. 2–4) decreases Ras signaling (p-ERK) and normalizes nerve tissue (H&E).

The next milestone consists of developing a pharmacological gene therapy approach to cure spontaneously developing pNF in germline mouse models. Toward this goal, a single dose one-month treatment of a Nf1 lentivirus was sufficient to significantly reduce tumor penetrance in our ipNF95.11b C/T xenograft assay. These results validate the intrathecal injection approach and the use of lentivirus as a potential strategy for NF patients. Further vector optimization, dose scheduling and evaluation of the duration of effect are beyond the scope of this initial proof-of-principle but required steps to complete pre-clinical studies. It is possible that a longer exposure time may be required to maximize potency^17,18. Although we use female mice grafted with a male cell line (ipNF95.11b C/T), we speculate that using a different combination (e.g., a male host mouse with a female or male cell line) should yield similar conclusions, as pNF is equally observed in males and females¹⁹.

One of the intriguing questions in the NF field is to what extent the expression level of NF1 needed to be rescued to regress tumor/normalize tissue. Nf1 ^+/− mice do not undergo loss-of-heterozygosity in the Schwann cell lineage²⁰ but it is unclear if NF1 ^+/− human Schwann cells do so in mice as it is the case in NF patients. In our xenograft, the NF1 ^+/− Schwann cell ipNF.95.11 C do not form tumor even after 3 months (Fig. 1A) but further repression by shRNA is sufficient to induce pNF within one month (Fig. 3C). Thus, it indicates that the required NF1 expression level to develop pNF is located somewhere between NF1 ^+/− and NF1 ^−/−.

Traditionally, neurofibromin was primarily viewed as a Ras GTPase-activating protein (GAP), and technical challenges in cloning the full-length 12 kb NF1 gene into expression vectors led many researchers to use the NF1-GRD as a substitute. However, introducing NF1-GRD in NF1 ^−/− Schwann cells partly rescues the Ras-GTP levels²¹ and cell proliferation²². A recent in vivo study indicates that NF1-GRD can alter/delay tumor growth of NF1 ^−/− cells²³ but it is unclear if this is sufficient to regress established neurofibroma and hence, suitable for NF1 gene therapy. An alternative strategy was recently put forward. Around 20% of NF1 mutations translate into pre-maturation stop codon mRNAs targeted for the nonsense mediated decay (NMD) mechanism, and hence lower NF1 expression. Ataluren is a small-molecule drug with proven efficacy in suppressing NMD. The group of Allan Jacobson hypothesized that ataluren treatment in tumor cells harboring an NF1 mutation creating a premature stop codon would lead to an increase in NF1 expression and, ultimately, milder clinical manifestations in NF1 patients. Towards this goal, they demonstrated that mouse neuronal cells with an Nf1 ^R683X/R683X mutation treated with ataluren can promote readthrough of the nonsense mutation at codon 683 of Nf1 mRNA in vitro but only show a modest increase in NF1 expression level²⁴. In vivo, results using the Dhh-cre Nf1 ^R683X/4F mice indicated that ataluren might slow the growth of neurofibromas and alleviate some paralysis phenotypes²⁵. Importantly, ataluren is not specific to NF1 mRNA and, hence, cannot be used to address the question of whether NF1 re-expression in established neurofibroma is sufficient to make tumor regress.

Another study investigated the concept of nonsense suppression in a novel NF1 animal model harboring a mutation creating a premature stop codon: the NF1 ^R1947/+ Ossabaw minipig model²⁶. In vitro, NF1 ^−/− Schwann cells isolated from neurofibroma and treated with any established nonsense suppression drugs (ataluren, gentamicin, or G418) did not reliably increase NF1 expression although some reduction in MAPK signaling was observed. These findings highlight the challenges and potential strategies for developing effective gene therapies for NF1.

We have established a proof-of-principle for NF1 gene therapy in plexiform neurofibroma mouse models using both genetic and pharmacological approaches. The concepts and methodology developed in this study may be extended to demonstrate the feasibility of NF1 gene therapy to address other clinical manifestations of the disorder. Ultimately, this research paves the way for developing pharmacological therapies that could benefit NF1 patients.

Methods

Mice husbandry

All mice NOD SCID (Charles River, #394) and athymic nude (Charles River, #490) are female and were housed in the animal facility of the Institut de recherche sur le Cancer de l’Université de Sherbrooke, Quebec, Canada and are 2-months old at the time of xenograft surgery. All procedures have been approved by the Animal Research Ethics Committee of the Faculty of Medicine and Health Sciences (FMSS) of the Université de Sherbrooke and strictly follow the Canadian Council on Animal Care (CCAC) requirements. We have complied with all relevant ethical regulations for animal use. All mice were housed in hermetically sealed cages and placed on a support equipped with a ventilation system. All mouse strains were maintained in a room with 12/12 (day/night) light cycle with a temperature of 70–72 °F.

Xenograft assay

Prior to surgery, we performed the following steps: 2-months old mice were anesthesized using 5% isoflurane (oxygen = 1–2 L/min), followed by a maintenance dose of 2% isoflurane. When using NOD SCID mice, back hair was shaved. Then, local skin anelgesia was performed using a 1:1 volume of Bupivacaïne (0,5%w/v) and Lidocaïne (2% v/v) diluted in sterile water and skin was desinfected using diluted chlorhexidine. Next, under constant isoflurane anesthesia, mice skin was cut open above the left quadriceps. The left sciatic nerve was located and damaged longitudinally over 5 mm using a 27 G needle to create physical space between nerve fibrils. 1 × 10⁶ cells in 100 uL of L-15 culture media (Wisent, 323-050-CL) were implanted in the vicinity of the left sciatic nerve. The skin was sutured with monocryl 4-0 (absorbable). The procedure was repeated similarly for the right sciatic nerve. Finally, mice were allowed to recover on a warm pad until fully awakened and placed back into a cage.

Intrathecal injection

Under constant isoflurane anesthesia, the injection site at the L5-L6 lumbar vertebra was cleaned, and sterilized with 70% ethanol, to minimize the risk of infection. A 30 G needle was inserted between the L5 and L6 vertebrae, and once the needle reaches the intradural space, 30 µL of 45 um-filtered and concentrated lentiviral particles (estimated at 1 × 10¹² particles/mL by spectrophotometry reading at OD260) were injected.

Cell lines

The hTERT and mCdk4 immortalized version of the ipn02.3 2λ (CRL-3392), ipNF95.11b C/T (CRL-03390), ipNF95.11 C (CRL-3391) initially developed by the Wallace lab¹⁰ were purchased from ATCC using the Neurofibromatosis Therapeutic Accelerating Program (NTAP) refund program. Cells were passaged and expanded in DMEM (Corning, 10-030-CV) + 10% FBS (Corning, 35-022-CV) at 37 °C in 5% CO₂.

The pLVX-TetOne-Nf1 vector was constructed by cloning the Nf1 mouse open reading frame from the host vector (Genecopoeia; EX-Mm04084-M02, NM_010897.2) in place of the GFP open reading frame of the recipient pLVX-TetOne-GFP vector (addgene; #171123) by Civic Bioscience Limitée (Simon Roy). More specifically, a 9637 bp fragment consisting of the 573 base pair rabbit b-globin intron joined in 5‘ of the Nf1 open reading frame was inserted into the BamHI cut pLVX-TetOne-GFP vector by high fidelity DNA assembly and transformed into Trans2-blue chemically competent bacterial cells (Transgenbiotech; CD411), resulting in a 18,8 kb vector. As it is acknowledged in the literature that full length Nf1 is toxic for bacteria and/or prone to mutation^27,28, full vector sequencing was performed by the Rnomic platform at UdeS using Oxford Nanopore technologies and confirms the 100% match between the ORF of our final vector (pLVX-TetOne-Nf1) and the ORF of NM_010897.2 (Supplementary Data 1). In addition to the experimental sequencing results, we also provide the theoric sequence (Supplementary Data 2).

The pLKO-TetON-shNF1 was constructed by cloning the following DNA oligonucleotides (shNF1_FWD 5‘-CCGGTAAGCGGCCTCACTACTATTTCTCGAGAAATAGTAGTGAGGCCGCTTATTTTTG-3‘ and shNF1_REV 5‘- AATTCAAAAATAAGCGGCCTCACTACTATTTCTCGAGAAATAGTAGTGAGGCCGCTTA-3‘) into the recipient Tet-pLKO-puro vector (addgene; #21915) using standard protocol from the manufacturer.

Stable cell line establishment was performed by infecting recipient cells with lentiviral particles generated from the desired inducible vector and the following 3 packaging vectors: plp1 (addgene; #6097); plp2 (addgene; #6098); plp/VSVG (addgene; #6099) in HEK293T cells followed by puromycin (Bioshop, PUR333) selection.

Doxycycline treatment

Doxycycline drinking water for mice was prepared by diluting 1 g of doxycycline (Sigma, D9891) in 500 mL of autoclaved water containing 5% sucrose (Bioshop, SUC507). In Figs. 2 and 4, doxycycline was started at 3 months of age. In Fig. 3, doxycycline was started 24 h after cell implantation. For in vitro studies, cells were treated with doxycycline at a final concentration of 2 ug/uL and incubated for 3 days.

Real-time PCR

RNA extraction was performed using the Qiazol reagent (Qiagen, 79306) following the manufacturer’s recommendation. The resulting total RNA extract was quantified using Nanodrop, and 150 ng from each sample were submitted to a reverse transcription reaction using Roche Transcriptor™ (Sigma-Aldrich, 41106313) and N6 Roche random (Sigma-Aldrich, 11034731001) primers in a 10 uL reaction. Finally, cDNA was diluted with 440 uL of RNase DNase free water and used as template for a qPCR reaction using NF1-specific primers (mouse-Nf1-FWD 5‘-GCAACTTGCCACTCCCTACTGA, mouse-Nf1-REV 5‘-ATGCTGTTCTGAGGGAAACGCT-3‘; human-NF1-FWD 5‘- AAAGGATCCCACTTCCGGTG-3‘; human-NF1-REV 5‘-CTTGGTCGCTCTCCCCACTA-3‘) and GAPDH as housekeeping gene (human-GAPDH-FWD 5‘-GTGAAGGTCGGAGTCAACGGATTT-3‘, human-GAPDH-REV 5‘-TGCCATGGGTGGAATCATATTGGA-3‘) in the 2X SyBr Green mix buffer (Quantabio, 95054-02 K) under the following cycling conditions: 95 °C, 3 min; [(95 °C, 15 s, 60°C, 30 s, 72 °C, 30 s) X 50], 72 °C, 30 sec. Relative expression of three biological replicates run in technical triplicates was calculated using qBase²⁹ and individual data points to build Figs. 2A, 3A can be found in Supplementary Data 3.

Histology

Mice were euthanized under isoflurane, followed by CO₂ asphyxia. Sciatic nerves were harvested and fixed in 10% formaldehyde (Sigma, HT501128-4L) for a minimum of 48 h. Tissues were washed in 70% ethanol, paraffin-embedded, and cut at 5 um. H&E slides were prepared from the resulting formalin-fixed paraffin-embedded (FFPE) slides by the Electron Microscopy and Histology Research Core of the Faculté de Médecine et des Sciences de la Santé at the Université de Sherbrooke and scanned on Hamamatsu nanozoomer. Tissues were classified as pNF when the 3 histological criteria were observed: higher cellularity compared to normal nerves, disorganized neural architecture characteristic of neurofibroma, and presence of thick collagen bundles.

Immunohistochemistry (IHC) of the sciatic nerves was performed based on the Vectastain ABC kit (Biolynx, VECTPK4000) and the DAB substrate kit peroxidase (Biolynx, VECTSK4103100) following the manufacturer’s recommendation. A boiling pH = 6 acetate buffer as the antigen retrieval solution and the following primary antibodies [(anti-S100, GA50461-2); (anti-Ku80, NEB 2180S, 1:200); (anti-pERK, Cell Signaling 4370S, 1:200); (anti-GFP, Roche 1181446001, 1:250)] and secondary antibodies (anti-rabbit-biotin (Jackson Immunoresearch, 711-065-152, 1:1000) (anti-mouse-biotin (Jackson Immunoresearch, 711-065-151) were used. The pathologist (S.G.) was blinded to the mice experimental group when performing the histological review.

Western-blot

ipNF95.11b C/T Tet-One-Nf1 cells were treated or not with doxycycline for 72 h. Proteins were extracted using RIPA+ buffer containing complete mini protease inhibitor 1X (Sigma, 20-188) and dosed using the BCA assay (Thermo Fisher Scientific 23223) according to the manufacturer recommendation. Proteins were resolved on a 8% SDS-PAGE and transferred on a nitrocellulose membrane. The membrane was then blocked with 5% milk and incubate with primary antibodies (anti-NF1, Bethyl A300-140A, 1:1000); (anti-actin, Cell Signaling Technology #4970, 1:1000) followed by secondary antibody [anti-rabbit-HRP (JacksonImmunoresearch, 711-035-152)]. Chemiluminesence was produced using the Clarity Western ECL (Biorad, 1705061) and imaging was performed on an iBright system (Thermo Fisher Scientific).

Statistics and Reproducibility

For all tumor penetrance comparison (Figs. 1B, 1C, 2D, 3D, 4B), a Fisher’s exact test (one-sided) was used to measure statistical significance (p ≤ 0.05) and the data are presented as stacked histograms to visualize the sample size. Tumor penetrance is an ordinal variable: tumor or normal. In most cases, each mouse was implanted with cells in the left and right sciatic nerves to generate two biological replicates per mice. Therefore, each sciatic nerve is a biological replicate (n = ) and the unit.

For all gene expression validation (Figs. 2A, 3A), a t-test was used to measure statistical significance (p ≤ 0.05) using Graph pad and the data are presented as bar graph with individual data points representing three biological replicates (different cell passage) to visualize the sample size.

Ethics and inclusion statement

We conduct research ethically and ensure that our data collection and analysis, animal experiments and authorship follow best practices.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Conceptualization: J.P.B.; Experiments: D.H.B., C. Plante, M.C., S. Gascon. Preliminary results: J.P.S.V., C. Poirier, M.L., M.D., B.L.; Histological evaluation: D.H.B, J.P.B., S. Geha. Supervision: J.P.B., B.L., B.C.; Draft writing: J.P.B.; Editing: J.P.B., B.L., B.C.; Funding: J.P.B.

Peer review

Peer review information

Communications Biology thanks Jody Fromm Longo and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Primary Handling Editors: Rosie Bunton-Stasyshyn & Benjamin Bessieres. A peer review file is available.

Data availability

Numerical source data for the bar graphs (Figs. 2A and 3A) and stacked histograms (Figs. 1B, 1C, 2D, 3D and 4B) can be found in Supplementary Data 3. All other data is available from the corresponding author on reasonable request.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s42003-026-09695-8.