Econazole selectively induces cell death in NF1-homozygous mutant tumor cells

Summary

Cutaneous neurofibromas (cNFs) are tumors that develop in more than 99% of individuals with neurofibromatosis type 1 (NF1). They develop in the dermis and can number in the thousands. cNFs can be itchy and painful and negatively impact self-esteem. There is no US Food and Drug Administration (FDA)-approved drug for their treatment. Here, we screen a library of FDA-approved drugs using a cNF cell model derived from human induced pluripotent stem cells (hiPSCs) generated from an NF1 patient. We engineer an NF1 mutation in the second allele to mimic loss of heterozygosity, differentiate the NF1^+/− and NF1^−/− hiPSCs into Schwann cell precursors (SCPs), and use them to screen a drug library to assess for inhibition of NF1^−/− but not NF1^+/− cell proliferation. We identify econazole nitrate as being effective against NF1^−/− hiPSC-SCPs. Econazole cream selectively induces apoptosis in Nf1^−/− murine nerve root neurosphere cells and human cNF xenografts. This study supports further testing of econazole for cNF treatment.

Graphical abstract

Highlights

• •There is no effective medical treatment for NF1 cutaneous neurofibromas (cNFs)
• •Econazole inhibits NF1-mutant cell proliferation in a screen of FDA-approved drugs
• •Econazole also induces apoptosis in cNF xenograft models
• •This study supports further testing of econazole for treatment of cNF

Cutaneous neurofibromas (cNFs) develop in the skin of more than 99% of patients with neurofibromatosis type 1 (NF1) and greatly affect quality of life. There is no FDA-approved drug treatment. In a screen of FDA-approved drugs using human NF1-mutant cells, Lakes et al. find that econazole selectively induces apoptosis in cNF xenograft models.

Introduction

Neurofibromatosis type 1 (NF1) is one of the most common inherited disorders, affecting one in every 3,000 individuals.¹ The disorder results from a loss of function mutation in the NF1 tumor suppressor gene encoding neurofibromin. Clinically, NF1 is characterized by widespread tumor formation in the peripheral and central nervous system, musculoskeletal anomalies, developmental disabilities, neurological deficits, cutaneous lesions, and increased risk of other malignancies.² NF1 can be inherited in an autosomal-dominant manner or acquired de novo, with subsequent loss of heterozygosity in the tumor cells of origin leading to tumorigenesis.

The development of neurofibromas is a cardinal feature of NF1. Neurofibromas are peripheral nerve sheath tumors that can be located in the dermis (cutaneous neurofibromas [cNFs]) or along larger peripheral nerves (plexiform neurofibromas [pNFs]). While many symptoms of NF1 present with phenotypic variability, cNF are found in more than 99% of individuals with NF1.³ They present as round, well-circumscribed, skin-colored nodules that can rapidly increase in number and size with disease progression, although eventually cNFs become quiescent with little to no growth. While cNFs are made up of many different cell types, Schwann-lineage cells (SLCs) are abundant and considered to be the tumor cells of origin.^4,5,6,7 That tumor regrowth can occur following incomplete resection of the tumor indicates that NF1^−/− Schwann cells have the ability to undergo additional rounds of proliferation.^7,8 Additionally, secretion of c-Kit from NF1^−/− Schwann cells plays an important role in multicellular recruitment of NF1^+/− cells to the tumor microenvironment.^9,10 Although they are not associated with malignancy, most patients with NF1 consider cNFs to be the most burdensome feature of the disorder due to visible disfigurement, unaesthetic appearance, pain and itching, and social stigma.¹¹

Despite the morbidity and negative impact on patient quality of life, there are currently no US Food and Drug Administration (FDA)-approved treatments for cNFs. The only options for patients are surgical removal or laser ablation, which can result in scarring or tumor regrowth at the tumor site.⁷ Additionally, due to the sheer number of cNFs on most patients, complete removal of all tumors is challenging. Loss of NF1 results in RAS (rat sarcoma) overactivation and dysregulation of downstream signaling pathway components, including the mitogen-associated protein kinase (MAPK) pathway, an obvious therapeutic target. Indeed, experiments in preclinical models have demonstrated that MEK (mitogen-activated protein kinase kinase) inhibition could be effective against pNFs,^8,12,13 which was borne out in clinical trials; in 2020, the FDA approved selumetinib, an oral MEK inhibitor, for treatment of symptomatic, inoperable pNFs in children.^14,15 So, while there have been advances in medical treatment for pNFs, the therapeutic options for cNFs remain limited. Considering the patient burden, the development of medical treatments that target cNF growth remains a clinical need. Specifically, the availability of topical treatments for cNFs would be particularly desirable because local application limits systemic exposure and prevents major adverse side effects.

In this study, we report a patient cell-based drug-screening platform and its use to screen FDA-approved drugs for their potential to be repurposed for the treatment of NF1-driven cNFs. We generated induced pluripotent stem cells (iPSCs) from NF1 individuals, introduced a mutation on the second NF1 allele to mimic loss of heterozygosity seen in neurofibroma development, and differentiated these iPSCs into proliferating Schwann cell precursors (SCPs). As would be done in a personalized medicine approach, we then used one of these lines to screen 2,570 FDA-approved drugs for their specific effects on NF1^−/− SCPs. One compound, econazole, stood out because it induced cell death at low concentrations. We found that econazole exhibits cell-type-specific effects on proliferation and survival of SCPs compared with differentiated Schwann cells and appears to work through a mechanism different from the approved pNF drug selumetinib. Last, we show that econazole induces apoptosis in human cNF tissue when topically applied to human neurofibroma biopsies in two ex vivo models. Thus, we propose that econazole may hold therapeutic potential for the safe and effective treatment of cNFs.

Results

Patient recruitment and generation of a human iPSC model of NF1

We recruited four individuals who had been diagnosed previously with NF1 and collected blood samples from each of them. Each patient completed a health questionnaire stating their disease history and current NF1-related symptoms. All blood samples and medical information were de-identified before proceeding. The recruitment period went on over a period of about a year. Peripheral blood mononuclear cells (PBMCs) from the patients’ blood samples were reprogrammed to iPSC using Sendai virus-mediated gene delivery. Five pluripotent stem cell lines per individual were established and stored. One cell line per individual was chosen, expanded, and stored for further experiments. We observed that the NF1^+/− iPSC lines exhibited a faster proliferation rate (i.e., shorter doubling time) compared with the pool of Harvard Stem Cell Institute (HSCI) iPSC lines, as expected, because NF1 acts as a tumor suppressor (Figure S1A). Patient 2 was chosen to be the primary subject for the drug discovery experiments. Chromosome analysis showed a normal karyotype of the iPSCs (Figure S1B). A second loss-of-function mutation was introduced into the cell line via CRISPR-Cas9 technology. We specifically targeted the wild-type allele, leaving the originally mutated allele untouched. Furthermore, the gene-targeting strategy was designed in a way that it disrupted and partly removed the GTPase-activating protein-related (GAP-related) domain (GRD) (Figure S1C), a functional domain that is responsible for the tumor suppressor function of NF1. Successful gene targeting was confirmed at the DNA and protein levels (Figure S1D).

Generation of target cells/neurofibroma-like cell types

While neurofibromas are heterogeneous tumors, SCPs have been reported to be the tumor cells of origin that initiate tumor formation following loss of NF1.^4,5,6,7 We established robust protocols that manipulate specific pathways in iPSCs with small chemical compounds to efficiently generate SCP-like cells (Figure 1A).^16,17,18 The differentiated cells express key markers of the Schwann cell lineage, can be maintained in a proliferative state for an extended amount of time, and have the ability to uniformly mature into differentiated Schwann cells (Figures 1B–1D). Additionally, the iPSCs-SCPs have multipotent, self-renewing capacity, as demonstrated by their ability to form spheres in culture (Figure 1C). We consider these cells to be a viable model to study human neurofibroma in vitro. Notably, we observed that NF1^−/− SCPs proliferated significantly faster than the isogenic NF1^+/− controls (Figure 1E) and that, under high-density culture conditions, NF1^−/− but not NF1^+/− SCPs formed 3D clusters (Figure 1F), indicating loss of contact inhibition. Accelerated proliferation and loss of contact inhibition are both classic features of tumor cell lines as well as tumors in vivo.

Figure 1.

NF1^−/− hiPSC-SPCs exhibit characteristics of tumor cells

(A) Diagram showing the protocol for differentiation of human induced pluripotent stem cells (hiPSCs) into Schwann cell precursors (SCPs).

(B) NF1^−/− hiPSCs differentiated to SCPs express the indicated SCP markers.

(C) Bright-field images showing uniform differentiation of NF1^−/− hiPSCs into SCPs.

(D) hiPSCs differentiated to mature Schwann cells express SOX10 (red) and S100β (green), markers of the Schwann cell lineage. Scale bars, 100 μm.

(E) NF1^−/− hiPSC-SCPs proliferate faster than NF1^+/− hiPSC-SCPs. To measure cell proliferation, a fixed number of cells was plated on day 0 in several wells, and then the cells were dissociated and counted on subsequent days as indicated. Data represent mean ± SEM, n = 5.

(F) NF1^−/− hiPSC-SCPs, but not NF1^+/− hiPSC-SCPs, form 3D clusters. Scale bars, 20 mm.

∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. See also Figure S2.

Chemical compound screen with FDA-approved drugs

NF1^−/− SCPs proliferate significantly faster than their NF1^+/− counterparts (Figure 1E). Because this was reproducibly observed, we chose to use this difference in proliferation rate as the screening assay readout. We scaled up the in vitro setup to a 96-well plate format and confirmed intra- and inter-plate repeatability of our screening procedures, a prerequisite for a reliable compound screen. The timeline for the screen is illustrated in Figure 2A and was as follows. On day −1, we transferred SCPs from maintenance medium to screening medium and plated 1,500 cells per well on the 96-well assay plates. On day 0, wells designated to measure the starting cell number were treated with the Hoechst nuclear stain and imaged on a plate reader. The remaining wells were then treated with a drug or the positive/negative controls. Each compound was tested on both NF1^+/− and NF1^−/− SCPs, each in duplicate, adding up to 4 measurements per compound. On days 1 and 2, the cells were left to incubate with the drug, and on day 3, the cells were stained with Hoechst and imaged on a plate reader. Nine images per well were taken and subsequently analyzed using CellProfiler software to calculate absolute cell numbers. Proliferation under different compounds was measured in relation to the unperturbed growth in the negative control and was then compared between NF1^+/− and NF1^−/− cells. Because filtering the results for either absolute (Figure 2B) or relative (Figure 2C) difference in cell proliferation reduction yields different sets of compounds, we devised a system to score each compound based on both absolute and relative differences between NF1^−/− SCPs and NF1^+/− controls. The 50 highest-scoring compounds (Figure 2D; Table S1) were chosen for further validation and evaluation.

Figure 2.

A screen for novel therapeutics for neurofibroma using hiPSCs identifies promising candidates

(A) Schematic of the timeline for the screen.

(B) Scatterplot of absolute difference of cell proliferation in NF1^−/− compared with NF1^+/− hiPSC-SCPs following treatment with compounds.

(C) Scatterplot of relative difference of cell proliferation in NF1^−/− compared with NF1^+/− hiPSC-SCPs following treatment with compounds.

(D) Scatterplot with the 50 highest scoring compounds indicated.

n = 2. See also Table S1.

Hit to lead validation

All 50 selected compounds were first re-tested in a 5-point dose-response assay in a concentration range from 1 nM to 100 μM to confirm the results of the primary screen and to identify the concentration range where the respective compound showed the highest differential effects on NF1^+/− and NF1^−/− cell proliferation (data not shown). Twenty-one compounds that were successfully confirmed were chosen for validation in a 10-point dose-response assay, with concentration ranges tailored to the individual compound (Figure 3). Every compound that caused a significant decrease in proliferation of NF1^−/− SCPs compared with NF1^+/− controls was deemed a positive outcome. This generated a final list of the nine most promising drug candidates: in alphabetical order, abemaciclib, centrimonium bromide, cobimetinib, econazole, fenretinide, methotrexate, pixantrone maleate, pyrvinium, and topotecan.

Figure 3.

Hit to lead validation with the 21 most effective compounds

Twenty-one compounds were chosen for further analysis in an extended dose-response assay. Data represent mean ± SEM, n = 3 technical replicates using the same pair of cell lines.

Following NF1 loss of heterozygosity, SCPs can initiate neurofibroma formation while differentiated Schwann cells cannot.^5,19,20 Therefore, all non-chemotherapeutics from the final list were selected to be tested for whether their effects on cell proliferation were specific for SCPs. Each of these four compounds, as well as the MEK inhibitors cobimetinib and selumetinib, was tested on both NF1^−/− and NF1^+/− SCPs and differentiated Schwann cells. One compound, econazole, exhibited strongly selective effects on the proliferation of SCPs while having minimal impact on differentiated Schwann cells (Figures 4A–4C). Nuclear Hoechst staining after incubation with econazole confirmed the cell death of SCPs and survival of differentiated Schwann cells after drug treatment (Figure 4D).

Figure 4.

Selective effects of econazole on NF1^−/− SCPs

(A and B) The indicated compounds were tested for their ability to affect cell proliferation of NF1^+/− hiPSCs differentiated to Schwann cells relative to NF1^+/− hiPSC-SCPs (A) and cell proliferation of NF1^−/− hiPSCs differentiated to Schwann cells relative to NF1^−/− hiPSC-SCPs (B).

(C) The indicated cells were treated with six different drugs at the indicated doses, and the relative cell proliferation was measured.

(D) hiPSC-SCPs (SCPs) and hiPSC-SCPs differentiated to Schwann cells (Schwann cells) treated with vehicle control (DMSO), 0.2 μM econazole, or 1.6 μM econazole for 72 h were stained with Hoechst to label cell nuclei. Scale bar, 100 μm.

Data represent mean ± SEM, n = 3 independent experiments using hiSCPs from the same differentiation. Rel., relative.

Further characterization showed that econazole treatment inhibits cell proliferation (Figure 5A) and induces apoptosis (Figure 5B), as shown by Ki67 and TUNEL (terminal deoxynucleotidyl transferase dUTP nick end labeling) staining, respectively. Crucially, econazole does not decrease levels of phosphorylated ERK (extracellular signal-regulated kinase) in SCPs and, thus, appears not to act by blocking MEK-ERK signaling (Figures 5C and 5D), unlike the approved drug selumetinib. This makes econazole an interesting candidate for combination treatment with selumetinib as a potentially more effective neurofibroma therapy. When we tested this combination treatment, we found that econazole plus selumetinib acts additively to inhibit SCP proliferation (Figure 5E). Additionally, as shown in Figure 1, NF1^−/− but not NF1^+/− SCPs lose contact inhibition and form 3D clusters when kept under high-density conditions for several weeks. An overview of the experimental timeline is shown in Figure S2A. We found that repeated econazole treatment inhibited NF1^−/− specific cluster formation in a dose-dependent manner (Figures S2B and S2C).

Figure 5.

Econazole decreases cell proliferation by inducing apoptosis

(A and B) The fraction of Ki67-positive (A) or TUNEL-positive (B) NF1^−/− hiPSC-SCPs was measured following treatment of the cells with different concentrations of econazole.

(C) Immunofluorescence imaging for pERK expression was performed on NF1^−/− hiPSC-SCPs treated with econazole or selumetinib at the indicated concentrations.

(D) Quantification of phospho-ERK (pERK) levels in (C) by relative fluorescence intensity.

(E) Proliferation of NF1^−/− hiPSC-SCPs was measured using CellTiter-Glo following treatment with increasing concentrations of econazole, selumetinib, or econazole + selumetinib (Sel).

Data represent mean ± SEM, n = 3. Scale bars, 100 μm.

Econazole treatment causes increased cell death in human cNF tissue

We next tested the effectiveness of econazole in two different ex vivo human neurofibroma models (Figure S3). Human cNFs were surgically removed from patients by the modified biopsy removal procedure described by Chamseddin et al.²¹ In one assay, these cutaneous tumors were hole punched to obtain equally sized tissue samples, and then these samples were placed with DMEM/F12 (10% fetal bovine serum [FBS]) medium in a 96-well plate with the epidermis portion facing upward. In a second assay, the cutaneous tumors were surgically grafted onto the backs of athymic nude mice. The tumor tissue was then treated with either commercially available econazole nitrate cream (1%), MEK inhibitor in a topical formulation, econazole nitrate cream plus MEK inhibitor, or petroleum jelly as a negative control. Application to the epidermis of the tumor tissue was done twice daily. We then performed TUNEL staining to measure apoptosis in the neurofibroma tissue following 24, 48, and 72 h of treatment. We found that econazole treatment alone efficiently induced apoptosis in the tumor tissue in both assays, exceeding the effect of topical MEK inhibitor (Figures 6A, 6B, S4A, and S4B). Histological comparison of neurofibroma and adjacent non-tumor skin revealed no significant morphological changes during these same time points; the tissue appeared intact with no signs of tissue necrosis at 24, 48, or 72 h. Hematoxylin and eosin (H&E) staining clearly depicts the histological features of a neurofibroma, including hypercellularity in the dermis and abundant spindle cells interspersed with collagen fibers (Figure 6C).

Figure 6.

Econazole treatment leads to apoptosis in the human ex vivo engraftment model of cNFs

(A) Representative images of TUNEL staining on human cNF tissue (neurofibroma) and non-tumor skin treated as indicated for 24, 48, or 72 h. Scale bars, 50 μm (left) and 500 μm (right).

(B) Quantification of TUNEL⁺ cells/DAPI⁺ cells in (A). Quantification was performed on multiple fields of view from 12 tumors and 6 adjacent skin samples from 3 individual patients. Each data point represents the approximate number of TUNEL⁺/DAPI⁺ cells in different fields of view from various regions of the tumor.

(C) H&E staining of the indicated tissue treated with Vaseline, econazole, MEK inhibitor, or econazole + MEK inhibitor (MEKi) for 24, 48, or 72 h. H&E staining was performed on 12 tumors and 6 adjacent skin samples from 3 individual patients. The H&E staining shows no signs of tissue necrosis at any time point and clearly depicts the histological features of neurofibroma, including hypercellularity in the dermis and abundant spindle-shaped Schwann cells (black arrows) interspersed with collagen fibers (red stars). Scale bar, 50 μm.

(D and E) Representative images (D) and quantification (E) of IHC analysis of the apoptotic markers cleaved PARP, cleaved caspase-3, and Bim after 24 h of the indicated treatment were performed on 8 tumors from 2 NF1 individuals .

(F and G) Representative images (F) and quantification (G) of IHC analysis of Ki67, pERK, and ERK after 24 h of the indicated treatment were performed on 8 tumors from 2 individual patients.

Scale bars, 50 μm (D and F). Data represent mean ± SEM. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. See also Figures S3–S5.

We also examined markers of apoptosis in the treated neurofibroma tissue with immunohistochemistry (IHC) and found that cleaved PARP (poly(ADP-ribose) polymerase), cleaved caspase-3, and BIM (BCL-2 interacting mediator of cell death) were significantly upregulated in the econazole-treated samples, with BIM levels highest in the combination-treated samples (Figures 6D and 6E). Similar to the econazole-treated human iPSCs (hiPSCs) (Figures 5C and 5D), the ex vivo human cNF tissue treated with econazole did not show significantly reduced levels of phospho-ERK while MEK inhibitor (MEKi)-treated cNFs did (Figures 6F and 6G). To further confirm our findings of apoptosis in econazole-treated samples, we performed immunofluorescence staining using the same markers. Our immunofluorescence data showed a similar increase in apoptotic marker expression, with cleaved caspase-3 significantly elevated in each of the three treatment groups to a comparable degree (econazole, MEKi, and combination-treated) (Figures S5A and S5B). We also saw decreased phospho-ERK expression in our samples treated with MEKi and econazole (Figures S5C and S5D). Because econazole can lead to apoptosis, phospho-ERK (pERK) levels may be expected to decrease over time because there are fewer viable cells present in the tissue. These ex vivo data not only confirm the in vitro results but also demonstrate the potential of using econazole nitrate cream as a topical therapeutic for cNFs.

To further confirm that econazole treatment induces apoptosis, we used cultured dorsal root ganglion/nerve root neurosphere cells (DNSCs), which have been shown previously to contain the cells of origin for neurofibroma.¹² We first tested different concentrations of econazole on Nf1^+/+, Nf1^+/−, and Nf1^−/− DNSCs and looked at the effect on cell proliferation at different time points (24, 48, and 72 h). We found that econazole treatment did not affect cell proliferation of wild-type Nf1 and Nf1-heterozygous DNSCs but significantly reduced cell proliferation of Nf1-mutant DNSCs (Figure 7A). The strongest decrease in Nf1^−/− DNSC proliferation was seen 24 h after econazole treatment. Additionally, we examined reactive oxygen species (ROS) activity in the Nf1^+/+, Nf1^+/−, and Nf1^−/− DNSCs because one purported mechanism of econazole is through the generation of ROS. After 1 h of econazole treatment, we observed ROS activation in both Nf1^+/− and Nf1^−/− DNSCs with econazole, MEKi, and combination treatment (Figures S6A and S6B). Because ROS activation was not statistically significant between the two cell genotypes, this suggests that econazole may be acting through alternative mechanisms in addition to the induction of ROS. We also observed that increasing concentrations of econazole induced apoptosis selectively in the Nf1^−/− DNSCs by TUNEL staining and increased cleaved caspase-3 levels while not affecting wild-type Nf1 or Nf1-heterozygous DNSCs (Figures 7B–7E and S6C). These data support the selective effect of econazole on NF1-null cells.

Figure 7.

Econazole inhibits cell growth and induces apoptosis in Nf1^−/− DNSCs

(A) Nf1^+/+, Nf1^+/−, and Nf1^−/− dorsal root ganglion/nerve root neurosphere cells (DNSCs) were treated with increasing doses of econazole, and cell proliferation was measured at 24, 48, and 72 h using CellTiter-Glo.

(B and C) Representative images of Nf1^+/+ and Nf1^−/− DNSCs treated with increasing doses of econazole and analyzed after 24 h for apoptosis by TUNEL assay (B) or expression of the apoptosis marker cleaved caspase-3 (CC3) (C). Nuclei were stained with DAPI.

(D) Quantification of TUNEL-positive cells in (B).

(E) Quantification of CC3⁺ cells in (C).

Experiments were performed three times with 3–5 technical replicates per experiment. Scale bars, 100 μm. Data represent mean ± SEM. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. See also Figure S6.

Discussion

We screened a commercially available (https://www.selleckchem.com/screening/fdaapproved-drug-library.html) library of 2,570 FDA-approved compounds on an hiPSC-based model of NF1-associated neurofibromas, using inhibition of cell proliferation as a readout. The primary compound screen and subsequent retesting identified nine compounds as potential therapeutics for NF1-associated neurofibromas. Further evaluation showed that the anti-fungal drug econazole exhibits selective effects on the proliferation rate of (1) SCPs and (2) the NF1^−/− genotype. Importantly, econazole inhibits proliferation of NF1^−/− SCPs at low concentrations, with an IC₅₀ (half-maximal inhibitory concentration) of about 0.15 μM.

Econazole is widely used as a topical cream for fungal infections of the human skin, such as tinea pedis, tinea cruris, etc., with little evidence of major side effects. Some minor side effects such as pruritis have been reported.²² Of note, pruritus is also reported in 19.4% of NF1 patients with cNFs,²³ and so this adverse event would need to be monitored for exacerbation in future preclinical and clinical trials with econazole. In fungal cells, econazole prevents cell wall synthesis by inhibiting the production of ergosterol, a key cell wall ingredient.²⁴ Other studies have shown that econazole also has anti-tumor activity in a wide range of cancer cell lines in vitro, including breast, gastric, colon, pancreatic, and lung cancer.^{25,26,27,28,29} Some purported mechanisms include cell-cycle arrest and apoptosis through the generation of ROS.^{25,26,27,28,29} Though we did observe ROS activation in both Nf1^−/− and Nf1^+/− DNSCs after treatment with econazole, we did not see decreased proliferation or the induction of apoptosis in Nf1^+/− cells, suggesting that econazole may be acting through an alternative mechanism to promote cell death. Because Nf1^+/− cells have been shown previously to affect cell proliferation and function in NF1, the induction of ROS may be affecting other downstream pathways in these cells.^9,30 In a pancreatic ductal adenocarcinoma cell line, econazole has been shown to increase induction of autophagy and decrease lysosomal biogenesis.²⁹ This resulted in accumulation of autophagosomes, which, in turn, caused endoplasmic reticulum (ER) stress-mediated apoptosis. However, the reported econazole concentrations needed to induce cell death in some of these cell lines are 100–200 times higher than what we observed in NF1^−/− SCPs. Importantly, these high concentrations are physiologically unachievable, thus making these results not translatable into therapy. Additionally, cell death caused by such high concentrations is likely due to non-specific toxicity rather than biological effects. In contrast, an IC₅₀ of 0.15 μM, as we observed in hiPSC-SCPs, is potentially achievable through topical application of commercially available anti-fungal creams containing econazole; however, this remains to be tested. We were able to effectively induce apoptosis in neurofibroma tissue in two ex vivo models of cNFs through topical application of econazole cream. Additionally, we did not see significant cell death in adjacent non-tumor tissue, suggesting that econazole is acting selectively on NF1^−/− cells. Histological comparison of neurofibroma and adjacent skin did not reveal any significant changes, suggesting that econazole is more likely to be acting on a molecular rather than morphological basis during this time course. Due to the time course (up to 72 h) necessitated by our ex vivo assays, we would not expect to see changes in the histological appearance of the tissue.

Because cell walls are absent in mammalian cells, econazole must act on SCPs through a different mechanism. In both NF1^−/− hiPSC-SCPs and Nf1^−/− DNSCs, we saw that econazole appears to not affect the Ras-MEK-ERK pathway signaling and, thus, is likely acting through pathways different from the approved drug selumetinib. Additionally, though the purported mechanism of econazole in other cancers has been studied previously, it is unclear how econazole may be acting in a benign tumor. More work must be done to elucidate its precise mechanism in neurofibroma.

One limitation of our study is that our screen was conducted using SCPs, but there are many other cell types in cNFs, including fibroblasts, neurons, mast cells, macrophages, and endothelial cells. Further studies will be needed to determine whether econazole has any effect on these other cell types and whether inhibiting SCP/SLCs alone is sufficient to treat cNFs. A second limitation is that we used only one pair of isogenic iPSCs for the screening and validation, and thus this particular NF1 mutation, together with the distinct genetic background of these cells, might cause a drug response that is not reproduced with other NF1 mutations. However, our in vitro experiments where we tested econazole on cNFs surgically removed from multiple NF1 patients suggest that the effect is likely applicable more generally. A third limitation comes from the fact that cNFs do not grow indefinitely; after initial rapid growth, they become quiescent. Thus, while we do see increased apoptosis in human cNFs in our ex vivo assays upon econazole treatment, if econazole also inhibits NF1^−/− Schwann cell proliferation, it may not be as effective in established cNFs. However, systemic administration could lead to apoptosis of proliferating NF1^−/− Schwann cells in pNFs and could be used in combination with MEK inhibitors. Finally, the cells we used in our in vitro experiments are SCPs and not the quiescent cells found in mature cNF tumors. NF1^−/− SCPs have a higher proliferation rate than their NF1^+/− and NF1^+/+ counterparts.^16,31 This difference in proliferation is likely due to the function of NF1 as a tumor suppressor. However, in other cell systems (for example, semi-immortalized human Schwann cells), proliferation rates were not increased in NF1-mutant cells, indicating that loss of NF1 alone may not be sufficient to drive an increased proliferation rate.³² Thus, the effects of econazole in our hiPSC model that initially identified the anti-fungal drug as well as in our ex vivo models may not be recapitulated in vivo given the multicellular and quiescent nature of the mature cNFs.

In summary, NF1 is the most common tumor disposition disorder, and more than 99% of NF1 patients develop cNFs. These tumors are often reported to be the most burdensome aspect of the disease, greatly reducing quality of life. However, to date, there are no approved therapies for cNFs. Utilizing a human stem cell-based disease model, we have identified the anti-fungal cream econazole as a very promising therapeutic drug with selective anti-tumor activity in NF1^−/− neurofibroma-like cells. Econazole is known to be extremely safe and is currently used topically to treat various fungal skin infections with minimal adverse effects. In this study, we showed that econazole treatment of human neurofibroma efficiently induces apoptosis selectively in the tumor tissue. We propose the use of topical econazole as a treatment for cNFs, helping to alleviate both the medical burden and psychosocial effects of these tumors. We also suggest the use of econazole and topical MEK inhibitors as a potentially more effective combination therapy. In the future, a new formulation of topical econazole with better penetration deep into the dermis will need to be tested for its effect on cNFs in human clinical trials to translate the in vitro and ex vivo effects we observed into a clinical setting.

Limitations of the study

A limitation of our study is that only one patient-derived hiPSC-SCP line was used for the compound screen, and it is possible that cells from an individual with a different NF1 mutation might not respond the same way. Additional screening using hiPSC-SCPs from other NF1 individuals would support the general effectiveness of econazole in treating cNFs. Although we tested the effects of econazole using ex vivo culture systems, in vivo testing using one of our genetically engineered mouse models of NF1 would also further bolster the efficacy results of econazole. Finally, the exact mechanism of action by which econazole induces apoptosis remains to be elucidated. Further studies to investigate this will be needed. We acknowledge that additional studies will be needed before proceeding to clinical testing.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
AP2	Abcam	ab61; RRID: AB_305292
BIM	Cell Signaling	2933S; RRID: AB_1030947
Cleaved Caspase 3	Cell Signaling	9661S; RRID: AB_2341188
Cleaved PARP	Cell Signaling	9541S; RRID: AB_331426
GAP43	Abcam	ab75810; RRID: AB_1310252
Ki67	Abcam	ab16667; RRID: AB_302459
NF1	Bethyl Laboratories	A300-140A; RRID: AB_2149790
P75	Abcam	ab8874; RRID: AB_306827
Phospho-ERK	Cell Signaling	9101S; RRID: AB_331646
Total-ERK	Cell Signaling	4695; RRID: AB_390779
Sox10	Thermo Fisher	AF2864; RRID: AB_442208
S100β	Abcam	ab52642; RRID: AB_882426
Bacterial and virus strains
Ad5CMVCre-eGFP	University of Iowa Viral Core	VVC-U of Iowa-1174-HT
Ad5CMVeGFP	University of Iowa Viral Core	VVC-U of Iowa-4
Chemicals, peptides, and recombinant proteins
Econazole nitrate	Selleckchem	S2535
Mirdametinib (PD0325901)	Selleckchem	S1036
Econazole nitrate (1%) cream	UT Southwestern Campus Pharmacy	N/A
B27	Fisher	12-587-010
N2	Gibco	17502–048
Fibronectin	Sigma	1801185647
EGF	Thermo Fisher	PHG0313
b-FGF	Thermo Fisher	PHG0360
RIPA buffer	Thermo Fisher	89901
DAPI	Thermo Fisher	0100–20
Accutase	STEMCELL Technologies	692 SEK
CHIR99021	STEMCELL Technologies	#72054
SB431542	R&D Systems	#TB1614-GMP
NRG1	PeproTech	GMP100-03
Hoechst 33342 Solution (20 mM)	Thermo Fisher	62249
FDA-approved Drug Library	Selleckchem	L1300
Critical commercial assays
DeadEnd™ Fluorometric TUNEL System	Promega	G3250
Promega CellTiter-Glo™ Luminescent Cell Viability Assay Kit	Promega	G7571
DCFDA/H2DCFDA - Cellular ROS Assay Kit	Abcam	ab113851
Pierce BCA protein assay kit	Thermo Fisher	PI23225
Experimental models: Cell lines
Human NF1+/− iPS cells	Broad Institute	N/A
Human NF1−/− iPS cells	Broad Institute	N/A
Experimental models: Organisms/strains
Foxn1-mutant mice	Charles River	553; RRID: IMSR_T001571
Oligonucleotides
sgRNA (mutant) 5′ GGCATCACTGAGGCACTGTA-3″	This paper	N/A
sgRNA GAP1 5′ TCCATGCAGACTCTCTTCCG-3″	This paper	N/A
SgRNA GAP2 5′ GCTAACATGTTGCCAATCAG-3′	This paper	N/A
Software and algorithms
ImageJ	Schneider et al.33	https://imagej.nih.gov/ij/
CellProfiler	Stirling et al.34	https://cellprofiler.org

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Lu Q. Le (lu.le@utsouthwestern.edu).

Materials availability

The CRISPR-Cas9 edited human iPS cells generated for this study may be requested under a material transfer agreement as per institutional requirements from lead contact.

Data and code availability

• •All data reported in this paper will be shared by the lead contact upon request.
• •This paper does not report original code.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Experimental model and study participant details

iPS cell generation from patient blood sample and CRISPR/Cas9 gene editing

We collected peripheral blood mononuclear cells from four individuals (aged <18 years), who had previously been diagnosed with NF1. The cells were reprogrammed with non-integrating Sendai virus into iPSC lines, which expressed pluripotency markers and were karyotypically normal. The GT duplication (c.3431_3432dupGT) in Exon 26 of individual #2 was identified by sequencing of the NF1 locus.

Guide RNAs (Synthego) were designed to a) directly bind to the site of the mutation (GGCATCACTGAGGCACTGTA), and b) span part of the GAP domain of the NF1 gene (TCCATGCAGACTCTCTTCCG or GCTAACATGTTGCCAATCAG). Two guide pairs were tested and overall 83 clones were isolated. Successful introduction of the large deletion was assessed by PCR, and the intact mutant allele carrying the germ line mutation was confirmed by sequencing. 5 clones were selected and expanded: 1 unedited, 1 carrying an indel mutation in the wild-type allele, and 3 carrying deletions in the wild-type allele. IPSC generation and genome-editing was performed in collaboration with the Harvard Stem Cell and Regenerative Biology (HSCRB) iPS Core.

Schwann cell induction and differentiation protocol

Human induced pluripotent stem cells (hiPSCs) were routinely maintained in Stemflex Medium (Gibco) on Matrigel-coated dishes until 80% confluency and passaged every 5 days at ratios between 1:6 and 1:10 using EDTA. Only hiPSCs between passages 12 and 30 were used for differentiation. Schwann cell precursor cells (SPCs) were generated from hiPSCs according to previously published protocols with minor modifications.^16,18 Briefly, hiPSCs were dissociated into single cells with Accutase (STEMCELL Technologies) and replated on Matrigel-coated dishes at a density of approximately 200,000 cells/cmˆ2. The next day, the culture medium was switched to DMEM/F12 with N2, B27 (all from Life Technologies), 10 μM SB431542 (R&D Systems), and 1.5 μM CHIR99021 (STEMCELL Technologies) for 5 days, followed by the addition of 50 ng/mL NRG1 (R&D Systems) until day 17. The cells were passaged at a ratio of 1:3 on days 5, 10, and 17. After day 17, SCPs were cultured in DMEM/F12 with N2, B27, 10 μM SB431542, 1.5 μM CHIR99021, 100 ng/mL NRG1 and 5 ng/mL FGF2 (R&D Systems), on Matrigel-coated plates. At 90% confluence, SCPs were dissociated with Accutase and re-plated at a ratio of 1:6. The cells were either maintained for up to 5 passages or cryopreserved in maintenance medium supplemented with 10% DMSO after passage 2.

For differentiation into Schwann cells, CHIR-99021 (a GSK3 inhibitor) and FGF2 were removed and the medium was switched to DMEM/F12 with N2, B27, 1% FBS, 10 μM SB431542, 100 ng/mL NRG1, 10 ng/mL PDGF-BB (Peprotech); after one week, SB431542 (a TGF-β receptor inhibitor) was removed from the medium.

Two minor modifications for SCP generation that were incorporated were: 1) The reduction of CHIR concentration from 3μM to 1.5μM since this led more robustly to uniformly SOX10+ proliferating populations, while 3 μM at times caused the cells to acquire a flat morphology early during differentiation. 2) The addition of low amounts of FGF2, which increased cell survival.

Drug screening growth assay and dose-response curves

For the compound growth assay, NF1+/− and NF1−/− SCPs were dissociated with Accutase, resuspended in DMEM/F12 with N2, B27, 1% FBS, 10 μM SB431542, 50 ng/mL NRG1β and 10 ng/mL PDGF-BB, and plated on Matrigel-coated 96-well microplates (PerkinElmer) at a density of 1,500 cells/well, using a Multidrop Combi Reagent Dispenser (Thermo Fisher). The cells were incubated for 24h to ensure recovery before compounds were added. A commercially available drug repositioning library (Selleckchem; catalog no. L1300) was used for the primary screen. At the time of purchase, the library contained 2,570 compounds. There was no further pre-selection of candidate compounds between purchase of the library and primary screen. BMP-inhibition with 1μM LDN-193189 (R&D Systems) and MEK inhibition with 10μM selumetinib served as positive controls, and 0.2% DMSO served as vehicle control. Separate all-vehicle control plates were prepared to account for edge effects. After 72h, the cells were stained with Hoechst, and the plates were imaged using a Cytation imaging reader (BioTek). Nine images at 4× magnification were taken of each well, covering nearly the entire well surface. Automated nuclei counting was performed in CellProfiler.³⁴

Cell numbers were normalized for the all-vehicle control plates and calculated as fractions of the negative control. Every compound was tested in duplicate at 1μM both in NF1 knockout (KO) and heterozygous SCPs. The effects on proliferation on KO and heterozygous SCPs were compared in absolute (ΔKO-ΔHET) or relative (ΔKO/ΔHET) terms and each compound was scored based on the number of standard deviations its specific effects on NF1−/− SCP proliferation exceeded the overall dataset. The 50 highest scored compounds were designated as hits. Hit compounds were re-ordered from Selleckchem and re-tested first in 5-point, then in 10-point dose-response curves.

For comparing compound effects on SCPs and Schwann cells, SCPs were first differentiated into Schwann cells for 2 weeks before being treated with compounds at different concentrations. The primary screen was performed in duplicates; all re-testing/dose-response curves were done in triplicates.

Human neurofibroma specimen collection

Five patients with NF1 gave informed consent to the study and underwent a surgical procedure to remove cNFs as previously reported²¹ at the UT Southwestern Dermatology Clinic. Patients were given a summary of the procedure, risks and benefits, and importance of postoperative care before they provided consent. There were no complications with the procedures. Prior to surgery, photographs of the tumors were taken to document their location. In the entire study, a total of 40 neurofibromas were removed and collected. Human subjects and cNF sample collection and use were approved by the Institutional Review Board at University of Texas Southwestern Medical Center.

Study approval

Animal care and use in this study were approved by the Institutional Animal Care and Use Committee at University of Texas Southwestern Medical Center. IRB approval was obtained from Harvard University (IRB17-0413) and University of Texas Southwestern Medical Center (STU 062014-015) for hiPS cells and human cNF tissue studies, respectively.

Method details

In vitro testing of econazole

The protocol to harvest Nf1−/−, Nf1^+/−, and Nf1+/+ mouse DNSCs was performed as previously reported.¹² Briefly, E13.5 mouse embryos were removed from anesthetized pregnant Nf1^fl/- and Nf1^fl/fl female mice and sacrificed. The spinal cord was removed, and DRGs/nerve roots were dissected from the vertebral column with the aid of a stereomicroscope. Fine scissors were used to cut and separate the nerve roots and DRGs. Cells were then plated on fibronectin coated plates in proliferation media: DMEM/F12 containing penicillin/streptomycin (0.1%); fungizone (40 μg/mL); B27 (without vitamin A), epidermal growth factor (20 ng/mL), and basic fibroblast growth factor (40 ng/mL; Sigma). Nf1^fl/- cells were infected with adenovirus carrying the Cre recombinase (Ad-CMV-Cre) to generate Nf1^−/− DNSCs, and Nf1^fl/- and Nf1^fl/fl DNSCs were infected with adeno-GFP to generate Nf1^+/− and Nf1^+/+ DNSCs. Deletion was confirmed by genomic analysis of Nf1. Nf1^−/− or Nf1^+/+ DNSCs were then replated, and once cells reached 70% confluency, they were treated with vehicle (DMSO), differing concentrations of econazole (0.1–10 μM), MEK inhibitor, or a combination of econazole plus MEK inhibitor. After 24 h of incubation, the cells were harvested for protein analysis.

Ex vivo testing of econazole treatment on human tumor samples in culture

Freshly harvested human cutaneous neurofibroma or nearby non-tumor skin samples were hole punched (6 mm punch) to get equivalent-sized samples and placed in 10% FBS medium in 96-well plates. Depending on the size of the neurofibroma that was surgically removed, the number of punches varied per tumor. The samples were placed in the wells with the epidermis pointing upward, thus exposing the neurofibroma to the air. The samples were then treated with either vehicle (petroleum jelly), econazole nitrate 1% cream (provided by the UT Southwestern Campus Pharmacy), MEK inhibitor or combination treatment changed twice daily. MEK inhibitor PD0325901 (PD901) was dissolved in 100% DMSO. Before topical application, this solution was diluted 1:1 with water. A total of 1 mg/mL PD901 was applied to the tumor twice daily. Tumors were harvested at different time points and collected for histological analysis.

For the grafting studies, nude (Foxn1-mutant) mice (purchased from Charles River Laboratory) were anesthetized (4 μL/g, body weight) by intraperitoneal injection using a mixture of ketamine (10 mg/mL) and xylazine (1 mg/mL) solution (provided by the UT Southwestern Animal Resource Center [ARC], Dallas, Texas, USA). Using fine scissors, a skin incision was made on the back of the animal. The human neurofibroma or non-tumor skin was grafted to the skin of the mouse with 4-0 proline suture. For these experiments, we chose smaller cNFs (approximately ≤1cm in size) to graft onto the mouse skin that did not require a hole punch.

Following the surgery, mice were placed into heated cages to recover from anesthesia. Vehicle (petroleum jelly), topical MEK inhibitor, econazole nitrate cream (1%), or combination treatment was applied to the epidermis of the cNF with a cotton-tipped applicator to fully cover the epidermis. Mice were housed in single cages to prevent tumor disruption. The mice were sacrificed and tumors harvested at different time points and collected for histological analysis or stored for RNA and protein analysis.

In both of these assays, econazole was applied on the epidermal side of the tumor, however it is possible that the drug may have penetrated along the side of the sample directly into the dermis.

Histology and immunostaining

For hematoxylin and eosin analysis, tissue specimens were harvested and fixed with 10% formalin in PBS for 1 day and subsequently embedded in paraffin. Sections (5 μm thick) were stained with H&E according to the manufacturer’s protocol (StatLab). For immunohistochemistry and immunofluorescence staining, frozen sections or paraffin sections after deparaffinization, rehydration, and antigen retrieval were used. Primary antibodies used: AP2 (#ab61, Abcam); BIM (#2933S, Cell Signaling); cleaved caspase-3 (#9661S, Cell Signaling); cleaved PARP (#9541S, Cell Signaling); GAP43 (#ab75810, Abcam); P75 (#ab8874, Abcam); NF1 (#A300-140A; Bethyl Laboratories) phospho-ERK (#9101S, Cell Signaling); total ERK (#4695, Cell Signaling); S100β (#ab52642, Abcam); SOX10 (#AF2864, Thermo Fisher); Ki67 (Abcam ab16667). For immunofluorescence staining, the primary antibodies were detected by secondary antibodies or streptavidin conjugated with Alexa 488 or Alexa Fluor 555 (Life Technologies) and nuclei were counterstained with DAPI (Thermo Fisher).

Western blot

Protein was extracted using RIPA buffer (Thermo Fisher, 89901). Protein concentration was determined using a Pierce BCA protein assay kit (Thermo Fisher, PI23225). Protein was separated by SDS-PAGE and transferred to PVDF membranes. Membranes were blocked for 1 h with 5% BSA, and then incubated with primary antibodies at 4°C overnight. Membranes were then incubated with HRP-conjugated secondary antibodies for 1 h. The ChemiDocTM Touch Imaging System (Bio-Rad) was used to capture the images.

Microscopy

Immunohistochemistry and TUNEL images were taken with an Olympus Ix-73 microscope.

CellTiter-Glo assay

Following 24-h treatment with vehicle or differing concentrations of econazole, the CellTiter-Glo® assay was performed in opaque-walled 96-well plates per the manufacturer’s protocol (Promega). Control samples containing media without any cells present were used as controls. Luminescence proportional to the amount of ATP present in each well was recorded.

TUNEL staining assay

The DeadEnd Fluorometric TUNEL assay was performed according to the manufacturer’s protocol (Promega) using paraffin-embedded human neurofibroma and non-tumor tissue samples. The assay measures the fragmented DNA of apoptotic cells by the incorporation of fluorescein-12-dUTP labeling. The fluorescently labeled DNA (green) was detected along with DAPI+ cells by the use of fluorescence microscopy. TUNEL-positive cells were quantified based on their total fluorescence and colocalization with DAPI+ cells.

ROS activity assay

The DCFDA/H2DCFDA - Cellular ROS Assay Kit (Abcam) was performed according to the manufacturer’s protocol, following 1 h treatment with vehicle, differing concentrations of econazole, MEK inhibitor, or combination. The assay measures hydroxyl, peroxyl, and other reactive oxygen species (ROS) activity within the cell. ROS generation was observed under the fluorescence microscope. ROS positive cells were quantified based on their total fluorescence and colocalization with Hoechst+ cells.

Quantification and statistical analysis

All data in this study are displayed as the mean ± SEM. For the iPSC experiments, the primary screen was performed with 2 technical replicates and the retesting of primary hits was performed with 3 technical replicates. All other results were obtained from at least 3 independent experiments. For the ex vivo experiments, at least 2 different biological replicates were used for each experiment with at least 2 experimental replicates. The experiments using mouse DNSCs were performed 3 times with 3–5 technical replicates per experiment. Comparisons among groups were performed by 1-way ANOVA or 2-tailed Student’s t test for pairwise comparisons. p values of 0.05 or less were considered statistically significant.

Published: December 19, 2023