The MEK-ERK signaling pathway is activated by growth factor-stimulated RAS binding to BRAF, which, when dysregulated, can initiate tumorigenesis by promoting uncontrolled cell growth (1) and cell fate changes (2, 3). Alterations in the BRAF gene, which encodes a central kinase in the MAPK signaling pathway, are prevalent in CNS tumors and are enriched in the pediatric population. BRAF protein is the most frequently mutated serine/threonine kinase in human cancer, with BRAF^V600E and KIAA1549::BRAF fusion being significant drivers of cellular transformation in pediatric CNS tumors, with high occurrences in low-grade gliomas (LGGs, up to 35%), and low occurrences in high-grade gliomas (HGGs, ~5% for BRAF^V600E) (4, 5). BRAF^V600E is a class I mutation, the most common class of mutations in glioma. Through a somatic point mutation at codon V600, valine is substituted by glutamate (BRAF^V600E), which is thought to mimic regulatory phosphorylation in the activation segment of the protein kinase domain (6). Consequently, BRAF^V600E-mutated kinase monomers constitutively activate the downstream MAPK-ERK signaling pathway, regardless of any external stimuli and independent of upstream Ras (7, 8).

Due to the high incidence of BRAF^V600E in several solid cancers, small molecule inhibitors against BRAF^V600E have been developed and have been actively tested in the clinic (9–11). These BRAF inhibitors rapidly suppress MAPK signaling but exhibit only transient effects, prompting their subsequent combination with MAPK inhibitors for treatment (12). BRAF^V600E and MEK inhibitor combination therapy (e.g., dabrafenib+trametinib) has been tested in the clinic against glioma and clinical response rates of 70% and 33% have been observed amongst adults with low- and high-grade glioma, respectively. Similar response rates in children have led to the recent approval of BRAF^V600E inhibitor dabrafenib and MEK inhibitor trametinib for pediatric patients one year of age and older with BRAF^V600E-mutated low-grade gliomas (LGG) who require systemic therapy (13–15). Despite these advances, further research is needed to develop new therapies and improve patient outcomes; clinical data shows tumor rebound when treatment is discontinued, and high-grade tumors frequently fail to respond or develop therapy resistance (16). Significant challenges stand in the way of improving current treatments, including that BRAF^V600E-mutated pediatric LGG (pLGG) are frequently found with secondary mutations, most commonly CDKN2A deletion, and at lower frequency with single nucleotide variations (SNVs) in NF1, FGFR1, KRAS, and H3FA (5). In addition, 24% of BRAF^V600E-mutated adult glioblastoma had mutations in the TERT promoter (17).

Alteration type influences the clinical course and therapy responsiveness of tumors. BRAF^V600E altered pLGG lacking tumor suppressor CDKN2A are at high risk for transforming into high-grade tumors when compared with BRAF^V600E CDKN2A balanced and BRAF wildtype tumors (18–20). Co-occurring alterations provide novel therapeutic vulnerabilities that can be targeted such as for example CDK4/6 inhibitors that target the p16/RB axis (21). Whether these drugs work synergistically or cooperatively with MAPK pathway inhibitors is not fully explored and additional preclinical testing is needed. Such preclinical testing requires preclinical models for BRAF^V600E CNS tumors that capture the genetic alterations found in human tumors. Mechanisms for cooperation of BRAF^V600E with CDKN2A loss for malignant progression have been investigated in genetically engineered mouse models (22). Other combinations of other co-occurring alterations (e.g., FGFR, H3FA) and BRAF^V600E are not yet available in models.

In conclusion, BRAF^V600E is a common kinase activating alteration in human cancer and in CNS with high prevalence in pLGG, whereby patients with BRAF^V600E altered tumors are routinely, successfully treated with molecular targeted therapy (BRAF inhibition). A unique feature is that BRAF^V600E-mutated hemispheric pLGG tend to progress from low to high grade tumors at which point they become difficult to manage and exhibit therapy resistance to BRAF inhibition (20, 23). Models that recapitulate BRAF^V600E pLGG and the progression from low- to high-grade are not available, which is why little is known about their biology and the underlying mechanisms of progression; this will be discussed in detail in the “Developing models for tumor progression from low to high grade is challenging” section in this review.

The clinicopathologic diversity of BRAF^V600E-mutated CNS tumors due to diverse co-occurring alterations makes treatment decisions ambiguous and raises the need for preclinical testing of combination therapies of BRAF/MEK inhibitors and therapies targeting co-occurring alterations. Such studies require generation of additional models which represent all types of genetic alterations combined with BRAF^V600E in human tumors.

BRAF^V600E CNS tumors are enriched in the pediatric population and occur most frequently in pLGG (17-35% of pLGG) and are found in a diverse range of histopathologic subtypes of glioma. Both pleomorphic xanthoastrocytoma/PXA (40~80%), ganglioglioma/GG (25~45%) and desmoplastic infantile gangioglioma/astrocytoma (DIG/DIA; 45%) are more likely to harbor the mutation than other pLGG, such as pilocytic astrocytoma/PA (5~16%); optic pathway glioma (OPG; ~2%), dysembryoplastic neuroepithelial tumors/DNT (5~10%), infantile LGG (iLGG; 18%), and diffuse pediatric-type LGGs (19, 24–26). BRAF^V600E occurs at lower frequency in pHGG (6-14%) a histopathologically and clinical diverse group of tumors (4). Importantly, BRAF^V600E expression when coinciding with CDKN2A deletion marks a clinicopathologic subgroup of pLGG at high-risk for progression (5, 18). Consistent with this high risk, BRAF^V600E is the most frequent recurrent mutation in secondary pHGGs, occurring in 39% of cases (27).

BRAF^V600E-mutated pLGG occurred most frequently in hemispheric regions (56%) but were also common in the diencephalon (29%). Interestingly, BRAF^V600E-mutated iLGG which present most frequently in the midline rarely progresses, and the mutation is absent in infantile HGG (iHGG), which is paradoxical to its occurrence in pHGG overall (5, 26). These data suggest that tumor location affects the clinical course of BRAF^V600E-mutated gliomas. Further to this point, BRAF^V600E-mutated PA are prevalent in diencephalic areas in particular in unresectable hypothalamic/optic pathway PA, in contrast to KIAA1549-BRAF fusion-positive PA, which are prevalent in the cerebellum (24). Amongst these hypothalamic/optic pathway PA, BRAF^V600E is more frequently associated with non-progression, and KIAA1549-BRAF fusion was more common in progressive tumors (28).

BRAF-mutated diencephalic LGGs rarely present malignant transformation, but because they are usually unresectable, novel targeted therapy is strongly warranted. Indeed, clinical reports show promising results with BRAF^V600E inhibitor therapy in optic pathway glioma (29). Spatial distribution patterns of the BRAF^V600E mutation in pediatric, infantile, and adult LGGs (25) and clinical trial results with molecular targeted therapies against BRAF-mutated tumors have been extensively reviewed elsewhere (8, 17, 30, 31). While BRAF-fusion events have been associated with a good prognosis, and SNVs (most prominently BRAF^V600E) have been associated with poor prognosis in pediatric LGG when all sites and histologies were included (5), the situation is reversed in BRAF-mutated pediatric diencephalic LGG such as optic pathway/hypothalamic glioma.

In conclusion, BRAF^V600E-altered CNS tumors occur in diverse neuroanatomical locations, predominantly in hemispheric areas and BRAF^V600E-altered PA exhibit a strong association with extra-cerebellar location. The microenvironment of tumors in distinct locations may determine the prognostic value of the BRAF alteration while responses to BRAF inhibitors appear to be universal. Due to the small study size of diencephalic tumors, further clinical and preclinical analyses are needed to test the neuroanatomical aspects of BRAF^V600E-altered CNS tumors.

BRAF^V600-mutated tumors present at a wide age range with a median age of 10.6 years in the pediatric population and are also found in adults. Genomic and epigenomic large scale analyses have detected BRAF^V600E in several adult CNS astrocytic tumors, including astroblastoma (24%-38%) (8, 25, 32), diffuse astrocytoma (DA; 30~40%), glioblastoma multiforme (GBM; ~ 6%) and astrocytoma (2~5%) (5, 17, 24, 33). Notably, the spectrum of histologic grades, pathology and nature of BRAF mutations in adults contrasted with that in pediatric tumors. In adults with BRAF-mutated CNS tumors, BRAF^V600E is most frequent in glioblastoma (~50%), and is common in epitheloid GBM (34, 35), followed by LGG (~22%), and PXA (18%). The mutation is notably absent in tumors with oligodendroglioma histology. Therefore, the histologic spectrum of BRAF^V600E CNS tumors is different in pediatric and adult patients. BRAF^V600E co-occurs with a similar set of alterations in pediatric and adult tumors, and both patient groups can benefit from BRAF-targeted therapy (17).

In non-CNS solid cancers, BRAF^V600E is frequently detected in malignant melanoma, hairy cell leukemia, colorectal carcinoma, and papillary thyroid carcinoma, as well as ovarian and lung tumors (36). Approximately 50% of melanomas harbor a BRAF^V600 mutation (37, 38). Melanoma has the highest propensity to form brain metastases of all malignancies (39) and given that melanoma patients have a high incidence (10-40%) of developing metastases in the brain (40), the clinical relevance of BRAF mutational status in melanoma brain metastasis has been studied (41, 42). Evidence showed that patients with BRAF^V600E-mutated melanoma were more likely to have CNS involvement than those with BRAF wild-type melanoma (43, 44). Clinical studies showed safety and efficacy of BRAF inhibitor therapy in patients with treated and untreated melanoma brain metastases more recently (45–47). In addition, BRAF and MEK inhibitor combination therapy provided promising clinical results against melanoma brain metastases (48, 49) which have prompted assessing baseline clinical features associated with outcomes (45, 46, 48–51). Genetic interaction studies in genetically engineered mouse models showed that BRAF ^V600E expression cooperates with Pten tumor suppressor loss to generate metastatic melanoma without CNS involvement (52). Expression of BRAF^V600E in Cdkn2a^Null mice generates melanomas without metastasis, but AKT1 activation promotes development of CNS metastases in this model (53).

In conclusion, BRAF^V600E expression is a driver mutation of CNS tumors that occur in various neuroanatomic locations and exhibit a wide range of age at presentation. In non-CNS cancers BRAF^V600E expression is not sufficient to generate CNS metastases and the type of co-occurring alterations determine whether CNS metastases form. More work is needed to unravel how BRAF^V600E mutation cooperates with co-alterations to increase invasive properties of tumor cells and generate brain metastasis.

Dysregulated neurodevelopmental programs underlying the distinct susceptibility of stem and progenitor cells to oncogenic alterations cause childhood CNS cancer (54). As stated above, the wide variance of disease location, age at onset, histopathology, and clinical outcome of BRAF^V600E-mutated CNS neoplasms is quite striking and quite unique amongst pediatric CNS cancer types. This diversity sets BRAF^V600E apart from other point mutations in pediatric CNS cancers, such as the histone 3 variants H3.3 and H3.1, which are strongly associated with unique locations, high grade and more narrow age range, and a uniformly dismal prognosis (55). It suggests that cells at distinct stages and neuroanatomical locations can be susceptible to transformation by this oncogene (55, 56). How a single oncogene can generate such diverse tumors and lead to age-dependent differences of prognosis is unknown. Tumor extrinsic factors, such as the tumor microenvironment and the immune system may play an important role in the development of these tumors and these factors need to be considered as important determinants of tumor subtypes when developing BRAF^V600E CNS tumor models. We are far from capturing the diversity of BRAF^V600E-mutated CNS tumor’s onset, histopathology, location, and prognosis with in vitro and in vivo models; thus, the underlying mechanisms leading to such diverse tumors remain poorly understood.

This represents a significant research problem. Despite the initial successes of treatment with FDA-approved first-generation small molecule inhibitors of BRAF^V600E, such as vemurafenib, dabrafenib, and encorafenib, still therapeutic effects are not durable. Although effective in treating other varieties of BRAF^V600E malignancies, tolerance towards BRAFi as a monotherapy has been shown in melanoma and HGG, commonly through (acquired, secondary) reactivation of the MEK pathway in melanoma, and resistance mechanisms are discussed in more detail below. Since treatment efficacy is limited by drug resistance in HGGs and BRAF^V600E -mutated LGGs frequently progress to high-grade, novel therapeutic approaches against BRAF^V600E -mutated gliomas in general are needed, raising the urgent need for further refinement of BRAF^V600E models for preclinical studies and their successful translation.

Here, we aim at providing an overview of mammalian models for BRAF^V600E mutated CNS tumors; we list existing models, compare their usefulness for basic mechanistic studies and preclinical approaches, and describe current strategies to analyze and quantify tumor phenotypes and vulnerabilities. The objectives of this review are to categorize available preclinical in vitro and in vivo models of BRAF^V600E-mutated CNS tumors. We highlight the applicability of these models for drug screening purposes and managing resistance. We describe their utility for disease characterization, including identifying and characterizing co-occurring mutations for their clinicopathologic characteristics and predictive and prognostic impact. In addition, we make suggestions on how to overcome existing limitations in model development. Preclinical studies should be performed in multiple models to avoid model-specific bias and ensure that preclinical data are robust and successfully translated to the clinic. Fortunately, several types of models are available and will be discussed in greater detail below.

Conventional cancer cell lines have been historically used for therapeutic screening due to their robust and uniform phenotypes and growth behavior as attached monolayer in high-serum culture conditions. Davies et al. identified BRAF^V600E mutations in established cancer cell lines DBTRG-05MG and AM38 (6), which subsequently became workhorses for preclinical testing, in particular of small molecule BRAF inhibitors, including PLX4720, a tool compound of Vemurafenib ( Table 1 ). PLX4720 suppressed cell viability in several BRAF^V600E altered cell lines and patient-derived orthotopic xenografts (PDoX) from DBTRG-05MG and AM38, with PDoX recapitulating the in vitro sensitivity to BRAF inhibitors ( Table 2 ) including PLX4720 monotherapy (79).

Like studies in melanoma, where BRAF inhibitor monotherapy shows rapid but transient MAPK pathway inhibition, PLX4720 treatment in glioma cell lines (M-38, DBTRG-05MG and NMC-G1) did not suppress MAPK pathway signaling in a durable way. This prompted testing of vertical dual inhibition of MAPK pathway with BRAF and MEK inhibitors, an approach that has improved response rates in melanoma. Several studies indeed showed that MAPK signaling rebound in glioma can be overcome by adding MEK inhibition, including our work with a BRAF^V600E murine glioma cell line (2341) which were derived from an adenovirus-induced tumor in a Cre-inducible mouse model in which BRAF^V600E is expressed under the endogenous promoter, and hence under physiologic levels (22) ( Figure 1 ). BRAF^V600E inhibitor monotherapy with dabrafenib of 2341 cells recapitulated the initial reduction followed by reactivation of MAPK signaling and accordingly 2341-derived orthotopic glioma transiently respond to BRAF^V600E inhibitor monotherapy with initial responses followed by continuous tumor growth. Combining BRAF and MEK inhibitors (dabrafenib+trametinib) suppressed MAPK pathway signaling in a more durable fashion (57, 80). Tumor regression from the combined dual MAPK pathway blockade with BRAF and MEK inhibitors was also confirmed in BT40 PDoX, which are described in greater detail below (81). In addition, combination therapy with PLX4720 + MEK inhibitor mirdametinib (PD0325901) (81) showed durable suppression of the MAPK pathway and this resulted in longer survival of animal subjects, further confirming our earlier data.

These studies combined prompted subsequent clinical evaluation of Vemurafenib against BRAF^V600E-mutated gliomas (11) and more recently clinical use of BRAF and MEK inhibitor combination treatment in patients with BRAF^V600E-mutated CNS tumors. Clinical responses showed an improvement in response rates with combination therapy over monotherapy (13). These data lead to the recent FDA approval of dabrafenib+trametinib for LGG and provide a paradigm for successful translation of preclinical data based on high-fidelity tumor models, including the 2341 model which expresses BRAF^V600E under the endogenous promoter like that in human tumors (57).

Continuous studies into novel combination therapies are important because of therapy resistance to vertical inhibition of MAPK signaling and cross-resistance to other BRAF/MEK inhibitors and BRAF/EGFR inhibitors in the clinic. Combinations investigated are of BRAF^V600E inhibitors with cdk4/6 inhibitor PD0332991 (22), PLK1 inhibitor tool compound BI2536 (82), radiation (83), MEK inhibitor AZD6224 + mTOR inhibitor everolimus (80), and EGFR inhibitor HKI-272 (84). These studies have used 2341 in addition to BRAF^V600E mutated CDKN2A^Null murine glioma cells derived from mice expressing BRAF^V600E using the CRE/Lox system under the nestin and GFAP-promoter, as well as established cancer cell lines.

Noteworthy, the 2341 orthotopic model can be implanted in fully immunocompetent mice allowing studies of immunomodulation and immunotherapy in the context of BRAF^V600E-mutated gliomas. Using the 2341-BRAF^V600E model, we have recently demonstrated that BRAF/MEK inhibition alters the tumor immune infiltrate and sensitizes tumors against immune checkpoint blockade by anti-PD-L1 and anti-CTLA4 treatment, which resulted in better clinical outcomes in mice implanted with BRAF^V600E mutated high-grade gliomas (3).

In conclusion, preclinical data obtained using murine glioma models that faithfully recapitulate the human tumors’ genetics and phenotypes have high predictive capacity.

Conventional cancer cell lines are customarily grown as adherent monolayers on plastic dishes or laminin covered flasks in high-serum medium (typically supplemented with 10% FBS), for multiple generations. Although they are robust and easy to grow in this manner, the high-serum medium prompts cells to differentiate towards astrocytic phenotype, thus generating homogeneous cell cultures, which is unrepresentative of glioma’s biology. Moreover, the parental tumor of conventional cancer cell lines is frequently unknown and/or uncharacterized. As such, these models are useful for assay optimizations, including commercially available kits, imaging, monitoring morphological changes and optimizing antibody staining conditions. However, 2D monolayer architecture poorly represents patient-specific heterogeneity and affects tumor intrinsic features related to cell signaling, phenotype, cell-to-cell interactions, and drug response. The 2D cell arrangement generates inappropriate cell density and nutrient gradients, unphysiological oxygen levels, underrepresented spatial context, and artificial interactions with extracellular matrix and tumor microenvironment (85). In contrast, patient-derived cell lines (PDCLs) are initiated from tissue obtained during craniotomies, biopsies, and autopsies. Rapidly dissociated surgical samples are typically seeded in low-attachment cell culture flasks, and generate 3D spheres within a few days. Moreover, the culturing medium is serum-free and is commonly supplemented with growth factors (EGF and FGF-2), which perpetuates the stem cell phenotypes frequently observed in glioma cells and establishes robust glioma stem cell (GSC) cultures. Such patient-derived 3D spheroid cultures preserve features of the original tumor better than conventional 2D cancer cells lines (86) because they demonstrate higher proliferation and invasive capacity, high tolerance towards chemoradiation and sensitivity towards T- and NK-mediated cell killing (85, 87). This being said, PDCLs can be used as personalized platforms for drug testing and therapy tailoring (88), such as evaluating the benefit of single BRAFi vs combined with MEK/mTOR/EGFR inhibitors. A side-by-side comparison of patient-derived GBM cells has clearly demonstrated loss of malignant potential, decreased morphologic, and cell type heterogeneity, and divergence of expression profiles in cells grown in 2D differentiating conditions (high serum), versus 3D un-differentiating (serum-free) medium (89). Proliferation rates are frequent measures of drug activity in cell-based assays. These are significantly increased in adherent cells, while 3D spheroid cultures exhibit more stable doubling times even after multiple rounds of passaging. Furthermore, conventional 2D cultures are unreliable for functional analysis, as they undergo genetic drift over time, emphasizing the need for 3D spheroid models of CNS tumors ( Table 1 ).

Our group recently generated a stable spheroid PDCL from a patient with a high-grade BRAF^V600E mutated tumor that also carried a TP53 and RB mutation (STN-10049) (3). This cell line was sensitive to the anti-proliferative effects of combined dabrafenib and trametinib treatment in vitro, although a marker for therapy resistant cells CD133 (82) was upregulated with dual MAPK inhibitor treatment, and the patient worsened rapidly despite aggressive chemotherapy and dabrafenib-trametinib treatment (3). Cells derived from the same patient’s tumor at autopsy showed decreased sensitivity to dabrafenib-trametinib, and increased expression of CD133, consistent with the patient’s lack of response and cancer progression on the drug. Thus, patient-derived cells derived from different timepoints (pre- and post-treatment) recapitulate longitudinal the patient response to therapy.

Several PDCLs representing various BRAF^V600E-mutated CNS tumor subtypes have been generated and published by individual labs (see Table 1 for a comprehensive list), and these lines are important novel tools available to investigators. Successful culturing and careful expansion of these models while monitoring for retention of driver alterations and genetic drift will ensure deep, comprehensive understanding of the biological and functional impact of BRAF^V600E mutation in glioma.

Molecular analyses of BRAF^V600E-mutated glioma PDCLs are still scarce ( Table 1 , Other Alterations Reported), and the congruence of these cell lines with their tumor of origin has yet to be determined. Large scale initiatives such as the Human Cancer Model Initiative (HCMI) in the USA have been launched with the purpose of generating next-generation models, whereby “next-generation” refers to models being clinically annotated and are characterized molecularly and phenotypically for their resemblance to their parental tumor (see HCMI searchable catalogue). HCMI cell lines can be purchased from the American Type Culture Collection (ATCC). As of May 2023, no BRAF^V600E-mutated CNS tumor cell lines have been reported in the HCMI searchable catalogue, but the efforts of the HCMI are ongoing, and expected to include more pediatric models soon.

In summary, PDCLs grown under 3D spheroid conditions are superior models of the human disease than conventional cancer cell lines and multiple individual labs. Novel BRAF^V600E PDCLs are continuously reported by individual investigators ( Table 1 ) and larger scale model development initiatives and collectively these models will provide an important toolkit for robust preclinical and mechanistic studies of drug resistance and targetability.

Certain types of CNS tumors, including low-grade glioma, have a high failure rate for developing into patient-derived cell lines. Implanting surgical tissue directly into mice to generate patient-derived xenografts (PDX) and derive cell lines from PDX tissue is a valid alternative approach to culturing cell lines from surgical tissue directly. The PDX-to-cell line approach was used to generate a heterozygous BRAF^V600E-mutated cell line BT-40 (63). BT-40 cells exhibited sensitivity to MAPK inhibitors selumetinib and trametinib (64, 105). A MAPK inhibitor library screen against BT-40 expressing a luciferase-based MAPK reporter assay revealed distinct sensitivities to single inhibitors, including next-generation RAF inhibitors, and synergistic effects of different inhibitor classes (106). BT-40 was also used to assess the clinical potential of ulixertinib, a first-in-class ERK inhibitor, and to test combinatorial effects with MEK inhibitors or BH3-mimetics (107).

However, it is unclear how well the low-grade parental tumor phenotype is preserved in BT-40 cells since they have been cultured in serum after passaging them as PDX. An additional caveat of patient-derived orthotopic xenografts (PDoX)-derived cell lines is potential contamination with mouse cells since they are hosted in mice brains; However, the BT-40 cell line had identical short tandem repeat analyses to the parental tumor and was free from mouse cells, as determined by LDH isozyme analyses. Whether BT-40 exhibits a gain of chromosome 7q identical to the xenograft has not been determined (105). In conclusion, PDX serve as an alternative source for cell lines and their ability to recapitulate the parental tumor needs to be monitored as it might change over time.

There are clear limitations with existing models, the most prevalent limitation being that most of these models represent high-grade CNS tumors with combined BRAF^V600E expression and CDKN2A deletion; progression models that recapitulate progression from low grade to high grade tumors are not available. Combinations of BRAF^V600E with other rarer coinciding alterations are rarely represented. CNS tumors display marked inter-tumoral heterogeneity, which emphasizes the need for using multiple cell lines for each tumor class. Unfortunately, there are not enough cell lines available to the community to capture representative patient-to-patient differences. Moreover, despite the relative frequency and poor clinical outcome of BRAF^V600E mutant-brain metastases, there is only one published genetically engineered model for BRAF^V600E melanoma metastatic to the brain (53, 108), which severely hinders studies of the mechanisms underlying the metastatic process.

To study the precise role of BRAF^V600E mutation in gliomagenesis in vivo, various mouse models of BRAF^V600E-mutated glioma have been developed to better understand the complexity of tumor biology. Transgenic or genetically engineered mouse models (GEMMs), syngeneic models, patient-derived xenografts (PDX) in immunocompromised mice, and virus-induced models are commonly used to recapitulate certain biological features of primary patient BRAF-mutated tumors ( Table 3 ). Although these models provide a valuable resource for studying how BRAF^V600E is involved in tumor development and recurrence following therapy, each model comes with its own benefits and limitations that need to be carefully considered.

These models involve the implantation of tissue into the flank and dissociated cells into the brain to obtain orthotopic xenografts (PDoX). Although the growth of PDoX is harder to follow due to their inaccessibility in the brain, they provide a more faithful representation of the tumor microenvironment and invasiveness of CNS tumors than flank tumors. Severe combined immunodeficient (SCID), non-obese/diabetic (NOD)/SCID, athymic nude, and NSG mice are typical hosts for PDX development, due to their impaired immune systems, which facilitates achieving graft survival without risk of short-term rejection. Although these models can mimic certain features of the original human tumors, such as genetic and molecular heterogeneity, the biological aspect of tumor initiation and growth cannot be studied. Moreover, they do not fully capture the immune system’s role in controlling glioma growth and progression, deeming them unsuitable for studying tumor immunology and evaluating immunotherapy response.

Establishing PDX models for BRAF^V600E-mutated gliomas can also be time-consuming with a low success rate of engraftment (109) depending on the availability of patient tumor tissue with several factors taken into consideration, such as how the tissue is collected and stored. Several PDX biobanks have been established by individual groups and institutions, including PDX panels from GBM (117), and some specifically for pediatric brain tumors (118, 119). BRAF missense mutations have been identified in these PDX by molecular profiling, but due to the relative scarcity of BRAF^V600E altered CNS tumors amongst brain tumor patients, PDX for these tumors are rare. A PDX biobank specifically focused on BRAF^V600E-mutated gliomas is currently lacking due to the scarcity of patient material, and the difficulty in retrieving complete patients’ clinical data, pathologies, gene expression profiles, and drug responsiveness (64). However, there is great potential for this to be an authoritative resource for developing new treatments and personalized medicine approaches.

These models are created by implanting an established tumor cell line derived from the same species as the host animal. The 2341 cell line of BRAF^V600E-mutated and CDKN2A-deleted glioma injected into both immunodeficient (NSG) and immunocompetent (FVB/N) mice have revealed successful engraftment and has been critical for the preclinical assessment of experimental therapeutics (57). The use of syngeneic models of BRAF^V600E-mutated glioma enables the study of the immune microenvironment associated with the tumor, and the testing of immunotherapies (3). However, syngeneic models may have limited translational potential since they do not fully recapitulate the biological and clinical features of human gliomas. Species-dependent differences in tumor microenvironment, immune response, tumor heterogeneity and therapeutic response are important factors that need to be taken into account. Multiple models generated from syngeneic mouse tumor lines for a particular type of glioma such as glioblastomas could also be variable in immune phenotypes among tumors, translating these findings into effective immunotherapeutic strategies for human glioma patients could be complex and warrants further research and validation (120). In addition, these models may exhibit inconsistent tumor growth rates, due to a functional immune response in host mice, which may affect the reliability of experimental results and require a large study cohort.

Knock-in GEMMs for BRAF^V600E-mutated gliomas are alternative models that involve conditional and inducible systems such as the Cre-loxP system to allow the expression of BRAF^V600E under the control of the endogenous promoter and enhancer sequences in a cell-specific or time-specific manner upon the activation of Cre recombinase (57, 121), which can be induced by delivering either tamoxifen or Cre-expressing viruses such as adeno-associated viruses and adenoviruses (22, 111). The use of the inducible Cre-loxP system combined with knock-in GEMMs has provided valuable tools for understanding the effects of BRAF^V600E mutation in tumor progression in the context of an intact immune system and microenvironment, and for developing potential therapies that specifically target these mutations.

Another type of GEMM for BRAF^V600E-mutated gliomas is the replication-competent avian sarcoma-leukosis retrovirus and the corresponding avian leukosis virus A (RCAS-TVA) system, which allows for somatic gene delivery to cells that are engineered to express cell surface receptor TVA using a cell type-specific promoter such as Nestin, which is a common marker for neural stem cells (112, 113). This enables investigation of the tumor cell of origin after BRAF^V600E mutation, as well as the role of BRAF in glioma development.

Transposon-based GEMMs have been developed for studying glioma development. These models include the Sleeping Beauty transposon system and the PiggyBac transposon system by which a transposon containing a mutation is integrated into the genomes of targeted cells of interest, resulting in the development of malignant tumors in the brain (122, 123). Specifically, BRAF^V600E mutation in neural progenitor cells induced by the binary PiggyBac transposase system has revealed the formation of GG-like tumors only in concert with the activation of Akt/mTOR signaling (115), as well as a hyperexcitable phenotype in neocortical pyramidal neurons with increased neuronal firing frequencies (114). This is also supported by another study showing a causal relationship between BRAF^V600E mutations and epileptic seizures, which are often associated with GG and PXA (111) ( Table 3 ).

The inter-tumoral heterogeneity of BRAF^V600E mutated gliomas provides an additional modeling challenge. PXA express reticulin fibers and are therefore thought to originate from transformed subpial fibrous astrocytes (124), whereas GG consisting of mixed neuronal and glial components points to a neural stem cell origin. Since there is an ongoing debate over the cell of origin for gliomas with Nestin-expressing neural stem and progenitor cells (NSCs) and oligodendrocyte progenitor cells (OPCs) being the two major candidates to transform into tumor cells and to shape tumor heterogeneity, it is important to characterize the relative contribution of each population to glioma initiation and progression in the context of BRAF^V600E mutation.

Moreover, no studies have yet been conducted to determine the OPC nature of BRAF-mutated tumors using either the RCAS-TVA system or the Cre-lox system with OPC-specific promoters such as PDGFRα or NG2 ( Table 3 ). This is particularly important since BRAF^V600E expression in Ink4a/Arf-deleted cells has been shown to increase the proportion of proliferative glial progenitor cells in adult mice in vivo, disrupt asymmetric cell division of OPCs, and increase OPC frequency in vitro at the expense of mature oligodendrocytes and astrocytes, respectively, without altering NSC frequency (3). These data suggest that OPCs are more sensitive to cellular transformation by concurrent BRAF^V600E expression and Ink4a/Arf loss than NSCs. These two alterations have yet to be combined in transgenic mouse models with OPC type-specific expression, to address whether OPCs can indeed be the origin of BRAF^V600E mutated tumors in vivo. Exploring whether BRAF^V600E mutation can also occur in mature astrocytes or oligodendrocytes that de-differentiate into stem or progenitor cells with tumorigenic potential would be intriguing to explore, since both astrocytes and neurons have been shown to function as the cells of origin for gliomas via dedifferentiation (125).

Nevertheless, the cellular origin of BRAF-mutated gliomas remains underappreciated, and further studies are required to understand the cellular and molecular mechanisms underlying the development of these tumors. Distinct cellular origins are just one potential explanation for the phenotypic diversity of BRAF^V600E-mutated tumors; alternative explanations such as microenvironmental diversity, and distinct neurodevelopmental origin amongst the tumors with distinct age presentation are important underlying factors to be considered.

Establishing robust in vitro models that are representative of low-grade CNS tumors, incuding BRAF^V600E -mutated CNS tumors is inherently challenging. Low grade tumors are notoriously difficult to model in vitro and in vivo, likely due to tumor cell differentiation states, slow growth senescence, and microenvironmental signals in culture media that favor outgrowth of normal brain cells. Indeed, certain copy number alterations have been associated with robust in vitro growth, including gain of chromosome 7p, and loss of chromosome 10q. Loss of PTEN (126) and CDKN2A loci (127) have been positively associated with the establishment of PDCLs. While high-grade tumors tend to have a higher frequency of establishing cell lines, it is challenging to maintain the phenotypic intra-tumoral heterogeneity that is so characteristic of high-grade CNS tumors; in addition, certain genetic alterations (ATRX, IDH1, and hTert) tend to get lost in cultured cells presumably due to negative selective pressure from culture conditions (128). Recent studies have successfully approached generating adult LGG marked by IDH1 alterations by picking slow-growing tumor cells and separating them from fast-growing normal fibroblasts, that typically outgrow from mixed patient-derived cultures (127). Moreover, patient-derived organoid cultures produced faithful in vitro models for adult IDH1-mutated LGG (129). Most in vitro BRAF^V600E-mutated CNS tumor models created thus far are high-grade models, including 2341, AM38, DBTRG-M05, STN-10049. Encouragingly, some models for LGG are also reported, including BT-40 and IC3635-PXA (see Tables 1 , 3 ). Care must be taken that these models retain their low-grade features mentioned above since it is well-known that tumor cells can spontaneously become more aggressive when passaged in vitro over time.

In conclusion, attempts to generate in vitro models for low-grade CNS tumors have been hampered by negative selection against tumor cells in culture and phenotypic drift to a high-grade tumor over time. Recent studies from adult tumor types have provided new guidance for generating urgently needed robust in vitro models of pediatric CNS tumors in general and BRAF^V600E pLGG in particular.

Adult LGG commonly evolve into higher grade tumors, while malignant transformation occurs less frequently in pLGG, with BRAF^V600E-mutated pLGG being an important exception (5, 18, 20, 27). Most in vitro and in vivo models in the field of neuro-oncology are for HGG, and genetically engineered mouse models recapitulating BRAF^V600E-mutated LGG and their progression to high-grade are lacking, due to the slow-growing nature of these tumors. Thus, the biological mechanisms underlying the onset of LGG formation remains poorly understood and molecular and cellular mechanisms for progression to a more aggressive clinical course of BRAF^V600E-mutated gliomas are unknown. To create an ideal preclinical model for BRAF^V600E-mutated LGG that represents the complex and dynamic tumor microenvironment of malignant transformation, the grading and classification of gliomas need to be recapitulated as closely as possible in animal models. As indicated in Figure 3 , the WHO grading system consisting of cellular density, nuclear atypia, mitosis, necrosis, and endothelial proliferation is commonly used to evaluate malignancy based on the tumor’s histopathological appearance with high grade tumors having more aggressive microscopic features than low grade tumors (130). High grade tumors often have increased nuclear pleomorphism with variability in shape and size, as well as elongated, large, and hyperchromatic nuclei. They also have increased cellularity, mitotic activity, necrosis, and vascular hyperproliferation. High grade tumors have a higher Ki67 proliferation index indicative of more aggressive tumor growth and they tend to exhibit high immunoreactivity for stem cell markers such as CD133, Nestin, Sox2 and Olig2. However, functional roles and contribution to tumor behavior during progression and treatment response of stem cells in LGG are still being investigated, and therefore the clinical significance of stem cell marker expression in glioma progression remains unclear. Asymmetric divisions of glioma cells regulate adult tumor cell fate and perpetuate a stem-like phenotype and contribute to tumor expansion (82, 131, 132). Regulation of asymmetric division in LGG has yet to be determined.

To further improve the classification of gliomas due to the limitations of the histological grading system such as subjectivity with inter-observer variabilities and intra-tumor heterogeneity (133), molecular markers are often integrated into a more clinically relevant grading scheme. Some alterations such as CDKN2A/B loss, PTEN, EGFR, NF1, and TERT promoter mutations have been shown to be associated with very poor clinical outcomes in patients with BRAF-altered gliomas (17). However, some studies reported conflicting results regarding the prognostic value of some molecular markers (133), emphasizing the importance of understanding the molecular mechanisms underlying glioma progression. This poses another challenge in generating preclinical models of LGG due to the difficulty in recapitulating the widespread genomic and epigenomic effects of BRAF alterations. Despite the mechanistic challenges in creating an ideal mouse model of BRAF^V600E-mutated LGG, it is feasible using the genetic engineering techniques with the following factors considered: 1) cell type-specific BRAF^V600E mutation induced by either Cre-loxP or RCAS-TVA system, 2) selection of the appropriate genetic background of the mouse which can influence the development and progression of tumors, 3) monitoring of tumor progression by MRI to detect the presence and growth of tumors, 4) validation of the model by comparing its histopathological and molecular characteristics to human BRAF-mutated LGG, and 5) the validated mouse model can be used for preclinical research to test drugs to better understand the underlying biological mechanisms of BRAF-mutated LGG and this will pave the way for the identification of therapeutic targets to block tumor progression.

In conclusion, LGG models are needed to better understand the role for stem cells and asymmetric division in tumor formation and progression. Due to the complexity of generating a mouse model that accurately recapitulates the genetic background, intra-tumoral heterogeneity, and tumor microenvironment that all closely resemble those of human tumors, multiple models may need to be generated and evaluated before an ideal model is established.

Recurring oncogenic alterations are difficult to maintain in vitro, including IDH1 mutations, which are reportedly frequently lost in PDCLs grown under standard serum-containing conditions (134). One study recently reported successfully generating several IDH1 mutated glioma cell lines by growing them under serum-free conditions and selecting cells manually based on their tumor cell morphology and separating them physically from the non-tumor associated cells that are IDH-wildtype (127). Such labor-intensive approaches might not be necessary when attempting to culture BRAF^V600E-mutated PDCLs, since BRAF^V600E expression was maintained in culture and upon xenografting in vivo, suggesting that cells remain dependent on activated MAPK pathway signaling (3). Notably, one study reports an increase in allelic frequency (from 28% in the patient tumor) to 70% in PDoX, and up to 69% in PDoX-derived cell lines, indicating that BRAF^V600E plays an important role in tumor progression (109). In conclusion, BRAF^V600E expression might be required for tumor cell growth and therefore maintained in culture. Whether BRAF^V600E expression provides a selective disadvantage or advantage in cultured cells has yet to be systematically explored.

One such technology is non-invasive imaging techniques including magnetic resonance imaging (MRI) and bioluminescence imaging (BLI), which enable the visualization and monitoring of tumor growth and assessment of tumor response to treatments in mice. MRI allows the identification of tumor size, shape, location, and invasion into surrounding tissues, whereas easily accessible and commonly used BLI can only monitor the growth of tumor cells that have been engineered to express luciferase, an enzyme that produces light that can be measured throughout BLI when the luciferin substrate is given. However, in contrast to MRI, there are some limitations of BLI that need to be seriously considered: 1) it has limited penetration depth of light so tumor cells located deep in the brain could be hard to detect, 2) it has a limited dynamic range, making it difficult to accurately measure tumor growth changes; the bioluminescence signal measured from tumor cells can sometimes saturate at high cell densities at later stages, and 3) unreliable quantification of BLI signal due to variations in the expression of the luciferase reporter gene in tumor cells. Indeed, a recently published study has demonstrated that the discrepancy in the evaluation of tumor growth by BLI and MRI was attributed to the instability in luciferase expression in the viral construct (136). In contrast, MRI allows for more accurate detection and measurement of tumor size and location in the brain, enabling researchers to perform longitudinal studies, where the tumor growth and progression can be monitored by analyzing changes in the tumor size and shape over time. As a result, BLI should ideally be used in conjunction with MRI if feasible at several time points to obtain complementary information for the accurate interpretive analysis of tumor growth and response to therapy in preclinical studies.

In terms of the assessment of the response to BRAF/MEK inhibitor combination therapy, biochemical analyses for evaluating kinase inhibitor activity in the brain are a time-consuming process that often involves tissue dissection, which may lead to the disruption of the activity being measured. Novel kinase-modulated bioluminescence indicators enable noninvasive imaging of signaling strength and could help tracking the activity of target kinases in response to targeted drug treatment (137). Such approaches represent valuable tools for improving drug discovery against BRAF^V600E mutated glioma and they might help with developing biomarkers which are urgently needed to stratify patients and identify those that will need additional treatments in addition to BRAFi+MEKi combination therapy.

Accumulating evidence suggests that BRAF^V600E-mutated gliomas are associated with an increased risk of epilepsy and seizure, which can have a range of behavioral consequences based on their frequency and severity. As indicated in Table 3 , some studies using mouse models revealed that BRAF^V600E mutation arising from progenitor cells during brain development leads to the acquisition of intrinsic epileptogenic properties in neuronal lineage cells and tumorigenic properties in glial lineage cells. This somatic mutation contributes to epileptogenesis by upregulating the expression of the RE1-silencing transcription factor which can repress a subset of genes coded for ion channels and receptors that are crucial to neuronal function (111, 138). It has also been demonstrated that BRAF^V600E mutation in neural progenitor cells results in a hyperexcitable neuronal phenotype (114). Since the BRAF^V600E mutation has been strongly linked to changes in neuronal excitability, it would be intriguing to determine whether there is a cognitive impairment as a result of glioma-associated epilepsy induced by BRAF^V600E mutation. Nevertheless, further studies are required to better understand the relationship between BRAF^V600E -mutated gliomas and epilepsy and the underlying mechanisms and potential therapeutic implications. Currently, behavioral assessment in mouse models of BRAF^V600E-mutated glioma that involves testing of motor and cognitive functions is lacking and thus is important to explore. This will also validate whether these preclinical models can mimic human patients’ cognitive decline and epileptogenic properties.

Intra-tumoral heterogeneity has been strongly implicated in affecting treatment response and conferring resistance in glioma with the underlying cellular and molecular mechanisms remaining fully unknown (139). Furthermore, the molecular responses of high versus low grade tumor cells carrying the BRAF^V600E mutation to BRAF and MEK inhibition have not been investigated in detail and thus the differential responses in patients with LGG and HGG to these therapies are poorly understood (13). Recent advances in single-cell sequencing technology would help address these questions since it allows for the integrative analysis of the genome, transcriptome, epigenome, proteome, and/or metabolome-based characterization of the individual cell among cancer cells within a tumor. Single-cell transcriptomic analysis of A375 (BRAF^V600E mutant) melanoma cell line described functional and phenotypic changes during development of drug resistance towards targeted therapy. Overall, four different treatment conditions were applied to the A375 cells (Vemurafenib monotherapy, V + Cobimetinib, V + trametinib, and untreated control). ScRNAseq revealed that the initial events of mono-drug resistance include loss of differentiated antigen presenting and neural crest-like cells and enrichment of undifferentiated, heterogeneous, highly proliferative stromal-like cells. Double-agent resistance featured variable amounts of dedifferentiated stromal cells because of heterogeneous effects of therapies on cellular states. In addition, specific survival machinery related to MAPK, p53 and PI3K/Akt reactivation and pluripotency induce distinctive cell populations (140).

Clonal barcoding is a powerful single-cell technique that allows researchers to track clonal evolution trajectories of tumor cells and to identify genetic changes and chromosomal aberrations that drive tumor growth and progression, or tumor cell resistance over time. A recently developed “CAPTURE” single-cell barcoding approach has been utilized in combination with an in-depth multi-omics analysis to reveal the clonal dynamics of BRAF^V600E melanoma cells’ response to vemurafenib and identify novel targetable vulnerabilities of resistant clones such as oxidative phosphorylation and E2F pathways (141). This approach can be applied to study treatment resistance in the context of BRAF^V600E mutated gliomas and will help uncover new mechanisms of tumor cell resistance in the brain. Another cell barcoding strategy was applied to study rare subpopulations, by cloning lentiviral library in which the barcode sequence is incorporated in the 3` untranslated region of the green fluorescence protein (GFP). Through sequencing, barcodes present in resistant samples were used to identify and trace rare, primed cells in the untreated samples. In this way, rare, drug-naive precursors were enriched by FACS and characterized and potential mechanisms of resistance were identified. Interestingly, shortly after drug treatment, these cells highly expressed phosphorylated ERK which confirmed earlier studies that BRAF inhibitor effects on MAPK signaling are only transient (142).

In conclusion, single-cell barcoding can be applied to study treatment resistance in the context of BRAF^V600E mutated gliomas and once applied to the brain will help uncover new mechanisms of tumor cell resistance.

CRISPR-Cas9 gene editing technology is now used more often in combination with other systems for developing glioma mouse models (143). Oldrini et al. combined the CRISPR-Cas9 system with the RCAS-TVA system to generate more precisely targeted glioma mouse models for biological and preclinical research including the one for BRAF-mutated glioma by inducing gain-of-function BRAFV637E mutation (orthologous to human BRAF^V600E) in NSCs (116). However, the CRISPR-Cas9 system suffers from a major limitation of off-target effects, which remain to be addressed.

Recently, 3D organoid culture systems generated from induced pluripotent stem cells (iPSCs), human embryonic stem cells (hESCs), or patient-derived adult stem cells from cancerous tissues have been developed to better capture complex molecular and cellular heterogeneity. In the search for pre-clinical glioma models, remarkable advancements have been made in creating organoid models for gliomas, such as neoplastic cerebral organoids (neoCORs), GLICO, and patient-derived glioblastoma organoids (GBOs) (144–148). These models retain many key features of primary patient tumors, providing accurate models for studying cellular and molecular interactions among and within tumors and thus represent new advanced ex vivo models for disease modeling, omics analysis, genetic manipulation, and personalized drug screening. However, the lack of vasculature, stromal and immune cells in these organoids makes it difficult to study tumor-immune microenvironment interactions. Nonetheless, the exciting organoid technology holds tremendous potential to significantly change our understanding of BRAF^V600E-mutated glioma tumor biology.

Due to low engraftment rates and inconsistent tumor growth of PDX and syngeneic models, various hydrogels have been developed to improve the reliability of mouse models. Although Matrigel is commonly used as a gold-standard hydrogel for 3D cell culture and experimental models (149), batch-to-batch variability in the complex composition of Matrigel can lead to inconsistencies in experimental results, especially when comparing data between different studies using different batches of Matrigel. Therefore, alternative hydrogel scaffolds such as the biomimetic EKGel, tissue-derived extracellular matrix hydrogels, and alginate-based hydrogels have been developed to overcome these limitations (150, 151). Applying one of these advanced hydrogels for culturing BRAF^V600E-mutated organoids or for encapsulating BRAF^V600E-mutated tumor cells upon intracranial delivery can improve the consistency of models of BRAF^V600E mutated gliomas.