Preclinical Characterization of a Next-Generation Brain Permeable, Paradox Breaker BRAF Inhibitor

In the last decade, the prognosis of metastatic melanoma has been greatly improved for BRAF V600E/K-positive tumors by targeted therapies and immunotherapies (1).

Three BRAF inhibitors, vemurafenib, dabrafenib, and encorafenib, are approved in combination with MEK inhibitors (cobimetinib, trametinib, and binimetinib) for the treatment of patients with BRAF V600E positive melanoma (2), as this results in superior efficacy over single-agent BRAFi and because BRAFi monotherapy is typically associated with the emergence of secondary neoplastic lesions due to their ability to trigger paradoxical RAF activation and due to rapid acquisition of drug resistance (3, 4).

The paradoxical RAF activation is the peculiar characteristic of first-generation BRAFi to selectively inhibit MAPK signaling in BRAF V600E/K cells while triggering MAPK signaling activation in other contexts such as in the presence of RAS (HRAS, KRAS, NRAS) and/or receptor tyrosine kinase (RTK) signaling. This phenomenon is mediated by the ability of BRAF V600E/K to promote MAPK activation as monomeric protein; a conformation effectively inhibited by available BRAFi whereas, in BRAF WT settings and in several documented BRAFi resistant states, the kinase forms homo or heterodimers with other RAF members (5–8). In the RAF dimeric state, one of the two protomers is bound to the inhibitor whereas the second acquires a conformation unfavorable for the drug binding which consequently enables signal propagation (9, 10).

Although MEKi co-administration with BRAFi can blunt the BRAFi-mediated paradoxical RAF activation by blocking MAPK from a downstream node, these agents present significant on-target intrinsic toxicity and a narrow therapeutic window due to the essentiality of MAPK for nontumoral tissue homeostasis (11). Moreover, resistance to BRAF/MEK combination is often driven by genetic events enabling RAF dimerization, thus suggesting the hypothesis that clinically achievable MEK inhibition might still be overcome by excessive activation of RAF-dimer mediated signaling in tumors (12, 13).

Brain metastases in melanoma represent an area of significant unmet medical need. Studies indicate that 7% to 25% of patients with melanoma present with brain metastasis already at diagnosis and an even higher rate of up to 75% are observed in the later disease stages (14–17).

Among the three approved BRAFi, dabrafenib, in combination with trametinib, has been extensively explored for the treatment of BRAF-mutant melanoma with brain metastasis. Although the response rate of that combination was similar between intracranial and extracranial lesions, the duration of response was significantly shorter in patients presenting with brain lesions (17).

Moreover, despite the initial response, in the majority of patients, disease relapse occurred only intracranially suggesting that suboptimal drug penetration might represent a limiting factor in controlling brain metastatic disease (17).

Here we present the preclinical characterization of a novel BRAFi, referred to as Compound Ia (Fig. 1A), a new molecule designed to address the key limitations of currently available BRAF-inhibitors, characterized by the ability of not inducing the RAF paradox (paradox breaker) in non-BRAF mutant settings and having excellent brain penetration properties.

The cell lines A375 (CRL-1619), A375 NRAS^Q61K (CRL-1619IG-2), HCT116 (CCL-247), Colo205 (CCL-222), YUMM1.7 (CRL-3362), Mia PaCa 2 (CRL-1420), and A549 (CCL-185) were obtained from the ATCC whereas Patu8902 cells (ACC 179) from the German Collection of Microorganisms and Cell Cultures (DSMZ). The A375-LUC line engineered to express firefly luciferase was generated in house through standard lentiviral transduction methods (Supplementary Materials and Methods). Cells were maintained in a humidified atmosphere with 5% CO₂ at 37°C. Cell lines authenticity was confirmed through short tandem repeats-PCR analysis performed at Microsynth. Absence of mycoplasma contamination was verified through testing of antibiotic-free cultured cells for 10 to 14 days through the kit MycoAlert Mycoplasma Detection Kit (#LT07–318, Lonza).

Encorafenib (HY-15605), dabrafenib (HY-14660A), binimetinib (HY-15202), and cobimetinib (HY-13064A) were purchased from MedChemExpress, PLX8394 from Selleckchem (S7965), the novel BRAFi Compound Ia was designed and synthesized (Supplementary Materials and Methods) taking into consideration previously reported paradox inducing BRAF inhibitors with the quinazolinone scaffold (18).

Binding affinity determination of Compound Ia was determined through the KinomeScan competition binding assay performed at Eurofin DiscoverEX. Enzymatic assays on the purified WT BRAF proteins and the mutants BRAF V600E/K/A/D were performed at Reaction Biology according to methods described previously (19). Figure 1B was generated using TREEspot Software Tool and reprinted with permission from KINOMEscan, a division of DiscoveRx Corporation, © DISCOVERX CORPORATION 2010.

Homogeneous time resolved fluorescence (HTRF) assay for the measurement of P-ERK levels was performed according to manufacturer's instruction by using the Advanced ERK phospho-T202/Y204 HTRF assay (Cisbio, #64AERPEH) and the Total ERK HTRF Assay Kit (Cisbio, #64NRKPEG). Cells were plated at a density of 8,000 cells/well in 12 μL into a low volume 384-well microplate (Greiner Bio-One, HiBase, #784–080) in DMEM without phenol red (Gibco, #21063029) supplemented with 10% FBS (Gibco, #10270–106). Compound treatment was performed for 1 hour with an 11-point dilution curve starting from the concentration of 10 μmol/L. FRET signal was measured using the PHERAstar FSX plate reader.

Western blot analysis was performed on lysates generated upon cell incubation with IP Lysis buffer (Pierce, #87788), supplemented with 1× protease/phosphatase inhibitor cocktail (Thermo Fisher Scientific, #78444). 5 to 30 μg of protein were mixed with Laemmli Sample Buffer (BioRad, #161–0747) and separated using gradient (4–20%) Criterion TGX precast gels (BioRad, #345–9898). Upon transfer to nitrocellulose membrane and saturation with a non-animal protein blocking buffer (Cell Signaling Technology, #15019), membranes were probed 1 hour at room temperature or overnight at 4°C with primary antibodies. Membranes were washed in TBS tween buffer, incubated with HRP-conjugated secondary antibody for 1 hour at room temperature. The detection of the bands was carried out through chemiluminescence reaction (Advansta, #K12043) and images collected through a Fusion FX system apparatus (Vilber Lourmat). The primary antibodies against P-ERK 1,2 (#4370), ERK 1,2 (#9102), P-MEK 1,2 (#9154), c-RAF (#53745), BRAF (#14814), and GAPDH (#5174) were purchased from Cell Signaling Technologies whereas anti Vinculin (#V9131) from Sigma Aldrich.

Cell line panel profiling was performed at Oncolead GmbH (Germany) on a panel of 94-cell lines representative of different tumor histologies but also comprehensive of freshly isolated peripheral blood mononuclear cell (PBMC) nontumoral control cells. Drug treatment was performed for 72 hours at the concentrations of 10, 1, 0.1, 10, 1, and 0.1 nmol/L. Cell cytotoxicity was determined through protein quantification based on a Sulforhodamine B staining method (20).

Colony forming assay: 500 to 1,500 cells were plated in 12-well plates and incubated overnight to allow seeding. Upon drug treatment cells were cultured for additional 7 to 10 days and subsequently fixed and stained with a solution of 10% methanol, 90% crystal violet.

Mice were maintained under specific pathogen-free conditions with daily cycles of 12-hour light/12-hour darkness according to international (Federation of European Laboratory Animal Science Associations) and national guidelines. The study protocols were reviewed and approved by the local government. In all experiments, female mice were 6 to 9 weeks of age at the experiment initiation.

Mouse strains utilized in the experiments were: CB.17 SCID mice for subQ xenografts relative to A375, Colo205, and A375 NRAS; NOD-scid IL2R gammanull (NSG) for the PDX J000106560, C57BL/6 for allografts derived from YUMM1.7 cells and Balb/c nude for the A375-LUC macro- and micro-disseminated brain metastatic models.

For oral administation Compound Ia was formulated in 10% PEG400 (Sigma Aldrich, #06855), 10% Kolliphor HS15 (Sigma Aldrich, #42966), dabrafenib in 0.5% hydroxypropyl methylcellulose (Sigma Aldrich, #09963), 0.1% Tween 80 (ACROS #278630010), encorafenib in 10% PEG400 (Sigma Aldrich, #P3265), 10% Soluplus (BASF), 0.2% SDS (Sigma-Aldrich, #71725) for, and binimetinib in 1% carboxymethylcellulose-Na (Sigma Aldrich, #C9481) 0.5% Tween 80.

Subcutaneous xenograft models were established through cell injection (see Supplementary Materials and Methods for details) of the right flank of mice and drug treatment initiated, after randomization, when tumors reached 100 to 200 mm³. Tumor volumes were monitored through regular caliper measurement.

The brain macrometastatic model was generated by injecting A375 stably expressing firefly luciferase (A375 LUC) into the forebrain. The inoculation procedure was performed with the support of a stereotactic device and anesthesia for the surgery carried out with isoflurane/O₂ using an inhalation mask.

The brain micrometastatic model was generated by injecting A375-LUC cells in the previously exposed carotid arteria interna. The surgery and inoculation procedure was performed with anesthesia carried out with isoflurane/O₂ using an inhalation mask.

Tumor monitoring in the brain macro- and micro-metastatic brain models was performed through bioluminescence imaging (BLI) upon intraperitoneal injection of 100 μg/mouse D-luciferin (Promega) 10 minutes before imaging.

Single-dose pharmacokinetic studies were generated for Compound Ia, dabrafenib, or encorafenib in male Wistar rats. In the first dose group per compound, three rats received 1 mg/kg by intravenous bolus and blood samples were collected from each animal until 24 hours post-dose. In a second group per compound, 3 animals received 20 mg/kg orally by gavage, blood samples were collected from each animal until 24 hours post-dose, and cerebrospinal fluid (CSF) samples were collected via a cannulation from the cisterna magna until 24 hours post-dose. Samples were stored at −80°C until processing prior to LC/MS-MS analysis to quantify compound concentration in samples.

Surplus, surgically excised tumors were provided by the biobank at the Department of Dermatology (University Hospital Zurich) after de-identified patients provided written, informed consent (BASEC-Nr.PB2017–00494). Tumor digests were prepared through incubation with an enzyme mixture containing 250 U/mL accutase (Sigma Aldrich, #A6964), 275 U/mL collagenase IV (Worthington, #LS004189), 10 U/mL DNAse I (Sigma Aldrich, #D5025), and 470 U/mL hyaluronidase (Sigma Aldrich, #H6254) for 30 minutes at 37°C and filtered with a 70 μmol/L cell strainer. Cell treatment was performed with Compound Ia at 100 nmol/L or 1 μmol/L for 2 hours followed by lysate preparation.

The chemical structure of a new ATP competitive BRAF inhibitor, hereafter referred to as Compound Ia, is reported in Fig. 1A. Biochemical profiling of the molecule indicated potent BRAF WT, BRAF V600E, and c-RAF binding at K_d of 0.6, 1.2, and 1.7 nmol/L, respectively. The selectivity profile conducted on 403 kinases (Fig. 1B; Supplementary Fig. S1A) revealed that except for the binding to the kinase BRK at a K_d of 156 nmol/L, no other kinases were bound by Compound Ia with K_d lower than 500 nmol/L indicating a highly selective profile.

The enzymatic activity characterization of Compound Ia on BRAF WT and the clinical relevant BRAF mutants V600E/K/A/D further confirmed the potent kinase inhibitory activity across all BRAF mutants tested with IC₅₀ at concentrations lower than 1.77 nmol/L (Supplementary Fig. S1B).

Compound Ia has a lower molecular weight of 461 Da compared with vemurafenib (489 Da), dabrafenib (519 Da), and encorafenib (540 Da); a property of relevance to facilitate brain penetration.

Compound Ia was designed to avoid paradoxical MAPK hyperactivation in non-BRAF V600E models, an undesired property of the three currently approved BRAFi. Immunoblotting (Fig. 1C) performed in a panel of KRAS mutant and BRAF WT models confirmed that while encorafenib and dabrafenib treatment promoted a marked increase of P-MEK and P-ERK starting from a concentration of 30 nmol/L, upon Compound Ia treatment, P-ERK and P-MEK levels remained substantially unaffected across all cell lines. Analogous results were observed with a more quantitative P-ERK measurement performed through a homogeneous time resolved fluorescence (HTRF) assay in HCT116 cells with Compound Ia avoiding P-ERK paradoxical induction to an extent analogous to that observed for the previously reported paradox breaker PLX8394 (21). Conversely, strong paradoxical MAPK induction with a bell-shaped dynamic was observed upon dabrafenib treatment (Fig. 1D). Furthermore, in agreement with previous reports on the activity of paradox breaker BRAFi in non V600E/K mutations (22, 23), Compound Ia but not dabrafenib, induced P-ERK repression in models presenting the type II and type III BRAF mutations G469A and G466V (Supplementary Fig. S2A).

We next examined the activity of Compound Ia in a collection of 94 cell lines representative of various histologic and genetic backgrounds. Compound Ia exerted cytotoxic activity within a range of 5.2 to 30.2 nmol/L, exclusively in the five BRAF V600E models represented within the tested collection and importantly, with the exception of HL60 BRAF WT cells where a IC₅₀ of 5.0 μmol/L was obtained. An IC₅₀ could not be determined for concentrations ranging from 0 to 10 μmol/L, in the other 88 BRAF WT cell lines, further proving the optimal on-target activity and selectivity profile of this molecule (Fig. 1E and F; Supplementary Fig. S2B). Additional experiments conducted in two organoids cultures derived from BRAF V600E mutant colorectal cancers further demonstrated the activity of Compound Ia which, in line with the expected activity in colorectal cancer (24), was strengthen by the combination with the EGFR antibody cetuximab (Supplementary Fig. S2C).

The pharmacokinetic (PK) and pharmacodynamic (PD) profiles of Compound Ia were evaluated in immunocompromised mice bearing the BRAF V600E mutant A375 melanoma xenografts. Compound Ia is characterized by a very low plasma clearance and fast and extensive oral absorption resulting in a high plasma drug exposure at low doses (Supplementary Tables SA and SB). In line with these compound properties, the treatment with Compound Ia triggered dose-dependent repression of P-ERK after single oral administration starting from a dose of 0.3 mg/kg and with the higher dose of 5 mg/kg driving robust and prolonged MAPK repression up to 24 hours (Fig. 2A and B). Furthermore, PD analysis conducted in A375 xenograft bearing mice treated for 8 days with Compound Ia at 1 mg/kg confirmed the potent MAPK repression observed upon multiple drug treatments (Fig. 2C). In the same xenograft model, the PD activity of Compound Ia at 5 and 20 mg/kg at 6 hours after oral administration was compared with dabrafenib at 100 mg/kg and encorafenib at 6 mg/kg; for the latter two agents, the doses were chosen based on literature reports evidencing effective tumor control of A375 derived xenografts (25). Results revealed that Compound Ia triggers superior MAPK repression in vivo compared with that achieved by dabrafenib and encorafenib (Fig. 2D).

The observed high efficacy of Compound Ia in inducing durable repression of the MAPK signaling at low doses prompted us to examine whether such PD activity translates into corresponding efficacious antitumor responses. We thus compared, in diverse BRAF V600E xenograft models, Compound Ia to dabrafenib and encorafenib, which were administered at doses that mimic clinically relevant unbound drug exposure (Supplementary Table SC) and demonstrate maximum antitumor activity, for these agents, in the A375 xenograft model (Supplementary Fig. S3A).

Compound Ia treatment induced dose-dependent tumor regression in A375-derived xenografts starting from a dose of 1 mg/kg and complete tumor remission was observed in Colo205 and in the Genetically Engineered Mouse Model-derived YUMM1.7 (26) starting from the dose of 5 and 1 mg/kg, respectively (Fig. 3). In the A375 and Colo205 xenograft studies, dabrafenib at 100 mg/kg led to tumor stasis and encorafenib, at 36 mg/kg, to tumor regression to a lower extent to that achieved by Compound Ia at lower doses. Moreover, in an independent experiment in the A375 model, complete tumor remission was observed, upon compound Ia treatment at 20 mg/kg (Supplementary Fig. S3A). On the basis of single-dose PK/PD studies in A375 tumor-bearing mice, Compound Ia demonstrated considerably higher efficacious (unbound) plasma exposure at an oral dose of 20 mg/kg compared with dabrafenib and encorafenib at doses of 100 and 6 mg/kg, respectively (Supplementary Table SA). This corroborates with the improved efficacy of Compound Ia.

Further validation was obtained in a BRAF V600E patient-derived xenograft (PDX) model of melanoma, which confirmed again the potent antitumor activity of Compound Ia eliciting marked tumor regression at 20 mg/kg, similar to that achieved with encorafenib at 36 mg/kg (Fig. 3C).

Collectively, these results support that Compound Ia presents excellent PK properties driving prolonged MAPK repression at low doses, which result in superior antitumor activity compared with encorafenib and dabrafenib.

The acquired ability to trigger RAF dimerization upon treatment with first-generation BRAFi is the predominant mechanism of resistance reported in melanoma. This is often enabled by the acquisition of genetic events that include acquisition of RAS mutations, expression of the p61 BRAF splice variant and BRAF amplification (10, 27).

We next tested the hypothesis that Compound Ia could retain activity in models presenting RAF-dimer–driven resistance.

We therefore utilized the isogenic model A375 harboring the NRAS Q61K mutation (A375 NRAS), which recapitulates a frequently occurring escape mechanism to BRAFi and BRAF/MEKi combinations in the clinic (28).

In this model, the acquired NRAS Q61K mutation, while promoting RAF-dimer–mediated signaling with consequent resistance to RAF-dimer-inducing BRAFi, does not abrogate the dependency from expression of BRAF V600E as the predominant driver oncogene. This is confirmed by BRAF knockdown experiments that, albeit being accompanied by some impact on c-RAF levels, confirms the BRAF essentiality for MAPK signaling maintenance in the A375 and A375 NRAS models (Supplementary Fig. S3B).

The inhibitory activity of Compound Ia on MAPK signaling is compared with encorafenib on MAPK signaling inhibition and to encorafenib and dabrafenib on cell clonogenicity (Fig. 4A and B). Our results reveal that, while Compound Ia triggers similar MAPK repression and inhibition of clonogenicity in A375 cells, effective MAPK repression, and antiproliferative activity in the A375 NRAS cell model were only observed with Compound Ia starting at a concentration of 100 or 300 nmol/L for P-ERK signaling and clonogenicity, respectively.

We next sought to validate these findings in vivo, on xenografts derived from the A375 NRAS model (Fig. 4C).

To better reflect efficacy data in the clinical setting, the paradox inducer BRAFi encorafenib was administered in combination with the MEKi binimetinib at doses approximating clinical exposures for both agents and compared with Compound Ia, given as single agent at doses of 20, 75, and 180 mg/kg. Higher doses of Compound Ia, compared with previous in vivo experiments, were chosen based on the shift of cytotoxic activity observed between A375 and A375 NRAS cells (Fig. 4B). Relative pharmacokinetic data for Compound Ia at these higher doses are reported in Supplementary Table SD.

Tumor growth kinetics (Fig. 4C) indicate that modest antitumor activity is achieved with encorafenib/binimetinib treatment thus further validating the relevance of this model in mimicking a state of acquired resistance not only to BRAFi but also to BRAF/MEKi combinations. Furthermore, in A375 NRAS xenografts, lack of activity was also observed for the paradox breaker BRAFi PLX8394 (Supplementary Fig. S3C and S3D).

Conversely, although Compound Ia at doses of 20 mg/kg induced tumor growth inhibition analogous to that achieved by encorafenib/binimetinib, it led to tumor stasis at 75 and 180 mg/kg.

Importantly, the treatment with compound Ia at the doses of 75 and 180 mg/kg was very well tolerated in mice and no neurologic or other adverse observations were reported in any of the studies conducted as also supported by the mouse body weight data (Supplementary Fig. S3E).

Taken together these data demonstrate that the paradox-breaker properties of Compound Ia, coupled with its high in vivo activity, clearly present a differentiated profile to first-generation BRAF-inhibitors with the potential to achieve activity in BRAF-driven tumors presenting resistance through RAF-dimer–based mechanisms.

Compound Ia is developed to efficiently permeate through the blood–brain barrier (BBB) and achieve efficacious exposure in the central nervous system (CNS) compartment, in order to have the potential to achieve durable responses in patients presenting brain metastatic disease.

P-glycoprotein is the main efflux transporter responsible for exclusion of drugs at the BBB (29). The apical efflux ratio (AP-ER) of Compound Ia, dabrafenib, and encorafenib was calculated in porcine kidney epithelial cells (LLC-PK1) overexpressing human P-gp (30). Compound Ia has a low human AP-ER of 1.5, indicating that it is not a substrate for P-gp mediated efflux. Conversely, dabrafenib and encorafenib have relatively high AP-ER of 3.6 and 6.9 (Supplementary Table SE), respectively, suggesting that their passage across the BBB is expected to be limited by P-gp-mediated efflux.

The brain penetration properties of Compound Ia, dabrafenib, and encorafenib were examined in vivo following a single dose of 1 mg/kg administered intravenously or 20 mg/kg administered orally to male Wistar rats. Following dosing, CSF and/or plasma were sampled serially up to 24 hours and compound concentrations were quantified by LC/MS-MS (Fig. 5A and B; Supplementary Table SB). PK parameters confirmed that Compound Ia presents with high oral plasma exposure and brain penetration potential as supported by the high exposure observed in plasma and CSF and by the high unbound CSF to plasma ratio (Kp,uu) value of 6.6. It is noteworthy that due to the very high plasma protein binding of Compound Ia in rat plasma (>99% protein bound), there may be some uncertainty with the derived fraction unbound in plasma (fu,p 0.0014) and subsequent Kp,uu determination. In comparison, dabrafenib oral plasma exposure was considerably lower although presented brain penetration potential with a CSF Kp,uu of 3.7, whereas encorafenib demonstrated high oral plasma exposure but a low brain penetration potential with a CSF Kp,uu of 0.014. These data well aligns with the previously reported pharmacodynamic experiments suggesting incomplete P-ERK abrogation after 6 hours of dabrafenib treatment and the reduced antitumor activity in xenograft experiments (Fig. 3). Collectively the PK evaluation of the three BRAFi examined here support that Compound Ia presents superior brain penetration properties and that concentrations achieved in the CNS compartment are expected to result in prolonged BRAF inhibition.

It is important to underline that the PK profiling was performed in animals with intact BBB thus providing an accurate evaluation of the brain penetration potential and without alteration or disruption of the BBB, an event expected in tumor lesions growing in the CNS compartment (29).

We next examined whether the high brain penetration potential of Compound Ia results in superior antitumor activity in brain metastatic models. Macro-metastatic disease was modeled by establishing tumors upon forebrain injection of A375 cells transduced with firefly luciferase to facilitate tumor growth monitoring through BLI.

Drug treatment was performed for 21 days and mice followed via tumor BLI and signs of body condition deterioration due to tumor progression.

Results demonstrate that dabrafenib at 100 mg/kg produced only a minor survival benefit and transiently controlled tumor progression as indicated by BLI data, whereas no tumor-related lethalities during treatment and a superior survival benefit were observed in the groups treated with 5 and 20 mg/kg of Compound Ia (Fig. 5C). These data are well in line with evidence of tumor regression observed by measuring BLI overtime (Fig. 5D).

It has been reported that brain macrometastasis could be associated with reduction of BBB integrity (29), therefore we further explored the impact of Compound Ia in a brain micrometastasis model aimed at mimicking early CNS metastatic spread. Disseminated micrometastasis were generated through intracarotid injections of A375-Luc cells and treatment initiated 13 days afterwards to permit cell seeding and micrometastasis establishment.

The activity of Compound Ia was compared with the combination encorafenib/binimetinib.

Tumor progression was monitored through BLI imaging, which revealed that encorafenib/binimetinib was effective in producing initial tumor control but followed by evidence of tumor progression despite treatment continuation (Fig. 5E and F).

Conversely, Compound Ia at 20 mg/kg triggered effective tumor regression throughout the duration of the study.

Taken together these data clearly support that Compound Ia presents high CNS permeability, which is associated with superior antitumor activity in preclinical models of brain metastasis.

We next sought to validate the activity of Compound Ia ex vivo using patient tumor explants. Resected metastatic lesions from nine patients with BRAF mutant-positive melanoma: three patients with brain metastatic lesions (Spec nos. 1, 7, 9), three with lymph node metastases (Spec nos. 2, 5, 6), and three with cutaneous or subcutaneous lesions (Spec nos. 3, 4, 8) were dissociated, and subsequently treated with Compound Ia for P-ERK level evaluation. Samples 1, 3, 8, and 9 were from patients who progressed to either BRAFi or BRAF/MEKi therapies (Supplementary Table SF).

The performed analysis through Western blot analysis on isolated tumor cells was preferred over other methods because it permits to account for different tumor cellularity across diverse specimens. Moreover, due to limited availability of material, an ideal comparative ex vivo analysis with available BRAF/MEK combination could not be conducted, thus, we limited this analysis to the investigation of Compound Ia activity in these patient derived samples.

Compound Ia triggered effective MAPK inhibition, starting from 100 nmol/L, in eight out of nine specimens tested (Fig. 6), including in three out of four specimens derived from patients previously treated with BRAFi or BRAF/MEKi.

Here we describe a comprehensive preclinical characterization of a next-generation BRAFi, referred to as Compound Ia, which present the potential to address the key limitations of the currently approved BRAF inhibitors.

Compound Ia was specifically designed to have paradox breaker properties thus providing the ability to effectively inhibit BRAF V600E/K-driven MAPK signaling without triggering MAPK hyperactivation in BRAF WT contexts.

This novel BRAFi presented an improved selectivity profile coupled with high potency against all relevant BRAF V600 substitutions as demonstrated in a large cell line panel where selective cytotoxicity in BRAF mutated models, and complete inactivity in all BRAF WT cell lines except one, was observed.

These properties suggest that Compound Ia may have a wide therapeutic index because the high selectivity could result in minimal undesired off target effects and, the absence of paradoxical activation, could avoid toxicities which are intrinsic to the mode of action of first-generation BRAFi.

The antitumor activity of Compound Ia, evaluated in relevant xenograft models, revealed effective and prolonged MAPK repression in tumors, which translates into potent tumor regressions at lower doses compared with dabrafenib and encorafenib, supporting the highly favorable properties of this agent.

In this regard, a limitation of this study is that in efficacy studies in models mimicking a BRAF/MEK naïve setting (Fig. 3), Compound Ia was compared only with single-agent dabrafenib or encorafenib and not to their respective combinations with MEKi, potentially underestimating the activity of the approved therapies in these experiments.

Besides the improved tolerability profile of paradox breaker BRAFi, this class of agents retains activity against several resistance mechanisms compared with approved BRAFi which are triggered by RAF dimerization. It has been reported, for example, that the paradox breaker PLX8394 manifests antineoplastic activity in models expressing the BRAF p61 splice variant, BRAF V600E amplification, and some BRAF fusions (23, 31).

In line with this profile, Compound Ia effectively inhibits MAPK signaling and reduces cell viability in a model of acquired resistance to BRAFi harboring the BRAF V600E/NRAS Q61K double mutation.

Importantly, the in vivo activity of Compound Ia as single agent in this BRAFi resistant model at 20 mg/kg was analogous to the one achieved by the combination of encorafenib/binimetinib at clinically relevant mouse equivalent doses whereas tumor control was observed at higher Compound Ia doses. These data suggest that the paradox breaker properties of Compound Ia might attenuate or even abolish the requirement for MEKi co-administration but it also indicates that relatively higher doses of Compound Ia might be needed to achieve efficacy in RAF dimer-mediated resistant tumors.

It is noteworthy that Compound Ia was well tolerated in mice at all doses tested (up to 180 mg/kg); thus, the preclinical data support that efficacious doses needed to drive efficacy in RAF-dimer resistant tumors could be reached. Moreover, Compound Ia could be tested in appropriate combination regimens to further boost its antitumor activity in tumors presenting resistance to available therapies but also, potentially, in BRAF/MEK naïve contexts.

Although the activity of paradox breaker BRAFi on some RAF dimer-mediated mechanism of resistance could be achieved and even broadened by PanRAF inhibitors, a class of agents equally active on monomeric and dimeric RAF (32, 33). These drugs inevitably inhibit RAF-driven signaling in healthy tissue leading to possible limitations due to tolerability (34, 35).

Metastatic spread into the CNS is a frequent dismal event for patients with melanoma, which is associated with poor prognosis.

The limited duration of response in brain metastatic lesions observed with available BRAFi in the clinic appears to correlate with the suboptimal brain penetration of these agents (17, 36).

Confirming this hypothesis, we observed limited exposure over time in a single-dose PK experiment for encorafenib and dabrafenib; conversely, Compound Ia, demonstrated high permeability across the BBB with an exposure predicted to drive antitumor cytotoxic activity.

In line with these findings, strong tumor regression, as indicated by BLI, and superior survival benefit within the treatment window was observed in Compound Ia-treated mice at 5 and 20 mg/kg; in contrast, 100 mg/kg dabrafenib provided only modest survival improvements.

Because in large brain metastatic lesions the BBB integrity could be compromised, we further validated the therapeutic potential of Compound Ia in a disseminated brain micrometastatic model where treatment with 20 mg/kg Compound Ia clearly demonstrated superiority in tumor control over encorafenib/binimetinib.

Finally, the activity of Compound Ia was tested ex vivo in metastatic lesions derived from nine patients with BRAF mutated melanoma. In line with the expected potent activity demonstrated by Compound Ia in vitro and in vivo, effective MAPK inhibition was observed in eight out of the nine tumors analyzed, including three out of four samples derived from patients who progressed to prior BRAFi or BRAF/MEKi treatment.

In conclusion, Compound Ia presents the profile of a next-generation BRAFi with superior preclinical activity and increased brain penetration and may circumvent the key limitations of available BRAFi therapies (Supplementary Fig. S4). A clinical study may further investigate its potential for the treatment of BRAF-V600 mutant patients with brain metastases, which are in need of more efficacious therapies.

J. Wichmann reports personal fees from F. Hoffmann-La Roche AG outside the submitted work; also has a patent for WO2021/116055 pending. T. Friess reports a patent for P36908 pending and a patent for P36928 pending. M. Kornacker reports employment with F. Hoffmann-La Roche AG and ownership of F. Hoffmann-La Roche AG stock. C. Handl reports personal fees from Hoffmann La Roche outside the submitted work. J. Emmenegger reports personal fees from F. Hoffmann-La Roche Ltd outside the submitted work. D. Hunziker reports personal fees from F. Hoffmann-La Roche AG outside the submitted work; also has a patent for WO2021/116055 pending. D. Krummenacher reports personal fees from F. Hoffmann-La Roche AG outside the submitted work; also has a patent for WO2021/116055 pending. D. Rüttinger reports employment with Hoffmann-La Roche at the time of manuscript preparation/submission. A. Ribeiro reports personal fees from Hoffmann-La Roche Inc. outside the submitted work. M. Bacac reports other support from Roche during the conduct of the study. A. Brigo reports personal fees from F. Hoffmann-La Roche Ltd. during the conduct of the study; personal fees from F. Hoffmann-La Roche Ltd. outside the submitted work. D.S. Hewings reports personal fees from F. Hoffmann-La Roche AG outside the submitted work; also has a patent for WO2021/116055 pending to F. Hoffmann-La Roche AG. R. Dummer reports personal fees for intermittent, project focused consulting, and/or advisory relationships with Novartis, Merck Sharp & Dhome (MSD), Bristol-Myers Squibb (BMS), Roche, Amgen, Takeda, Pierre Fabre, Sun Pharma, Sanofi, Catalym, Second Genome, Regeneron, Alligator, T3 Pharma, MaxiVAX SA, Pfizer, and touchIME outside the submitted work. M.P. Levesque reports grants from Roche, Novartis, and Molecular Partners outside the submitted work. G. Schnetzler reports personal fees from Roche during the conduct of the study. B. Martoglio reports employment with F. Hoffmann-La Roche Inc. J.R. Bischoff reports employment with Hoffmann-La Roche Ltd and ownership of F. Hoffmann-La Roche Ltd. stock. P. Pettazzoni reports personal fees from F. Hoffmann-La Roche AG outside the submitted work; also has a patent for WO2021/116055 pending. No disclosures were reported by the other authors.

J. Wichmann: Conceptualization, supervision, investigation, writing–original draft. C. Rynn: Conceptualization, data curation, writing–original draft, writing–review and editing. T. Friess: Conceptualization, validation, investigation. J. Petrig-Schaffland: Methodology. M. Kornacker: Writing–review and editing. C. Handl: Investigation. J. Emmenegger: Investigation. J. Eckmann: Conceptualization, validation, investigation, methodology. F. Herting: Conceptualization, validation, investigation, methodology. N. Frances: Investigation, methodology. D. Hunziker: Investigation, methodology. D. Krummenacher: Resources, methodology. D. Rüttinger: Resources, supervision, methodology. A. Ribeiro: Resources, investigation, methodology. M. Bacac: Resources, investigation, methodology. A. Brigo: Investigation, methodology. D.S. Hewings: Investigation, methodology. R. Dummer: Conceptualization, investigation, methodology, writing–original draft, writing–review and editing. M.P. Levesque: Conceptualization, investigation, methodology, writing–original draft, writing–review and editing. G. Schnetzler: Conceptualization, supervision, methodology, writing–original draft, writing–review and editing. B. Martoglio: Conceptualization, supervision, investigation, methodology, writing–original draft, writing–review and editing. J.R. Bischoff: Conceptualization, data curation, supervision, investigation, writing–original draft, writing–review and editing. P. Pettazzoni: Conceptualization, resources, data curation, supervision, investigation, writing–original draft, writing–review and editing.

We thank for technical assistance: Sylvia Herter, Monika Aegler, Peter Schrag, Christelle Rapp, Véronique Dall'Asen, Marie Stella Gruyer, Didier Wolter, Damien Docquir, Bastian Riviere, Claudia Senn, and Wolfgang Jacob. Pawel Dzygiel is greatly appreciated for the conduct of the single-dose rat PK studies and corresponding sample bioanalysis. Special thanks to the in vivo pharmacology and bioimaging groups, particularly to Stefanie Lechner, Christa Bayer, Petra Ulrich, Daniela Geiss, Stefan Hoert, Gunter Muth, Carsten Wolter, Franz Osl, and Thomas Poeschinger. Thanks also for the studies on CRC organoids to Carla Verissimo and Esmee Koeedot at HUB organoids.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.