Differences in sustained cellular effects of MET inhibitors are driven by prolonged target engagement and lysosomal retention

Intracellular distribution of drug compounds is dependent on physicochemical characteristics and may have a significant bearing on the extent of target occupancy and, ultimately, drug efficacy. We assessed differences in the physicochemical profiles of MET inhibitors capmatinib, crizotinib, savolitinib, and tepotinib and their effects on cell viability and MET phosphorylation under steady-state and washout conditions (to mimic an open organic system) in a human lung cancer cell line. To examine the differences of the underlying molecular mechanisms at the receptor level, we investigated the residence time at the kinase domain and the cellular target engagement. We found that the ranking of the drugs for cell viability was different under steady-state and washout conditions, and that under washout conditions, tepotinib displayed the most potent inhibition of phosphorylated MET. Post-washout effects were correlated with the partitioning of the drug into acidic subcellular compartments such as lysosomes, and the tested MET inhibitors were grouped according to their ability to access lysosomes (crizotinib and tepotinib) or not (capmatinib and savolitinib). Reversible lysosomal retention may represent a valuable intracellular storage mechanism for MET inhibitors, enabling prolonged receptor occupancy in dynamic, open physiologic systems, and may act as a local drug reservoir. The use of washout conditions to simulate open systems and investigate intracellular drug distribution is a useful characterization step that deserves further investigation.


Introduction
The mesenchymal-epithelial transition factor kinase (MET; also called the hepatocyte growth factor [HGF] receptor) is a tyrosine kinase proto-oncogene that is preferentially expressed in epithelial cells. Although the physiologic role of MET is not fully understood, it is involved in several developmental stages and is normally strictly regulated. MET signaling can be dysregulated by a variety of MET alterations, leading to oncogenesis through hyperactive or otherwise aberrant MET signaling, metabolic reprogramming, increased resistance to cancer treatment, enhanced metastatic properties, and reinforcement of cancer stem cell properties (Fu et al., 2021). Epidemiologic evidence supports the therapeutic potential of MET inhibition (Gherardi et al., 2012).
Crizotinib was one of the first small multikinase inhibitors with reasonable activity in MET-dependent in vivo xenograft models (Heigener and Reck, 2018), and is approved for the treatment of anaplastic lymphoma kinase (ALK)-or ROS1-positive, metastatic non-small cell lung cancer (NSCLC) (FDA, 2021). In addition to ALK and ROS1 activity, crizotinib shows orthosteric adenosine triphosphate (ATP)-competitive MET inhibition and was originally developed as a MET inhibitor (Drilon et al., 2020). The increasing confidence in MET as a relevant oncogene culminated in the identification of a diverse set of selective class Ib MET inhibitors (Salgia et al., 2020). Although clinical development was discontinued for many of them, capmatinib and tepotinib have been approved for clinical use internationally (Paik et al., 2020;Wolf et al., 2020;Hong et al., 2021), and savolitinib has been approved more recently in China (Lu et al., 2021).
In clinical settings, MET inhibitors can be differentiated by pharmacokinetic characteristics such as volume of distribution and half-life, which not only influence drug dosage (Jones et al., 2020;Xiong et al., 2021), but also potentially modulate primary pharmacology and the risk:benefit ratio of therapeutic agents. In particular, volume of distribution is largely driven by specific and unspecific binding of the therapeutic agent to intracellular proteins and membranes, and to a lesser extent by the stability of the drug-target complex and NanoBRET Cellular Target Engagement Assay HEK293T cells were obtained from DSMZ (Braunschweig, Germany) and cultured at 37°C/5% CO 2 in Dulbecco's Modified Eagle Medium (DMEM)-high glucose, supplemented with 10% FBS (PAA Laboratories), 2 mM L-glutamine, 1 mM sodium pyruvate, and 1x Non-Essential Amino Acids Solution (all Gibco, Life Technologies, Carlsbad, CA, USA). All reagents for the NanoBRET assay and the C-terminal NanoLuc MET plasmid were obtained from Promega (Madison, WI, USA), including transient transfected cell batch MET TE HEK293 (part number: CS1810E18, batch number: 433041) and the NanoBRET TE in-cell Kinase Kit, (part number: N2501, batch number: 288446).
The NanoBRET tracer displacement assay was used to measure cellular target engagement. The bioluminescence resonance energy transfer (BRET) signal occurs between the chemiluminescence signal of the product of the NanoLuc-tagged MET and the tracer molecule (NanoBRET Kinase Tracer K5) labeled with BODIPY fluorophore (Promega) in live cells transiently transfected by the Fugene method 24 hours prior to the assay. The tested MET inhibitor competes with tracer for MET-NanoLuc protein binding and reduces the BRET signal in a concentration-dependent manner. The NanoBRET assay was performed according to the manufacturer's instructions with some adaptations: 120 nl serially diluted inhibitor (in DMSO, final concentration 30 µM to 1 nM) resp. DMSO control was dispensed to 384-well plates (GreinerBio, PP, F-Bottom, 781201) by Labcyte Echo (Beckman Coulter, Brea, CA, USA). 40 µl of a suspension of cMet-NanoLuctransfected cells were added, reseeded at a density of 20,000 cells/ml after trypsinization and resuspending in Opti-MEM without phenol red (Life Technologies). The compounds and cells were incubated for 1 hour at 5% CO 2 and 37°C. For the washout NanoBRET assay, cells were handled as described above. For HEK293T cells, centrifugation was adapted to 4 minutes at 300 xg. After washout, cells were resuspended in 40 µl Opti-MEM without phenol red, and 20 µl were transferred to 384-well plates (Greiner Bio-One, white, F-bottom, 781080).
The washout cells were incubated for 4 hours at 5% CO 2 and 37°C to allow equilibration. For the NanoBRET assay, 750 nM NanoBRET Kinase Tracer K5 (in DMSO) was added to the cell suspension by a Tecan D300e dispenser; after a short mix, the plates were incubated for 2 hours at 5% CO 2 and 37°C for the standard NanoBRET assay. The plates were equilibrated to room temperature for 30 minutes, and Nano-Glo Substrate and Extracellular NanoLuc Inhibitor were added according to the manufacturer's instructions to measure the NanoBRET signal. After a 1-hour incubation in the dark at room temperature, luminescence was measured at 450 nm and fluorescence at 610 nm (acceptor) on a PerkinElmer EnVision multimode reader, and the NanoBRET ratio signal was calculated. Four-parameter curve-fitting was performed with Genedata screener This article has not been copyedited and formatted. The final version may differ from this version. savolitinib, EBC-1 cells were treated as described above (cell viability assay), with cell count and volumes adapted for 96-well plates (part number: 651261, Greiner Bio-One). The treatment concentrations were 0.01, 0.1, 1.0, and 5 µM for tepotinib and 1.0, 5.0, 7.5, and 10 µM for capmatinib, crizotinib, and savolitinib. The difference in the concentration ranges is based on different sensitivities in analyte measurement via LC-MS/MS.
After the treatment period, the incubation media were transferred to a fresh polypropylene PCR plate and diluted were prepared by dispensing a 10 mM DMSO-based stock solution to 100 µl ACN with a Tecan D300e.
Calibration curves were calculated for each compound via linear regression with 1/x 2 weighting.

Sample preparation for tepotinib: protein precipitation
Supernatant samples from the incubation of EBC-1 cells with tepotinib were diluted with MEM (supplemented with 10% FBS) to a final dilution of 1:100, or 1:10 to match the calibration range. Washed cell pellets from the compound incubation step were lysed by adding 100 µl ACN. After incubation with organic solvent for 1 hour at 300 rpm and room temperature, the cell lysates were diluted with ACN in the same manner as supernatant samples. For tepotinib, 30 µl from each calibrator and sample were transferred to a polypropylene PCR plate with 90 µl internal standard solution (ACN with 4 ng/ml stable isotope labeled tepotinib). The PCR plate was mixed for 10 minutes at 1,200 rpm and subsequently centrifuged at 1,700 xg for 10 minutes at 20°C.
To adjust the organic solvent ratio in the final samples, 50 µl supernatants were transferred to 100 µl ammonium bicarbonate buffer (20 mM, pH 6.9). Afterwards, 7.5 µl was injected into the LC-MS/MS system.

Sample preparation for capmatinib, crizotinib, and savolitinib: liquid-liquid extraction
Capmatinib, crizotinib, and savolitinib samples were purified by liquid-liquid extraction due to high matrix effects from MEM. A volume of 50 µl standard solution, diluted supernatant or cell lysate sample was transferred to a 1 ml deep well plate. For pH adjustment, 50 µl buffer (ammonium bicarbonate buffer 20 mM, pH 6.9 for capmatinib and crizotinib, and ultrapure water with 0.025% ammonia [pH 12.0] for savolitinib) were added. For extraction, 600 µl MTBE with internal standard (stable isotope labeled forms with 2.5 ng/ml for capmatinib, 0.25 ng/ml for crizotinib, and 0.5 ng/ml of a structural analog as internal standard for savolitinib) were added to each well. Afterwards, the plate was incubated for 10 minutes at 800 rpm and room temperature.
Then, the plate was centrifuged at 4,700 xg for 10 minutes at 15°C. In the final step, 200 µl supernatant were transferred to a 300 µl PCR plate. The solvent phase was evaporated by TurboVap LV (Zymark, Hopkinton, MA, USA) for 30 minutes at 40°C. Finally, the dried extract was reconstituted with 100 µl Eluent A (ultrapure water with 0.025% ammonia) and ACN (80/20 [v/v]), and 7.5 µl aliquots were injected into the LC-MS/MS system.

LC-MS/MS conditions
For the separation of tepotinib, capmatinib, and crizotinib, a Waters ACQUITY UPLC ® CSH™ C18 (130 Å, 1.7 µm particle size, 2.1 x 50 mm) column was used. A gradient method at 40°C was applied with a flow of 0.75 ml/min using two eluents: Eluent A (ultrapure water, 0.1% formic acid and 10 mM ammonium formate), and Eluent B (acetonitrile). The initial eluent composition of the gradient was 80% Eluent A and 20% Eluent B (0-This article has not been copyedited and formatted. The final version may differ from this version. 1.5 minutes) followed by a linear gradient to 60% Eluent B (1.5-2.8 minutes). Eluent B was then increased to 95% (2.8-2.9 minutes) followed by a plateau until 3.0 minutes. From minute 3.0 to 5.0, starting conditions were applied (80% Eluent A and 20% Eluent B).
Finally, all analytes and internal standard compounds were detected by multiple-reaction monitoring using ionization in positive mode with the following parameters: curtain gas: 35 psi, ion spray voltage 3,500 V, temperature 500°C; ion source gas 1: 30 psi; ion source gas 2: 70 psi. Compound concentrations were determined with Analyst 1.6.3 (AB SCIEX, Framingham, MA, USA) by calculating the peak area ratio of analyte and internal standard.
LysoTracker Displacement Assay and Immunofluorescence A549 cells (MET wild-type) were purchased from ATCC (Manassas, VA, USA) and grown in DMEM with 10% (v/v) FBS in 10% CO 2 at 37°C. For all assays, 50 ng/ml Gibco (Life Technologies, Carlsbad, CA, USA) penicillin/streptomycin was added. A549 tumor cells were seeded after using a cell strainer with 7,500 cells/well in 100 µl in clear bottom, black 96-well plates (part number: 655090, Greiner Bio-One) and incubated overnight at 10% CO 2 and 37°C in medium without phenol red. Treatment with compounds dissolved in DMSO was conducted using a Tecan D300 dispenser. Chloroquine diphosphate salt (part number: C6628, Sigma Aldrich) 50 µM was used as a control for 0% lysosomal staining, and 0.5% DMSO vehicle for 100% lysosomal staining.
Thirty minutes after compound treatment, 25 nM LysoTracker Red DND-99 (Life Technologies Europe, Netherlands) was added manually (10 µl prediluted). After shaking for 5 seconds, the plates were incubated for another 30 minutes at 37°C and 10% CO 2 , then washed once with phosphate-buffered saline (PBS) and fixed with 4% paraformaldehyde in PBS for 15 minutes at room temperature. After two washing steps with PBS, Hoechst (Molecular Probes, 1:10,000) was added. Imaging was performed using an ImageXpress Micro Confocal system (Molecular Devices, San Jose, CA, USA).
For washout conditions, 10,000 A549 tumor cells per well were seeded in 100 µl in 96-well polypropylene v-bottom plates (part number: 651261, Greiner Bio-One), and treated and washed as described above.
Afterwards, cells were resuspended in fresh medium and transferred to a black 96-well plate with a clear bottom.
In parallel, cells from control plates (no washing steps in the polypropylene plate) were transferred to black 96-This article has not been copyedited and formatted. The final version may differ from this version. well plates with a clear bottom. Control and washout plates were incubated at 37°C, 10% CO 2 for 3 hours. After incubation, chloroquine was added, with LysoTracker Red added 30 minutes later, and plates were processed as described above for the standard conditions. For lysosomal-associated membrane protein 1 (LAMP1) staining, fixed cells were permeabilized with 0.1% Triton X-100 for 10 minutes and washed once with PBS and blocked with 1% goat serum (part number: B15-035, PAA Laboratories), 0.1% BSA (part number: K11-013, PAA Laboratories) and 0.1% sodium azide (the healthcare business of Merck KGaA, Darmstadt, Germany) for 1 hour. LAMP1 primary antibody (part number: 9091, D2D11 XP Rabbit mAb, Cell Signaling Technology, MA, USA) was diluted 1:100 in PBS containing 0.1% BSA and incubated overnight at 4°C. After washing three times the next day, the secondary goat antirabbit Alexa Fluor 488 antibody (part number: A-11070, batch number: 2018208, Thermo Fisher Scientific, IL, USA) was diluted 1:2,000 in PBS containing 0.1% BSA and incubated for 75 minutes. Cells were washed three times and wells filled up to 100 µl; PBS and 0.05% sodium azide was added for long-term storage. Imaging was performed using an 20x objective and the ImageXpressULTRA High Throughput Confocal Imaging System and the associated software MetaXpress, version 6.6.1.42 (both from Molecular Devices). Images from LysoTracker and LAMP1 staining were overlayed with FiJi 1.53c (Schindelin et al., 2012).

Statistical Analysis
All experimental results are expressed as mean ± standard deviation (SD) or mean ± standard error of the mean (SEM) of three or more independent experiments; details are specified in the figure legends. Statistical analysis was performed using GraphPad Prism software (GraphPad Software, CA, USA). The statistical significance of differences between controls and washout conditions for IC 50 values and target engagement was determined using multiple unpaired Student t-tests. We corrected for multiple comparisons using the Holm-Šidák method and considered a p-value of less than 0.05 to be significant. This article has not been copyedited and formatted. The final version may differ from this version.

Cell Viability and pMET Persistence After Washout
After a 14-hour washout, tepotinib displayed persistent pMET inhibitory effects in A549 cells stimulated with HGF (Schadt et al., 2010;Bladt et al., 2013). In the present study, the human lung cancer cell line EBC-1 was used to evaluate cell viability, pMET status and to assess tepotinib's cell persistence in addition to that of the MET inhibitors capmatinib, savolitinib, and crizotinib. EBC-1 was selected as a model cell line because EBC-1 cells harbor MET amplification, express high levels of MET, have high basal MET phosphorylation, and their viability is sensitive to MET inhibition (Lutterbach et al., 2007). Additionally, EBC-1 cells allowed a robust assay readout in washout settings and polypropylene plates, which was not the case for other tested cell lines (data not shown). Fig. 1A and 1B illustrate the potency shifts for tepotinib and capmatinib after washout; doseresponse curves for crizotinib and savolitinib are available in Supplementary Fig. S1A and B. Fig. 1C summarizes cell viability (IC 50 ) after treatment and IC 50 values after washout. Tepotinib, capmatinib, and savolitinib showed similar potencies in the viability assay in steady-state conditions (IC 50 1-2 nM), and crizotinib showed a lower potency (11 nM), consistent with previously reported data (Baltschukat et al., 2019).
Applying washout conditions, more precisely 1-hour compound treatment followed by three washing steps and after 72-hour viability assessment, mean IC 50 values increased for all inhibitors but differed in their extent. The IC 50 value of tepotinib and crizotinib shifted 14-and 10-fold, respectively, while the IC 50 for capmatinib and savolitinib shifted by factors of ≥300. We also investigated the more direct effect of these four compounds on MET phosphorylation (Y1234/1235, Fig. 1 Fig. S1C and D), which paralleled the effect observed for viability. The pMET IC 50 values for tepotinib and crizotinib shifted by a factor of 5, and shifts observed for capmatinib and savolitinib were again more extreme, with factors >150. To ensure that the washout effects were specific to the compound and cell interaction, we excluded plastic binding of the inhibitors (Palmgrén et al., 2006) by using polypropylene plates and cell-free controls (see Supplementary Fig. S2).

Effect of Drug-Target Complex Duration on the Differences Between the MET Inhibitors
Tepotinib has a long residence time of >1,000 minutes (Willemsen-Seegers et al., 2017), which could contribute to persistent effects in a washout set-up. We analyzed residence times and dissociation rate constants (k off ) for all four MET inhibitors ( Fig. 2A); crizotinib had the fastest k off and tepotinib the slowest. Fig. 2B shows best estimates of residence times from k off values averaged over all time points with inhibitor occupancies higher than 10%; the residence time of tepotinib was estimated at 20 hours, followed by capmatinib (14 hours), savolitinib (10 hours), and crizotinib (4 hours).
This article has not been copyedited and formatted. The final version may differ from this version. Target engagement kinetics were analyzed in a cellular system using a MET NanoBRET assay (Fig. 2C) (Robers et al., 2015). In non-washout conditions, target engagement potencies ranged from 9-28 nM, with crizotinib less potent than the other inhibitors. We tried to recapitulate the washout conditions of the EBC-1 viability and pMET assays but had to adjust the protocol to HEK293T cells, a standard model cell line for mechanistic NanoBRET assays. Keeping the compound incubation time constant at 1 hour and applying three washing steps in polypropylene plates, cells had to be transferred to cell culture plates and different incubation times were addressed. After 2 hours, we saw initial splits in potencies, which increased at 4 hours ( Supplementary Fig. S3). Extended incubation did not further increase potency shifts, but this could be due to technical limitations of the assay set-up at later time points (data not shown). Fig. 2C illustrates the potency shift for tepotinib and shift factors for all MET inhibitors 4 hours after washout. Additional replicates can be found in Supplementary Fig. S4. Despite having the shortest residence time, the mean IC 50 value for crizotinib shifted by a factor of 16, which is comparable to the shift of tepotinib (factor 19, longest residence time). The difference in ranking order in the persistence of effects in the cellular NanoBRET assay after washout and protein-based surface plasmon resonance (SPR) assessment indicates that parameters other than on-target residence time contribute to the persistent cellular effects of tepotinib and crizotinib described above (Fig. 1).

Sustained Effects After Washout and Intracellular Compound Concentrations
Intracellular concentrations of MET inhibitors from EBC-1 cell lysates and corresponding extracellular compound concentrations in the incubation medium (after 1-hour treatment, before washout) were investigated as additional parameters influencing their sustained cellular effects. Fig. 3A shows the measured compound concentration after cell lysis with 100 µl acetonitrile normalized to 10,000 cells/well. Fig. 3B shows the detailed compound recovery from cell lysates and incubation medium for the 5 µM treatment. The partition coefficient reflects the ratio between the intracellular drug amount and the remaining compound in the incubation medium after treatment. It was calculated by building the arithmetic mean of the partition coefficient from all incubation concentrations for each MET inhibitor. A partition coefficient of 1.75 ± 0.61 was determined for crizotinib, 0.68 ± 0.39 for tepotinib, 0.003 ± 0.002 for capmatinib, and 0.001 ± 0.0007 for savolitinib. At 5 µM, tepotinib and crizotinib showed >50% intracellular compound concentrations, indicating a cellular accumulation of the compounds, which could be cell organelle or membrane-binding associated.

MET Inhibitor Differences in Physicochemical Profile
Based on initial structural considerations, we hypothesized that physicochemical differences might explain differences in intracellular compound concentrations. To further substantiate this working hypothesis, the major This article has not been copyedited and formatted. The final version may differ from this version.  (Benet et al., 2016), characterizing pharmaceutically relevant physicochemical property ranges. Despite the molecular weight of the MET inhibitors covering a considerable range (345-493 g/mol) and excepting the slightly increased number of hydrogen bond donators of crizotinib, the numbers of hydrogen bond acceptors and donators, as well as the calculated TPSA, were comparable. However, lipophilicity (logP) and the acid dissociation constant (pKa1) separate the four compounds into two categories: lipophilic weak bases (tepotinib and crizotinib) and neutral molecules (capmatinib and savolitinib).

Reservoirs
Lipophilic weak bases are known to accumulate in acidic cell compartments such as lysosomes, so we assessed lysosomal retention using two approaches: effects of co-treatments with chloroquine and an indirect fluorescence-based assay (Schmitt et al., 2019). Co-treatment with a lysosomotropic agent (e.g. chloroquine) should compete with lipophilic weak bases (tepotinib and crizotinib) for storage capacities of acidic cell compartments and, therefore, reduce the cellular compound concentration and activity in washout settings. We applied selected MET inhibitor concentrations within their dose-response range and combined this treatment with either DMSO control or chloroquine at 10 or 50 µM. For tepotinib, Fig. 4A shows reduced inhibition of MET phosphorylation with increasing concentrations of chloroquine co-treatments, indicating a competition of tepotinib and chloroquine for acidic compound reservoirs. We observed similar effects for crizotinib, but capmatinib and savolitinib activities were not affected by co-treatments with chloroquine ( Supplementary Fig.   S5). A similar competition of lysosomal storage capacities should occur between compounds and the fluorescence probe LysoTracker. Lysosomal retention is expected to result in dye displacement and reduction of the fluorescence signal. We selected A549 for quantification analysis based on assay robustness, but similar effects were detected in the EBC-1 cell line ( Supplementary Fig. S6A). Additionally, LAMP1 lysosome immunofluorescence staining did not detect differences between DMSO controls and tepotinib-treated samples ( Supplementary Fig. S6B), and co-localization of LAMP1 and LysoTracker (DMSO control) confirmed the staining of the desired subcellular compartment. Fig. 4B shows reduced LysoTracker fluorescence with tepotinib, detected by confocal imaging. In Fig. 4C, this effect is quantified using image analysis. Chloroquine was used as a reference compound, and the IC 50 value determined by Schmitt et al. using  respectively. Capmatinib and savolitinib did not affect the fluorescent probe signals. Lastly, we investigated the reversibility of the lysosomal retention of tepotinib. Tepotinib was incubated for 1 hour, and cells were washed and then transferred from polypropylene plates to imaging-compatible plates and left for 4 hours to attach. At 12.5 µM tepotinib (Fig. 4D)

Discussion
Prominent differences existed between the MET inhibitors in washout experiments, altering the potency ranking of the tested compounds. We observed a link between the compounds' physicochemical properties (high pKA and logP) and persistence in washout conditions. We suggest that compound distribution within cells is highly dynamic and influenced by multiple factors (Fig. 5). Besides lysosomal retention and residence time, transport across cell membranes and the amount of unbound intracellular compound could contribute to persistency.

Duration of the Drug-Target Complex and its Effect on Target Inhibition in Open Systems
Drug-target complex stability is determined by its dissociation rate constant. Long residence time and a rapid rebinding driven by the high proximal drug concentration would extend the overall duration of target occupancy (Copeland, 2016). The recovery rate of MET activity after washout has been shown to increase with decreasing MET receptor residence time for a series of MET tyrosine kinase inhibitors (Farrell et al., 2017). In the present study, Fig. 5A illustrates the impact of a stable drug-target complex on overall cellular compound distribution; a slow k off rate enables persistent target inhibition, despite extracellular compound depletion by metabolism or excretion in vivo (or washing steps in vitro). Our analysis of MET inhibitor/target residence times by SPR ( Fig. 2A and B) allowed us to rank dissociation rates of the four MET inhibitors, starting with the slowest k off : tepotinib, capmatinib, savolitinib, and crizotinib. This ranking was not reflected in the potency shifts in NanoBRET washout data nor pMET or cell viability, where potency shifts (lowest first) were ranked: tepotinib/crizotinib, capmatinib, and savolitinib. Our explanation for the discrepancy between protein and cellular target engagement is that additional cellular effects, such as lysosomal retention, mask the effect of residence time. Additional (later) time points for the NanoBRET assay would strengthen our hypothesis but were not feasible using the assay setup and are subject to future investigations. Looking at the two non-lysosomotropic agents, savolitinib and capmatinib, the potency shift ranking matches the residence time ranking, and the better washout performance of capmatinib could be explained by its longer residence time. More molecules with different residence times and similar physicochemical properties could be analyzed further.

Contribution of Lysosomal Retention to Sustained Target Inhibition
Intracellular acidic/lipophilic cell compartments, including lysosomes, are known to be accessible by certain compounds (Miao et al., 2013). The mechanism of lysosomal retention of compounds has been studied since it was described by de Duve et al. (de Duve et al., 1974). Lysosomes in normal cells have a significantly lower pH (~4-5) than the surrounding cytosol, and this acidic subcellular compartment enables lysosomotropic behavior, This article has not been copyedited and formatted. The final version may differ from this version. which is well described for lipophilic weak bases. Intra-lysosomal concentration/capacity depends on the pKa of the compound and the permeability coefficient α of its neutral and protonated forms (Duvvuri et al., 2004(Duvvuri et al., , 2005. The bidirectional passive permeation of the lysosomal membrane also depends on the compound's lipophilicity (logP). Furthermore, lysosomal retention depends on secondary factors, including compound size, intramolecular dipole, active transport, and other mechanisms affecting general biodistribution. Focusing on the measured pKa and lipophilicity descriptors of the four MET inhibitors, the compounds were separated into lipophilic weak bases (tepotinib and crizotinib) and neutral molecules (capmatinib and savolitinib). We have shown that tepotinib and crizotinib are lysosomotropic and capmatinib and savolitinib are non-lysosomotropic agents.
Such a lysosomal compound reservoir could contribute to persistent effects of a compound. To our knowledge, there is only one washout study published, which shows that the CDK4/6 inhibitor palbociclib is reversibly stored in lysosomes, linking the lysosomal compound reservoir to long-term activity of the inhibitor (Llanos et al., 2019). We started to show the reversibility of lysosomal retention experimentally by reduced LysoTracker displacement after washout (Fig. 4D). Two limitations to our dataset could be addressed by expanding cell lines and compounds to confirm the benefits of lysosomal retention and by expanding kinetic analyses to confirm the reversibility of lysosomal retention.
As depicted in Fig. 5, kinetics are important in the reversibility of lysosomal retention; as the non-charged molecule is membrane-permeable, the equilibrium will be restored after extracellular compound removal by washing steps. Cytosolic compound could cross the cell membrane, non-charged compound could cross the lysosomal membrane, and within the lysosome, deprotonation of compounds will restore the equilibrium. In summary, the degree of lysosomal retention, membrane transport, and resulting kinetics depend on individual compound physicochemical characteristics, and our data indicate benefits for the sustained target inhibition of MET by lysosomotropic agents such as crizotinib and tepotinib.
Lysosomal trapping, or sequestration, is associated with a separate set of effects and implications, including reduced effective cellular concentrations or resistance mechanisms (Gong et al., 2003;Duvvuri et al., 2006;Ndolo et al., 2010;de Klerk et al., 2018;Englinger et al., 2018;Halaby, 2019). However, most of these aspects have been studied in steady-state experimental settings, where lysosomal retention generally reduces cytoplasmic concentration and prevents pharmacologic activity at the target site. This change in a compound's unbound intracellular concentration contributes to asymmetry between the extracellular and intracellular drug concentration: the "potency drop-off" (Hann and Simpson, 2014;Trünkle et al., 2020). Lysosomal retention may also be associated with negative phenomena such as phospholipidosis (Lowe et al., 2012;Muehlbacher et al., This article has not been copyedited and formatted. The final version may differ from this version. target could be an advantage, assuming an intact equilibrium between compartments. Once the cytosolic drug concentration decreases, a back diffusion from the lysosomal compartment into the cytosol restores compound availability (Kazmi et al., 2013). These reservoirs could maintain high levels of target engagement in dynamic systems to support C min -driven pharmacology.
Our findings indicate that physicochemical differences are closely linked to intracellular accumulation of MET inhibitors during washout. Basic compounds often show a high volume of distribution (Peters, 2011) and high partition into tissue. The tissue distribution also depends on the tissue type and its relative content of lysosomes (Schmitt et al., 2021). Johne et al. demonstrated in a mass balance study that tepotinib, a lipophilic weak base, distributes preferably into tissue, and was found in higher concentrations in xenograft tumors compared to plasma concentration (Bladt et al., 2013;Johne et al., 2020). In the context of cancer, tumor cells tend to have a higher intracellular pH value compared with normal differentiating cells, but there is increased acidification in the tumor cell environment (White et al., 2017) and lipophilic weak basic compounds distribute preferably to the reduced pH value of the tumor tissue. The reversed pH gradient in tumor tissues is extensively described in the literature (Webb et al., 2011).
Metabolism could additionally affect biodistribution in multiple ways. We considered efflux pumps such as P-glycoprotein (P-gp, ABCB1). However, P-gp should have minimal or no effect on our results because all four MET inhibitors are substrates of P-gp (Cortot et al., 2022). Additionally, the experiments in this study are based on the lung cancer cell lines EBC-1 and A549 with low P-gp expression (Cancer Dependency Map Portal [RRID:SCR_017655]). Therefore, the contribution of lysosomes to sustained target engagement is a reasonable assumption. Further studies are needed to investigate if the lysosomotropic compounds tepotinib or crizotinib induce lysosomal swelling, lysosomal leakage, and potentially a lysosome-mediated cell death and how our observations translate to in vivo and/or clinical settings.

Importance of Sustained Effects to Induce the Desired Phenotype by MET Inhibition
As lysosomal retention is compound-and context-dependent, and consequently able to modulate desired and undesired pharmacologic effects, these correlations offer the opportunity to optimize pharmacokinetic and pharmacodynamic characteristics. Depending on the target, persistent inhibition may be critical to induce a and U87-MG mouse models suggest that near-complete inhibition of MET phosphorylation (>90% inhibition) for the duration of administration is necessary to maximize therapeutic benefit (Zou et al., 2007), and customized regimens of different inhibitors (including tepotinib) may be required to achieve this goal (Srivastava et al., 2018). Thirdly, pharmacokinetic-pharmacodynamic modeling for tepotinib and savolitinib have been based around the need for prolonged, high levels of MET inhibition for clinical activity and identified once-and twicedaily dosing, respectively, as optimal (Jones et al., 2020;Xiong et al., 2021). In the present study, we confirm that persistent target inhibition results in a stronger, phenotypic viability effect as our washout effects on pMET and viability correlate well (Fig. 1). We also show that persistent effects in our in vitro setting depend on the molecular physicochemical properties and could be engineered to favor lysosomal compound reservoirs.   Activity of co-treatments compared to tepotinib mono-treatment. Mean ± SEM; N = 3; two-way ANOVA; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. LysoTracker staining of A549 cells (green) and nuclear counterstain using DAPI (blue). Imaging was performed using the high content imager Micro Confocal  This article has not been copyedited and formatted. The final version may differ from this version.  Tables   Table 1 Physicochemical profiles    This article has not been copyedited and formatted. The final version may differ from this version.

Fig. 3
This article has not been copyedited and formatted. The final version may differ from this version.

Fig. 4
This article has not been copyedited and formatted. The final version may differ from this version.    This article has not been copyedited and formatted. The final version may differ from this version.